Compare commits

..

2 Commits

Author SHA1 Message Date
lizzie
45d513e114 add svg 2025-12-24 09:17:03 +00:00
lizzie
e71f44cc55 [docs/dynarmic] move docs to root 2025-12-24 09:15:32 +00:00
64 changed files with 894 additions and 4027 deletions

View File

@@ -96,8 +96,8 @@
"package": "VVL",
"repo": "KhronosGroup/Vulkan-ValidationLayers",
"tag": "vulkan-sdk-%VERSION%",
"git_version": "1.4.335.0",
"git_version": "1.4.328.1",
"artifact": "android-binaries-%VERSION%.zip",
"hash": "48167c4a17736301bd08f9290f41830443e1f18cce8ad867fc6f289b49e18b40e93c9850b377951af82f51b5b6d7313aa6a884fc5df79f5ce3df82696c1c1244"
"hash": "5ec895a453cb7c2f156830b9766953a0c2bd44dea99e6a3dac4160305041ccd3e87534b4ce0bd102392178d2a8eca48411856298f9395e60117cdfe89f72137e"
}
}

View File

@@ -13,3 +13,4 @@ This contains documentation created by developers. This contains build instructi
- **[The NVIDIA SM86 (Maxwell) GPU](./NvidiaGpu.md)**
- **[User Handbook](./user)**
- **[Release Policy](./ReleasePolicy.md)**
- **[Dynarmic](./dynarmic)**

View File

@@ -343,3 +343,279 @@ SetTerm(IR::Term::If{cond, term_then, term_else})
This terminal instruction conditionally executes one terminal or another depending
on the run-time state of the ARM flags.
# Register Allocation (x64 Backend)
`HostLoc`s contain values. A `HostLoc` ("host value location") is either a host CPU register or a host spill location.
Values once set cannot be changed. Values can however be moved by the register allocator between `HostLoc`s. This is
handled by the register allocator itself and code that uses the register allocator need not and should not move values
between registers.
The register allocator is based on three concepts: `Use`, `Def` and `Scratch`.
* `Use`: The use of a value.
* `Define`: The definition of a value, this is the only time when a value is set.
* `Scratch`: Allocate a register that can be freely modified as one wishes.
Note that `Use`ing a value decrements its `use_count` by one. When the `use_count` reaches zero the value is discarded and no longer exists.
The member functions on `RegAlloc` are just a combination of the above concepts.
The following registers are reserved for internal use and should NOT participate in register allocation:
- `%xmm0`, `%xmm1`, `%xmm2`: Used as scratch in exclusive memory access.
- `%rsp`: Stack pointer.
- `%r15`: JIT pointer
- `%r14`: Page table pointer.
- `%r13`: Fastmem pointer.
The layout convenes `%r15` as the JIT state pointer - while it may be tempting to turn it into a synthetic pointer, keeping an entire register (out of 12 available) is preferable over inlining a directly computed immediate.
Do NEVER modify `%r15`, we must make it clear that this register is "immutable" for the entirety of the JIT block duration.
### `Scratch`
```c++
Xbyak::Reg64 ScratchGpr(HostLocList desired_locations = any_gpr);
Xbyak::Xmm ScratchXmm(HostLocList desired_locations = any_xmm);
```
At runtime, allocate one of the registers in `desired_locations`. You are free to modify the register. The register is discarded at the end of the allocation scope.
### Pure `Use`
```c++
Xbyak::Reg64 UseGpr(Argument& arg);
Xbyak::Xmm UseXmm(Argument& arg);
OpArg UseOpArg(Argument& arg);
void Use(Argument& arg, HostLoc host_loc);
```
At runtime, the value corresponding to `arg` will be placed a register. The actual register is determined by
which one of the above functions is called. `UseGpr` places it in an unused GPR, `UseXmm` places it
in an unused XMM register, `UseOpArg` might be in a register or might be a memory location, and `Use` allows
you to specify a specific register (GPR or XMM) to use.
This register **must not** have it's value changed.
### `UseScratch`
```c++
Xbyak::Reg64 UseScratchGpr(Argument& arg);
Xbyak::Xmm UseScratchXmm(Argument& arg);
void UseScratch(Argument& arg, HostLoc host_loc);
```
At runtime, the value corresponding to `arg` will be placed a register. The actual register is determined by
which one of the above functions is called. `UseScratchGpr` places it in an unused GPR, `UseScratchXmm` places it
in an unused XMM register, and `UseScratch` allows you to specify a specific register (GPR or XMM) to use.
The return value is the register allocated to you.
You are free to modify the value in the register. The register is discarded at the end of the allocation scope.
### `Define` as register
A `Define` is the defintion of a value. This is the only time when a value may be set.
```c++
void DefineValue(IR::Inst* inst, const Xbyak::Reg& reg);
```
By calling `DefineValue`, you are stating that you wish to define the value for `inst`, and you have written the
value to the specified register `reg`.
### `Define`ing as an alias of a different value
Adding a `Define` to an existing value.
```c++
void DefineValue(IR::Inst* inst, Argument& arg);
```
You are declaring that the value for `inst` is the same as the value for `arg`. No host machine instructions are
emitted.
## When to use each?
* Prefer `Use` to `UseScratch` where possible.
* Prefer the `OpArg` variants where possible.
* Prefer to **not** use the specific `HostLoc` variants where possible.
# Return Stack Buffer Optimization (x64 Backend)
One of the optimizations that dynarmic does is block-linking. Block-linking is done when
the destination address of a jump is available at JIT-time. Instead of returning to the
dispatcher at the end of a block we can perform block-linking: just jump directly to the
next block. This is beneficial because returning to the dispatcher can often be quite
expensive.
What should we do in cases when we can't predict the destination address? The eponymous
example is when executing a return statement at the end of a function; the return address
is not statically known at compile time.
We deal with this by using a return stack buffer: When we execute a call instruction,
we push our prediction onto the RSB. When we execute a return instruction, we pop a
prediction off the RSB. If the prediction is a hit, we immediately jump to the relevant
compiled block. Otherwise, we return to the dispatcher.
This is the essential idea behind this optimization.
## `UniqueHash`
One complication dynarmic has is that a compiled block is not uniquely identifiable by
the PC alone, but bits in the FPSCR and CPSR are also relevant. We resolve this by
computing a 64-bit `UniqueHash` that is guaranteed to uniquely identify a block.
```c++
u64 LocationDescriptor::UniqueHash() const {
// This value MUST BE UNIQUE.
// This calculation has to match up with EmitX64::EmitTerminalPopRSBHint
u64 pc_u64 = u64(arm_pc) << 32;
u64 fpscr_u64 = u64(fpscr.Value());
u64 t_u64 = cpsr.T() ? 1 : 0;
u64 e_u64 = cpsr.E() ? 2 : 0;
return pc_u64 | fpscr_u64 | t_u64 | e_u64;
}
```
## Our implementation isn't actually a stack
Dynarmic's RSB isn't actually a stack. It was implemented as a ring buffer because
that showed better performance in tests.
### RSB Structure
The RSB is implemented as a ring buffer. `rsb_ptr` is the index of the insertion
point. Each element in `rsb_location_descriptors` is a `UniqueHash` and they
each correspond to an element in `rsb_codeptrs`. `rsb_codeptrs` contains the
host addresses for the corresponding the compiled blocks.
`RSBSize` was chosen by performance testing. Note that this is bigger than the
size of the real RSB in hardware (which has 3 entries). Larger RSBs than 8
showed degraded performance.
```c++
struct JitState {
// ...
static constexpr size_t RSBSize = 8; // MUST be a power of 2.
u32 rsb_ptr = 0;
std::array<u64, RSBSize> rsb_location_descriptors;
std::array<u64, RSBSize> rsb_codeptrs;
void ResetRSB();
// ...
};
```
### RSB Push
We insert our prediction at the insertion point iff the RSB doesn't already
contain a prediction with the same `UniqueHash`.
```c++
void EmitX64::EmitPushRSB(IR::Block&, IR::Inst* inst) {
using namespace Xbyak::util;
ASSERT(inst->GetArg(0).IsImmediate());
u64 imm64 = inst->GetArg(0).GetU64();
Xbyak::Reg64 code_ptr_reg = reg_alloc.ScratchGpr(code, {HostLoc::RCX});
Xbyak::Reg64 loc_desc_reg = reg_alloc.ScratchGpr(code);
Xbyak::Reg32 index_reg = reg_alloc.ScratchGpr(code).cvt32();
u64 code_ptr = unique_hash_to_code_ptr.find(imm64) != unique_hash_to_code_ptr.end()
? u64(unique_hash_to_code_ptr[imm64])
: u64(code->GetReturnFromRunCodeAddress());
code->mov(index_reg, dword[code.ABI_JIT_PTR + offsetof(JitState, rsb_ptr)]);
code->add(index_reg, 1);
code->and_(index_reg, u32(JitState::RSBSize - 1));
code->mov(loc_desc_reg, u64(imm64));
CodePtr patch_location = code->getCurr<CodePtr>();
patch_unique_hash_locations[imm64].emplace_back(patch_location);
code->mov(code_ptr_reg, u64(code_ptr)); // This line has to match up with EmitX64::Patch.
code->EnsurePatchLocationSize(patch_location, 10);
Xbyak::Label label;
for (size_t i = 0; i < JitState::RSBSize; ++i) {
code->cmp(loc_desc_reg, qword[code.ABI_JIT_PTR + offsetof(JitState, rsb_location_descriptors) + i * sizeof(u64)]);
code->je(label, code->T_SHORT);
}
code->mov(dword[code.ABI_JIT_PTR + offsetof(JitState, rsb_ptr)], index_reg);
code->mov(qword[code.ABI_JIT_PTR + index_reg.cvt64() * 8 + offsetof(JitState, rsb_location_descriptors)], loc_desc_reg);
code->mov(qword[code.ABI_JIT_PTR + index_reg.cvt64() * 8 + offsetof(JitState, rsb_codeptrs)], code_ptr_reg);
code->L(label);
}
```
In pseudocode:
```c++
for (i := 0 .. RSBSize-1)
if (rsb_location_descriptors[i] == imm64)
goto label;
rsb_ptr++;
rsb_ptr %= RSBSize;
rsb_location_desciptors[rsb_ptr] = imm64; //< The UniqueHash
rsb_codeptr[rsb_ptr] = /* codeptr corresponding to the UniqueHash */;
label:
```
## RSB Pop
To check if a predicition is in the RSB, we linearly scan the RSB.
```c++
void EmitX64::EmitTerminalPopRSBHint(IR::Term::PopRSBHint, IR::LocationDescriptor initial_location) {
using namespace Xbyak::util;
// This calculation has to match up with IREmitter::PushRSB
code->mov(ecx, MJitStateReg(Arm::Reg::PC));
code->shl(rcx, 32);
code->mov(ebx, dword[code.ABI_JIT_PTR + offsetof(JitState, FPSCR_mode)]);
code->or_(ebx, dword[code.ABI_JIT_PTR + offsetof(JitState, CPSR_et)]);
code->or_(rbx, rcx);
code->mov(rax, u64(code->GetReturnFromRunCodeAddress()));
for (size_t i = 0; i < JitState::RSBSize; ++i) {
code->cmp(rbx, qword[code.ABI_JIT_PTR + offsetof(JitState, rsb_location_descriptors) + i * sizeof(u64)]);
code->cmove(rax, qword[code.ABI_JIT_PTR + offsetof(JitState, rsb_codeptrs) + i * sizeof(u64)]);
}
code->jmp(rax);
}
```
In pseudocode:
```c++
rbx := ComputeUniqueHash()
rax := ReturnToDispatch
for (i := 0 .. RSBSize-1)
if (rbx == rsb_location_descriptors[i])
rax = rsb_codeptrs[i]
goto rax
```
# Fast memory (Fastmem)
The main way of accessing memory in JITed programs is via an invoked function, say "Read()" and "Write()". On our translator, such functions usually take a sizable amounts of code space (push + call + pop). Trash the i-cache (due to an indirect call) and overall make code emission more bloated.
The solution? Delegate invalid accesses to a dedicated arena, similar to a swap. The main idea behind such mechanism is to allow the OS to transmit page faults from invalid accesses into the JIT translator directly, bypassing address space calls, while this sacrifices i-cache coherency, it allows for smaller code-size and "faster" throguhput.
Many kernels however, do not support fast signal dispatching (Solaris, OpenBSD, FreeBSD). Only Linux and Windows support relatively "fast" signal dispatching. Hence this feature is better suited for them only.
![Host to guest translation](./HostToGuest.svg)
![Fastmem translation](./Fastmem.svg)
In x86_64 for example, when a page fault occurs, the CPU will transmit via control registers and the stack (see `IRETQ`) the appropriate arguments for a page fault handler, the OS then will transform that into something that can be sent into userspace.
Most modern OSes implement kernel-page-table-isolation, which means a set of system calls will invoke a context switch (not often used syscalls), whereas others are handled by the same process address space (the smaller kernel portion, often used syscalls) without needing a context switch. This effect can be negated on systems with PCID (up to 4096 unique IDs).
Signal dispatching takes a performance hit from reloading `%cr3` - but Linux does something more clever to avoid reloads: VDSO will take care of the entire thing in the same address space. Making dispatching as costly as an indirect call - without the hazards of increased code size.
The main downside from this is the constant i-cache trashing and pipeline hazards introduced by the VDSO signal handlers. However on most benchmarks fastmem does perform faster than without (Linux only). This also abuses the fact of continous address space emulation by using an arena - which can then be potentially transparently mapped into a hugepage, reducing TLB walk times.

View File

Before

Width:  |  Height:  |  Size: 128 KiB

After

Width:  |  Height:  |  Size: 128 KiB

View File

Before

Width:  |  Height:  |  Size: 98 KiB

After

Width:  |  Height:  |  Size: 98 KiB

View File

@@ -49,7 +49,7 @@ Important API Changes in v6.x Series
Documentation
-------------
Design documentation can be found at [docs/Design.md](docs/Design.md).
Design documentation can be found at [./Design.md](./Design.md).
Usage Example

View File

@@ -119,10 +119,10 @@
"package": "VulkanUtilityLibraries",
"repo": "scripts/VulkanUtilityHeaders",
"tag": "%VERSION%",
"git_version": "1.4.335",
"git_version": "1.4.328",
"artifact": "VulkanUtilityHeaders.tar.zst",
"git_host": "git.crueter.xyz",
"hash": "16dac0e6586702580c4279e4cd37ffe3cf909c93eb31b5069da7af36436d47b270a9cbaac953bb66c22ed12ed67ffa096688599267f307dfb62be1bc09f79833"
"hash": "9922217b39faf73cd4fc1510f2fdba14a49aa5c0d77f9ee24ee0512cef16b234d0cabc83c1fec861fa5df1d43e7f086ca9b6501753899119f39c5ca530cb0dae"
},
"spirv-tools": {
"package": "SPIRV-Tools",

View File

@@ -217,6 +217,7 @@ else()
--disable-ffmpeg
--disable-ffprobe
--disable-network
--disable-postproc
--disable-swresample
--enable-decoder=h264
--enable-decoder=vp8

View File

@@ -39,6 +39,7 @@ android {
buildFeatures {
viewBinding = true
buildConfig = true
}
compileOptions {

View File

@@ -14,8 +14,6 @@ android.useAndroidX=true
# Kotlin code style for this project: "official" or "obsolete":
kotlin.code.style=official
kotlin.parallel.tasks.in.project=true
android.defaults.buildfeatures.buildconfig=true
# Android Gradle plugin 8.0.2
android.suppressUnsupportedCompileSdk=34
android.native.buildOutput=verbose

View File

@@ -34,8 +34,8 @@ struct Member {
struct RoomInformation {
std::string name; ///< Name of the server
std::string description; ///< Server description
u32 member_slots{}; ///< Maximum number of members in this room
u16 port{}; ///< The port of this room
u32 member_slots; ///< Maximum number of members in this room
u16 port; ///< The port of this room
GameInfo preferred_game; ///< Game to advertise that you want to play
std::string host_username; ///< Forum username of the host
};
@@ -46,8 +46,8 @@ struct Room {
std::string id;
std::string verify_uid; ///< UID used for verification
std::string ip;
u32 net_version{};
bool has_password = false;
u32 net_version;
bool has_password;
std::vector<Member> members;
};

View File

@@ -21,7 +21,7 @@ namespace Core::Frontend {
struct CabinetParameters {
Service::NFP::TagInfo tag_info;
Service::NFP::RegisterInfo register_info;
Service::NFP::CabinetMode mode{};
Service::NFP::CabinetMode mode;
};
using CabinetCallback = std::function<void(bool, const std::string&)>;

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -23,9 +20,9 @@ struct KeyboardInitializeParameters {
std::u16string initial_text;
char16_t left_optional_symbol_key;
char16_t right_optional_symbol_key;
u32 max_text_length{};
u32 min_text_length{};
s32 initial_cursor_position{};
u32 max_text_length;
u32 min_text_length;
s32 initial_cursor_position;
Service::AM::Frontend::SwkbdType type;
Service::AM::Frontend::SwkbdPasswordMode password_mode;
Service::AM::Frontend::SwkbdTextDrawType text_draw_type;
@@ -37,12 +34,12 @@ struct KeyboardInitializeParameters {
};
struct InlineAppearParameters {
u32 max_text_length{};
u32 min_text_length{};
f32 key_top_scale_x{};
f32 key_top_scale_y{};
f32 key_top_translate_x{};
f32 key_top_translate_y{};
u32 max_text_length;
u32 min_text_length;
f32 key_top_scale_x;
f32 key_top_scale_y;
f32 key_top_translate_x;
f32 key_top_translate_y;
Service::AM::Frontend::SwkbdType type;
Service::AM::Frontend::SwkbdKeyDisableFlags key_disable_flags;
bool key_top_as_floating;
@@ -53,7 +50,7 @@ struct InlineAppearParameters {
struct InlineTextParameters {
std::u16string input_text;
s32 cursor_position{};
s32 cursor_position;
};
class SoftwareKeyboardApplet : public Applet {

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
@@ -82,7 +79,7 @@ using DeviceHandle = u64;
// This is nn::nfc::TagInfo
struct TagInfo {
UniqueSerialNumber uuid{};
UniqueSerialNumber uuid;
u8 uuid_length;
INSERT_PADDING_BYTES(0x15);
NfcProtocol protocol;

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
@@ -318,7 +315,7 @@ static_assert(sizeof(ModelInfo) == 0x40, "ModelInfo is an invalid size");
// This is nn::nfp::RegisterInfo
struct RegisterInfo {
Service::Mii::CharInfo mii_char_info;
WriteDate creation_date{};
WriteDate creation_date;
AmiiboName amiibo_name;
u8 font_region;
INSERT_PADDING_BYTES(0x7A);

View File

@@ -164,7 +164,7 @@ IPlatformServiceManager::IPlatformServiceManager(Core::System& system_, const ch
// Rebuild shared fonts from data ncas or synthesize
impl->shared_font = std::make_shared<Kernel::PhysicalMemory>(SHARED_FONT_MEM_SIZE);
for (auto& font : SHARED_FONTS) {
for (auto font : SHARED_FONTS) {
FileSys::VirtualFile romfs;
const auto nca =
nand->GetEntry(static_cast<u64>(font.first), FileSys::ContentRecordType::Data);
@@ -261,7 +261,7 @@ Result IPlatformServiceManager::GetSharedFontInOrderOfPriority(
out_font_sizes.size(), impl->shared_font_regions.size()});
for (size_t i = 0; i < max_size; i++) {
auto& region = impl->GetSharedFontRegion(i);
auto region = impl->GetSharedFontRegion(i);
out_font_codes[i] = static_cast<u32>(i);
out_font_offsets[i] = region.offset;

File diff suppressed because it is too large Load Diff

View File

@@ -1,19 +0,0 @@
# Fast memory (Fastmem)
The main way of accessing memory in JITed programs is via an invoked function, say "Read()" and "Write()". On our translator, such functions usually take a sizable amounts of code space (push + call + pop). Trash the i-cache (due to an indirect call) and overall make code emission more bloated.
The solution? Delegate invalid accesses to a dedicated arena, similar to a swap. The main idea behind such mechanism is to allow the OS to transmit page faults from invalid accesses into the JIT translator directly, bypassing address space calls, while this sacrifices i-cache coherency, it allows for smaller code-size and "faster" throguhput.
Many kernels however, do not support fast signal dispatching (Solaris, OpenBSD, FreeBSD). Only Linux and Windows support relatively "fast" signal dispatching. Hence this feature is better suited for them only.
![Host to guest translation](./HostToGuest.svg)
![Fastmem translation](./Fastmem.svg)
In x86_64 for example, when a page fault occurs, the CPU will transmit via control registers and the stack (see `IRETQ`) the appropriate arguments for a page fault handler, the OS then will transform that into something that can be sent into userspace.
Most modern OSes implement kernel-page-table-isolation, which means a set of system calls will invoke a context switch (not often used syscalls), whereas others are handled by the same process address space (the smaller kernel portion, often used syscalls) without needing a context switch. This effect can be negated on systems with PCID (up to 4096 unique IDs).
Signal dispatching takes a performance hit from reloading `%cr3` - but Linux does something more clever to avoid reloads: VDSO will take care of the entire thing in the same address space. Making dispatching as costly as an indirect call - without the hazards of increased code size.
The main downside from this is the constant i-cache trashing and pipeline hazards introduced by the VDSO signal handlers. However on most benchmarks fastmem does perform faster than without (Linux only). This also abuses the fact of continous address space emulation by using an arena - which can then be potentially transparently mapped into a hugepage, reducing TLB walk times.

View File

@@ -1,97 +0,0 @@
# Register Allocation (x64 Backend)
`HostLoc`s contain values. A `HostLoc` ("host value location") is either a host CPU register or a host spill location.
Values once set cannot be changed. Values can however be moved by the register allocator between `HostLoc`s. This is
handled by the register allocator itself and code that uses the register allocator need not and should not move values
between registers.
The register allocator is based on three concepts: `Use`, `Def` and `Scratch`.
* `Use`: The use of a value.
* `Define`: The definition of a value, this is the only time when a value is set.
* `Scratch`: Allocate a register that can be freely modified as one wishes.
Note that `Use`ing a value decrements its `use_count` by one. When the `use_count` reaches zero the value is discarded and no longer exists.
The member functions on `RegAlloc` are just a combination of the above concepts.
The following registers are reserved for internal use and should NOT participate in register allocation:
- `%xmm0`, `%xmm1`, `%xmm2`: Used as scratch in exclusive memory access.
- `%rsp`: Stack pointer.
- `%r15`: JIT pointer
- `%r14`: Page table pointer.
- `%r13`: Fastmem pointer.
The layout convenes `%r15` as the JIT state pointer - while it may be tempting to turn it into a synthetic pointer, keeping an entire register (out of 12 available) is preferable over inlining a directly computed immediate.
Do NEVER modify `%r15`, we must make it clear that this register is "immutable" for the entirety of the JIT block duration.
### `Scratch`
```c++
Xbyak::Reg64 ScratchGpr(HostLocList desired_locations = any_gpr);
Xbyak::Xmm ScratchXmm(HostLocList desired_locations = any_xmm);
```
At runtime, allocate one of the registers in `desired_locations`. You are free to modify the register. The register is discarded at the end of the allocation scope.
### Pure `Use`
```c++
Xbyak::Reg64 UseGpr(Argument& arg);
Xbyak::Xmm UseXmm(Argument& arg);
OpArg UseOpArg(Argument& arg);
void Use(Argument& arg, HostLoc host_loc);
```
At runtime, the value corresponding to `arg` will be placed a register. The actual register is determined by
which one of the above functions is called. `UseGpr` places it in an unused GPR, `UseXmm` places it
in an unused XMM register, `UseOpArg` might be in a register or might be a memory location, and `Use` allows
you to specify a specific register (GPR or XMM) to use.
This register **must not** have it's value changed.
### `UseScratch`
```c++
Xbyak::Reg64 UseScratchGpr(Argument& arg);
Xbyak::Xmm UseScratchXmm(Argument& arg);
void UseScratch(Argument& arg, HostLoc host_loc);
```
At runtime, the value corresponding to `arg` will be placed a register. The actual register is determined by
which one of the above functions is called. `UseScratchGpr` places it in an unused GPR, `UseScratchXmm` places it
in an unused XMM register, and `UseScratch` allows you to specify a specific register (GPR or XMM) to use.
The return value is the register allocated to you.
You are free to modify the value in the register. The register is discarded at the end of the allocation scope.
### `Define` as register
A `Define` is the defintion of a value. This is the only time when a value may be set.
```c++
void DefineValue(IR::Inst* inst, const Xbyak::Reg& reg);
```
By calling `DefineValue`, you are stating that you wish to define the value for `inst`, and you have written the
value to the specified register `reg`.
### `Define`ing as an alias of a different value
Adding a `Define` to an existing value.
```c++
void DefineValue(IR::Inst* inst, Argument& arg);
```
You are declaring that the value for `inst` is the same as the value for `arg`. No host machine instructions are
emitted.
## When to use each?
* Prefer `Use` to `UseScratch` where possible.
* Prefer the `OpArg` variants where possible.
* Prefer to **not** use the specific `HostLoc` variants where possible.

View File

@@ -1,157 +0,0 @@
# Return Stack Buffer Optimization (x64 Backend)
One of the optimizations that dynarmic does is block-linking. Block-linking is done when
the destination address of a jump is available at JIT-time. Instead of returning to the
dispatcher at the end of a block we can perform block-linking: just jump directly to the
next block. This is beneficial because returning to the dispatcher can often be quite
expensive.
What should we do in cases when we can't predict the destination address? The eponymous
example is when executing a return statement at the end of a function; the return address
is not statically known at compile time.
We deal with this by using a return stack buffer: When we execute a call instruction,
we push our prediction onto the RSB. When we execute a return instruction, we pop a
prediction off the RSB. If the prediction is a hit, we immediately jump to the relevant
compiled block. Otherwise, we return to the dispatcher.
This is the essential idea behind this optimization.
## `UniqueHash`
One complication dynarmic has is that a compiled block is not uniquely identifiable by
the PC alone, but bits in the FPSCR and CPSR are also relevant. We resolve this by
computing a 64-bit `UniqueHash` that is guaranteed to uniquely identify a block.
```c++
u64 LocationDescriptor::UniqueHash() const {
// This value MUST BE UNIQUE.
// This calculation has to match up with EmitX64::EmitTerminalPopRSBHint
u64 pc_u64 = u64(arm_pc) << 32;
u64 fpscr_u64 = u64(fpscr.Value());
u64 t_u64 = cpsr.T() ? 1 : 0;
u64 e_u64 = cpsr.E() ? 2 : 0;
return pc_u64 | fpscr_u64 | t_u64 | e_u64;
}
```
## Our implementation isn't actually a stack
Dynarmic's RSB isn't actually a stack. It was implemented as a ring buffer because
that showed better performance in tests.
### RSB Structure
The RSB is implemented as a ring buffer. `rsb_ptr` is the index of the insertion
point. Each element in `rsb_location_descriptors` is a `UniqueHash` and they
each correspond to an element in `rsb_codeptrs`. `rsb_codeptrs` contains the
host addresses for the corresponding the compiled blocks.
`RSBSize` was chosen by performance testing. Note that this is bigger than the
size of the real RSB in hardware (which has 3 entries). Larger RSBs than 8
showed degraded performance.
```c++
struct JitState {
// ...
static constexpr size_t RSBSize = 8; // MUST be a power of 2.
u32 rsb_ptr = 0;
std::array<u64, RSBSize> rsb_location_descriptors;
std::array<u64, RSBSize> rsb_codeptrs;
void ResetRSB();
// ...
};
```
### RSB Push
We insert our prediction at the insertion point iff the RSB doesn't already
contain a prediction with the same `UniqueHash`.
```c++
void EmitX64::EmitPushRSB(IR::Block&, IR::Inst* inst) {
using namespace Xbyak::util;
ASSERT(inst->GetArg(0).IsImmediate());
u64 imm64 = inst->GetArg(0).GetU64();
Xbyak::Reg64 code_ptr_reg = reg_alloc.ScratchGpr(code, {HostLoc::RCX});
Xbyak::Reg64 loc_desc_reg = reg_alloc.ScratchGpr(code);
Xbyak::Reg32 index_reg = reg_alloc.ScratchGpr(code).cvt32();
u64 code_ptr = unique_hash_to_code_ptr.find(imm64) != unique_hash_to_code_ptr.end()
? u64(unique_hash_to_code_ptr[imm64])
: u64(code->GetReturnFromRunCodeAddress());
code->mov(index_reg, dword[code.ABI_JIT_PTR + offsetof(JitState, rsb_ptr)]);
code->add(index_reg, 1);
code->and_(index_reg, u32(JitState::RSBSize - 1));
code->mov(loc_desc_reg, u64(imm64));
CodePtr patch_location = code->getCurr<CodePtr>();
patch_unique_hash_locations[imm64].emplace_back(patch_location);
code->mov(code_ptr_reg, u64(code_ptr)); // This line has to match up with EmitX64::Patch.
code->EnsurePatchLocationSize(patch_location, 10);
Xbyak::Label label;
for (size_t i = 0; i < JitState::RSBSize; ++i) {
code->cmp(loc_desc_reg, qword[code.ABI_JIT_PTR + offsetof(JitState, rsb_location_descriptors) + i * sizeof(u64)]);
code->je(label, code->T_SHORT);
}
code->mov(dword[code.ABI_JIT_PTR + offsetof(JitState, rsb_ptr)], index_reg);
code->mov(qword[code.ABI_JIT_PTR + index_reg.cvt64() * 8 + offsetof(JitState, rsb_location_descriptors)], loc_desc_reg);
code->mov(qword[code.ABI_JIT_PTR + index_reg.cvt64() * 8 + offsetof(JitState, rsb_codeptrs)], code_ptr_reg);
code->L(label);
}
```
In pseudocode:
```c++
for (i := 0 .. RSBSize-1)
if (rsb_location_descriptors[i] == imm64)
goto label;
rsb_ptr++;
rsb_ptr %= RSBSize;
rsb_location_desciptors[rsb_ptr] = imm64; //< The UniqueHash
rsb_codeptr[rsb_ptr] = /* codeptr corresponding to the UniqueHash */;
label:
```
## RSB Pop
To check if a predicition is in the RSB, we linearly scan the RSB.
```c++
void EmitX64::EmitTerminalPopRSBHint(IR::Term::PopRSBHint, IR::LocationDescriptor initial_location) {
using namespace Xbyak::util;
// This calculation has to match up with IREmitter::PushRSB
code->mov(ecx, MJitStateReg(Arm::Reg::PC));
code->shl(rcx, 32);
code->mov(ebx, dword[code.ABI_JIT_PTR + offsetof(JitState, FPSCR_mode)]);
code->or_(ebx, dword[code.ABI_JIT_PTR + offsetof(JitState, CPSR_et)]);
code->or_(rbx, rcx);
code->mov(rax, u64(code->GetReturnFromRunCodeAddress()));
for (size_t i = 0; i < JitState::RSBSize; ++i) {
code->cmp(rbx, qword[code.ABI_JIT_PTR + offsetof(JitState, rsb_location_descriptors) + i * sizeof(u64)]);
code->cmove(rax, qword[code.ABI_JIT_PTR + offsetof(JitState, rsb_codeptrs) + i * sizeof(u64)]);
}
code->jmp(rax);
}
```
In pseudocode:
```c++
rbx := ComputeUniqueHash()
rax := ReturnToDispatch
for (i := 0 .. RSBSize-1)
if (rbx == rsb_location_descriptors[i])
rax = rsb_codeptrs[i]
goto rax
```

View File

@@ -48,7 +48,7 @@ private:
void Save();
PlayTimeDatabase database;
u64 running_program_id{};
u64 running_program_id;
std::jthread play_time_thread;
};

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2017 Citra Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -56,7 +53,7 @@ struct ChatEntry {
/// Represents a system status message.
struct StatusMessageEntry {
StatusMessageTypes type{}; ///< Type of the message
StatusMessageTypes type; ///< Type of the message
/// Subject of the message. i.e. the user who is joining/leaving/being banned, etc.
std::string nickname;
std::string username;

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -545,9 +542,8 @@ Id EmitImageFetch(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id c
lod = Id{};
}
const ImageOperands operands(lod, ms);
return Emit(&EmitContext::OpImageSparseFetch, &EmitContext::OpImageFetch, ctx, inst,
ctx.F32[4], TextureImage(ctx, info, index), coords, operands.MaskOptional(),
operands.Span());
return Emit(&EmitContext::OpImageSparseFetch, &EmitContext::OpImageFetch, ctx, inst, ctx.F32[4],
TextureImage(ctx, info, index), coords, operands.MaskOptional(), operands.Span());
}
Id EmitImageQueryDimensions(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id lod,

View File

@@ -315,9 +315,6 @@ void DefineSsbos(EmitContext& ctx, StorageTypeDefinition& type_def,
ctx.Decorate(id, spv::Decoration::Binding, binding);
ctx.Decorate(id, spv::Decoration::DescriptorSet, 0U);
ctx.Name(id, fmt::format("ssbo{}", index));
if (!desc.is_written) {
ctx.Decorate(id, spv::Decoration::NonWritable);
}
if (ctx.profile.supported_spirv >= 0x00010400) {
ctx.interfaces.push_back(id);
}
@@ -1435,9 +1432,6 @@ void EmitContext::DefineInputs(const IR::Program& program) {
}
if (info.uses_sample_id) {
sample_id = DefineInput(*this, U32[1], false, spv::BuiltIn::SampleId);
if (stage == Stage::Fragment) {
Decorate(sample_id, spv::Decoration::Flat);
}
}
if (info.uses_is_helper_invocation) {
is_helper_invocation = DefineInput(*this, U1, false, spv::BuiltIn::HelperInvocation);
@@ -1448,13 +1442,6 @@ void EmitContext::DefineInputs(const IR::Program& program) {
subgroup_mask_le = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupLeMaskKHR);
subgroup_mask_gt = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupGtMaskKHR);
subgroup_mask_ge = DefineInput(*this, U32[4], false, spv::BuiltIn::SubgroupGeMaskKHR);
if (stage == Stage::Fragment) {
Decorate(subgroup_mask_eq, spv::Decoration::Flat);
Decorate(subgroup_mask_lt, spv::Decoration::Flat);
Decorate(subgroup_mask_le, spv::Decoration::Flat);
Decorate(subgroup_mask_gt, spv::Decoration::Flat);
Decorate(subgroup_mask_ge, spv::Decoration::Flat);
}
}
if (info.uses_fswzadd || info.uses_subgroup_invocation_id || info.uses_subgroup_shuffles ||
(profile.warp_size_potentially_larger_than_guest &&
@@ -1462,9 +1449,7 @@ void EmitContext::DefineInputs(const IR::Program& program) {
AddCapability(spv::Capability::GroupNonUniform);
subgroup_local_invocation_id =
DefineInput(*this, U32[1], false, spv::BuiltIn::SubgroupLocalInvocationId);
if (stage == Stage::Fragment) {
Decorate(subgroup_local_invocation_id, spv::Decoration::Flat);
}
Decorate(subgroup_local_invocation_id, spv::Decoration::Flat);
}
if (info.uses_fswzadd) {
const Id f32_one{Const(1.0f)};
@@ -1476,9 +1461,6 @@ void EmitContext::DefineInputs(const IR::Program& program) {
}
if (loads[IR::Attribute::PrimitiveId]) {
primitive_id = DefineInput(*this, U32[1], false, spv::BuiltIn::PrimitiveId);
if (stage == Stage::Fragment) {
Decorate(primitive_id, spv::Decoration::Flat);
}
}
if (loads[IR::Attribute::Layer]) {
AddCapability(spv::Capability::Geometry);
@@ -1570,21 +1552,17 @@ void EmitContext::DefineInputs(const IR::Program& program) {
if (stage != Stage::Fragment) {
continue;
}
const bool is_integer = input_type == AttributeType::SignedInt ||
input_type == AttributeType::UnsignedInt;
if (is_integer) {
switch (info.interpolation[index]) {
case Interpolation::Smooth:
// Default
// Decorate(id, spv::Decoration::Smooth);
break;
case Interpolation::NoPerspective:
Decorate(id, spv::Decoration::NoPerspective);
break;
case Interpolation::Flat:
Decorate(id, spv::Decoration::Flat);
} else {
switch (info.interpolation[index]) {
case Interpolation::Smooth:
break;
case Interpolation::NoPerspective:
Decorate(id, spv::Decoration::NoPerspective);
break;
case Interpolation::Flat:
Decorate(id, spv::Decoration::Flat);
break;
}
break;
}
}
if (stage == Stage::TessellationEval) {

View File

@@ -7,6 +7,7 @@
#pragma once
#include <algorithm>
#include <cstring>
#include <memory>
#include <numeric>
@@ -15,6 +16,8 @@
#include "video_core/guest_memory.h"
#include "video_core/host1x/gpu_device_memory_manager.h"
#include "video_core/texture_cache/util.h"
#include "video_core/polygon_mode_utils.h"
#include "video_core/renderer_vulkan/line_loop_utils.h"
namespace VideoCommon {
@@ -353,14 +356,37 @@ void BufferCache<P>::UpdateComputeBuffers() {
template <class P>
void BufferCache<P>::BindHostGeometryBuffers(bool is_indexed) {
const auto& draw_state = maxwell3d->draw_manager->GetDrawState();
if (is_indexed) {
BindHostIndexBuffer();
} else if constexpr (!HAS_FULL_INDEX_AND_PRIMITIVE_SUPPORT) {
const auto& draw_state = maxwell3d->draw_manager->GetDrawState();
if (draw_state.topology == Maxwell::PrimitiveTopology::Quads ||
draw_state.topology == Maxwell::PrimitiveTopology::QuadStrip) {
runtime.BindQuadIndexBuffer(draw_state.topology, draw_state.vertex_buffer.first,
draw_state.vertex_buffer.count);
} else {
if constexpr (!P::IS_OPENGL) {
const auto polygon_mode = VideoCore::EffectivePolygonMode(maxwell3d->regs);
if (draw_state.topology == Maxwell::PrimitiveTopology::Polygon &&
polygon_mode == Maxwell::PolygonMode::Line && draw_state.vertex_buffer.count > 1) {
const u32 vertex_count = draw_state.vertex_buffer.count;
const u32 generated_count = vertex_count + 1;
const bool use_u16 = vertex_count <= 0x10000;
const u32 element_size = use_u16 ? sizeof(u16) : sizeof(u32);
auto staging = runtime.UploadStagingBuffer(
static_cast<size_t>(generated_count) * element_size);
std::span<u8> dst_span{staging.mapped_span.data(),
generated_count * static_cast<size_t>(element_size)};
Vulkan::LineLoop::GenerateSequentialWithClosureRaw(dst_span, element_size);
const auto synthetic_format = use_u16 ? Maxwell::IndexFormat::UnsignedShort
: Maxwell::IndexFormat::UnsignedInt;
runtime.BindIndexBuffer(draw_state.topology, synthetic_format,
draw_state.vertex_buffer.first, generated_count,
staging.buffer, static_cast<u32>(staging.offset),
generated_count * element_size);
}
}
if constexpr (!HAS_FULL_INDEX_AND_PRIMITIVE_SUPPORT) {
if (draw_state.topology == Maxwell::PrimitiveTopology::Quads ||
draw_state.topology == Maxwell::PrimitiveTopology::QuadStrip) {
runtime.BindQuadIndexBuffer(draw_state.topology, draw_state.vertex_buffer.first,
draw_state.vertex_buffer.count);
}
}
}
BindHostVertexBuffers();
@@ -689,6 +715,44 @@ void BufferCache<P>::BindHostIndexBuffer() {
const u32 offset = buffer.Offset(channel_state->index_buffer.device_addr);
const u32 size = channel_state->index_buffer.size;
const auto& draw_state = maxwell3d->draw_manager->GetDrawState();
if constexpr (!P::IS_OPENGL) {
const auto polygon_mode = VideoCore::EffectivePolygonMode(maxwell3d->regs);
const bool polygon_line =
draw_state.topology == Maxwell::PrimitiveTopology::Polygon &&
polygon_mode == Maxwell::PolygonMode::Line;
if (polygon_line && draw_state.index_buffer.count > 1) {
const u32 element_size = draw_state.index_buffer.FormatSizeInBytes();
const size_t src_bytes = static_cast<size_t>(draw_state.index_buffer.count) * element_size;
const size_t total_bytes = src_bytes + element_size;
auto staging = runtime.UploadStagingBuffer(total_bytes);
std::span<u8> dst_span{staging.mapped_span.data(), total_bytes};
std::span<const u8> src_span;
if (!draw_state.inline_index_draw_indexes.empty()) {
const u8* const src =
draw_state.inline_index_draw_indexes.data() +
static_cast<size_t>(draw_state.index_buffer.first) * element_size;
src_span = {src, src_bytes};
} else if (const u8* const cpu_base =
device_memory.GetPointer<u8>(channel_state->index_buffer.device_addr)) {
const u8* const src = cpu_base +
static_cast<size_t>(draw_state.index_buffer.first) * element_size;
src_span = {src, src_bytes};
} else {
const DAddr src_addr =
channel_state->index_buffer.device_addr +
static_cast<DAddr>(draw_state.index_buffer.first) * element_size;
device_memory.ReadBlockUnsafe(src_addr, dst_span.data(), src_bytes);
src_span = {dst_span.data(), src_bytes};
}
Vulkan::LineLoop::CopyWithClosureRaw(dst_span, src_span, element_size);
buffer.MarkUsage(offset, size);
runtime.BindIndexBuffer(draw_state.topology, draw_state.index_buffer.format,
draw_state.index_buffer.first, draw_state.index_buffer.count + 1,
staging.buffer, static_cast<u32>(staging.offset),
static_cast<u32>(total_bytes));
return;
}
}
if (!draw_state.inline_index_draw_indexes.empty()) [[unlikely]] {
if constexpr (USE_MEMORY_MAPS_FOR_UPLOADS) {
auto upload_staging = runtime.UploadStagingBuffer(size);

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: 2021 yuzu Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
@@ -20,7 +17,7 @@ void Scheduler::Push(s32 channel, CommandList&& entries) {
std::unique_lock lk(scheduling_guard);
auto it = channels.find(channel);
ASSERT(it != channels.end());
auto& channel_state = it->second;
auto channel_state = it->second;
gpu.BindChannel(channel_state->bind_id);
channel_state->dma_pusher->Push(std::move(entries));
channel_state->dma_pusher->DispatchCalls();

View File

@@ -0,0 +1,46 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
#pragma once
#include "video_core/engines/maxwell_3d.h"
namespace VideoCore {
inline Tegra::Engines::Maxwell3D::Regs::PolygonMode EffectivePolygonMode(
const Tegra::Engines::Maxwell3D::Regs& regs) {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
const bool cull_enabled = regs.gl_cull_test_enabled != 0;
const auto cull_face = regs.gl_cull_face;
const bool cull_front = cull_enabled && (cull_face == Maxwell::CullFace::Front ||
cull_face == Maxwell::CullFace::FrontAndBack);
const bool cull_back = cull_enabled && (cull_face == Maxwell::CullFace::Back ||
cull_face == Maxwell::CullFace::FrontAndBack);
const bool render_front = !cull_front;
const bool render_back = !cull_back;
const auto front_mode = regs.polygon_mode_front;
const auto back_mode = regs.polygon_mode_back;
if (render_front && render_back && front_mode != back_mode) {
if (front_mode == Maxwell::PolygonMode::Line || back_mode == Maxwell::PolygonMode::Line) {
return Maxwell::PolygonMode::Line;
}
if (front_mode == Maxwell::PolygonMode::Point || back_mode == Maxwell::PolygonMode::Point) {
return Maxwell::PolygonMode::Point;
}
}
if (render_front) {
return front_mode;
}
if (render_back) {
return back_mode;
}
return front_mode;
}
} // namespace VideoCore

View File

@@ -525,24 +525,18 @@ BlitImageHelper::BlitImageHelper(const Device& device_, Scheduler& scheduler_,
nullptr, PUSH_CONSTANT_RANGE<VK_SHADER_STAGE_FRAGMENT_BIT, sizeof(float) * 4>))),
full_screen_vert(BuildShader(device, FULL_SCREEN_TRIANGLE_VERT_SPV)),
blit_color_to_color_frag(BuildShader(device, BLIT_COLOR_FLOAT_FRAG_SPV)),
blit_depth_stencil_frag(device.IsExtShaderStencilExportSupported()
? BuildShader(device, VULKAN_BLIT_DEPTH_STENCIL_FRAG_SPV)
: vk::ShaderModule{}),
blit_depth_stencil_frag(BuildShader(device, VULKAN_BLIT_DEPTH_STENCIL_FRAG_SPV)),
clear_color_vert(BuildShader(device, VULKAN_COLOR_CLEAR_VERT_SPV)),
clear_color_frag(BuildShader(device, VULKAN_COLOR_CLEAR_FRAG_SPV)),
clear_stencil_frag(BuildShader(device, VULKAN_DEPTHSTENCIL_CLEAR_FRAG_SPV)),
convert_depth_to_float_frag(BuildShader(device, CONVERT_DEPTH_TO_FLOAT_FRAG_SPV)),
convert_float_to_depth_frag(BuildShader(device, CONVERT_FLOAT_TO_DEPTH_FRAG_SPV)),
convert_abgr8_to_d24s8_frag(device.IsExtShaderStencilExportSupported()
? BuildShader(device, CONVERT_ABGR8_TO_D24S8_FRAG_SPV)
: vk::ShaderModule{}),
convert_abgr8_to_d24s8_frag(BuildShader(device, CONVERT_ABGR8_TO_D24S8_FRAG_SPV)),
convert_abgr8_to_d32f_frag(BuildShader(device, CONVERT_ABGR8_TO_D32F_FRAG_SPV)),
convert_d32f_to_abgr8_frag(BuildShader(device, CONVERT_D32F_TO_ABGR8_FRAG_SPV)),
convert_d24s8_to_abgr8_frag(BuildShader(device, CONVERT_D24S8_TO_ABGR8_FRAG_SPV)),
convert_s8d24_to_abgr8_frag(BuildShader(device, CONVERT_S8D24_TO_ABGR8_FRAG_SPV)),
convert_abgr8_srgb_to_d24s8_frag(device.IsExtShaderStencilExportSupported()
? BuildShader(device, CONVERT_ABGR8_SRGB_TO_D24S8_FRAG_SPV)
: vk::ShaderModule{}),
convert_abgr8_srgb_to_d24s8_frag(BuildShader(device, CONVERT_ABGR8_SRGB_TO_D24S8_FRAG_SPV)),
convert_rgba_to_bgra_frag(BuildShader(device, CONVERT_RGBA8_TO_BGRA8_FRAG_SPV)),
convert_yuv420_to_rgb_comp(BuildShader(device, CONVERT_YUV420_TO_RGB_COMP_SPV)),
convert_rgb_to_yuv420_comp(BuildShader(device, CONVERT_RGB_TO_YUV420_COMP_SPV)),
@@ -673,11 +667,6 @@ void BlitImageHelper::ConvertR16ToD16(const Framebuffer* dst_framebuffer,
void BlitImageHelper::ConvertABGR8ToD24S8(const Framebuffer* dst_framebuffer,
const ImageView& src_image_view) {
if (!device.IsExtShaderStencilExportSupported()) {
// Shader requires VK_EXT_shader_stencil_export which is not available
LOG_WARNING(Render_Vulkan, "ConvertABGR8ToD24S8 requires shader_stencil_export, skipping");
return;
}
ConvertPipelineDepthTargetEx(convert_abgr8_to_d24s8_pipeline, dst_framebuffer->RenderPass(),
convert_abgr8_to_d24s8_frag);
Convert(*convert_abgr8_to_d24s8_pipeline, dst_framebuffer, src_image_view);
@@ -713,11 +702,6 @@ void BlitImageHelper::ConvertS8D24ToABGR8(const Framebuffer* dst_framebuffer,
void BlitImageHelper::ConvertABGR8SRGBToD24S8(const Framebuffer* dst_framebuffer,
const ImageView& src_image_view) {
if (!device.IsExtShaderStencilExportSupported()) {
// Shader requires VK_EXT_shader_stencil_export which is not available
LOG_WARNING(Render_Vulkan, "ConvertABGR8SRGBToD24S8 requires shader_stencil_export, skipping");
return;
}
ConvertPipelineDepthTargetEx(convert_abgr8_srgb_to_d24s8_pipeline,
dst_framebuffer->RenderPass(),
convert_abgr8_srgb_to_d24s8_frag);

View File

@@ -15,6 +15,7 @@
#include "video_core/engines/draw_manager.h"
#include "video_core/renderer_vulkan/fixed_pipeline_state.h"
#include "video_core/renderer_vulkan/vk_state_tracker.h"
#include "video_core/polygon_mode_utils.h"
namespace Vulkan {
namespace {
@@ -59,13 +60,13 @@ void FixedPipelineState::Refresh(Tegra::Engines::Maxwell3D& maxwell3d, DynamicFe
raw1 = 0;
extended_dynamic_state.Assign(features.has_extended_dynamic_state ? 1 : 0);
extended_dynamic_state_2.Assign(features.has_extended_dynamic_state_2 ? 1 : 0);
extended_dynamic_state_2_logic_op.Assign(features.has_extended_dynamic_state_2_logic_op ? 1 : 0);
extended_dynamic_state_2_extra.Assign(features.has_extended_dynamic_state_2_extra ? 1 : 0);
extended_dynamic_state_3_blend.Assign(features.has_extended_dynamic_state_3_blend ? 1 : 0);
extended_dynamic_state_3_enables.Assign(features.has_extended_dynamic_state_3_enables ? 1 : 0);
dynamic_vertex_input.Assign(features.has_dynamic_vertex_input ? 1 : 0);
xfb_enabled.Assign(regs.transform_feedback_enabled != 0);
ndc_minus_one_to_one.Assign(regs.depth_mode == Maxwell::DepthMode::MinusOneToOne ? 1 : 0);
polygon_mode.Assign(PackPolygonMode(regs.polygon_mode_front));
polygon_mode.Assign(PackPolygonMode(VideoCore::EffectivePolygonMode(regs)));
tessellation_primitive.Assign(static_cast<u32>(regs.tessellation.params.domain_type.Value()));
tessellation_spacing.Assign(static_cast<u32>(regs.tessellation.params.spacing.Value()));
tessellation_clockwise.Assign(regs.tessellation.params.output_primitives.Value() ==
@@ -157,7 +158,7 @@ void FixedPipelineState::Refresh(Tegra::Engines::Maxwell3D& maxwell3d, DynamicFe
return static_cast<u16>(array.stride.Value());
});
}
if (!extended_dynamic_state_2_logic_op) {
if (!extended_dynamic_state_2_extra) {
dynamic_state.Refresh2(regs, topology_, extended_dynamic_state_2);
}
if (!extended_dynamic_state_3_blend) {

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -23,8 +20,7 @@ using Maxwell = Tegra::Engines::Maxwell3D::Regs;
struct DynamicFeatures {
bool has_extended_dynamic_state;
bool has_extended_dynamic_state_2;
bool has_extended_dynamic_state_2_logic_op;
bool has_extended_dynamic_state_2_patch_control_points;
bool has_extended_dynamic_state_2_extra;
bool has_extended_dynamic_state_3_blend;
bool has_extended_dynamic_state_3_enables;
bool has_dynamic_vertex_input;
@@ -190,7 +186,7 @@ struct FixedPipelineState {
u32 raw1;
BitField<0, 1, u32> extended_dynamic_state;
BitField<1, 1, u32> extended_dynamic_state_2;
BitField<2, 1, u32> extended_dynamic_state_2_logic_op;
BitField<2, 1, u32> extended_dynamic_state_2_extra;
BitField<3, 1, u32> extended_dynamic_state_3_blend;
BitField<4, 1, u32> extended_dynamic_state_3_enables;
BitField<5, 1, u32> dynamic_vertex_input;

View File

@@ -0,0 +1,68 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
#pragma once
#include <algorithm>
#include <cstring>
#include <span>
#include "common/assert.h"
#include "common/common_types.h"
namespace Vulkan::LineLoop {
inline void CopyWithClosureRaw(std::span<u8> dst, std::span<const u8> src, size_t element_size) {
ASSERT_MSG(dst.size() == src.size() + element_size, "Invalid line loop copy sizes");
if (src.empty()) {
if (!dst.empty()) {
std::fill(dst.begin(), dst.end(), u8{0});
}
return;
}
std::memcpy(dst.data(), src.data(), src.size());
std::memcpy(dst.data() + src.size(), src.data(), element_size);
}
inline void GenerateSequentialWithClosureRaw(std::span<u8> dst, size_t element_size,
u64 start_value = 0) {
if (dst.empty()) {
return;
}
const size_t last = dst.size() - element_size;
size_t offset = 0;
u64 value = start_value;
while (offset < last) {
std::memcpy(dst.data() + offset, &value, element_size);
offset += element_size;
++value;
}
std::memcpy(dst.data() + offset, &start_value, element_size);
}
template <typename T>
inline void CopyWithClosure(std::span<T> dst, std::span<const T> src) {
ASSERT_MSG(dst.size() == src.size() + 1, "Invalid destination size for line loop copy");
if (src.empty()) {
if (!dst.empty()) {
dst.front() = {};
}
return;
}
std::copy(src.begin(), src.end(), dst.begin());
dst.back() = src.front();
}
template <typename T>
inline void GenerateSequentialWithClosure(std::span<T> dst, T start_value = {}) {
if (dst.empty()) {
return;
}
const size_t last = dst.size() - 1;
for (size_t i = 0; i < last; ++i) {
dst[i] = static_cast<T>(start_value + static_cast<T>(i));
}
dst.back() = start_value;
}
} // namespace Vulkan::LineLoop

View File

@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -323,44 +326,9 @@ VkShaderStageFlagBits ShaderStage(Shader::Stage stage) {
}
VkPrimitiveTopology PrimitiveTopology([[maybe_unused]] const Device& device,
Maxwell::PrimitiveTopology topology) {
switch (topology) {
case Maxwell::PrimitiveTopology::Points:
return VK_PRIMITIVE_TOPOLOGY_POINT_LIST;
case Maxwell::PrimitiveTopology::Lines:
return VK_PRIMITIVE_TOPOLOGY_LINE_LIST;
case Maxwell::PrimitiveTopology::LineLoop:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
case Maxwell::PrimitiveTopology::LineStrip:
return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP;
case Maxwell::PrimitiveTopology::Triangles:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
case Maxwell::PrimitiveTopology::TriangleStrip:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP;
case Maxwell::PrimitiveTopology::TriangleFan:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN;
case Maxwell::PrimitiveTopology::LinesAdjacency:
return VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY;
case Maxwell::PrimitiveTopology::LineStripAdjacency:
return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_WITH_ADJACENCY;
case Maxwell::PrimitiveTopology::TrianglesAdjacency:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY;
case Maxwell::PrimitiveTopology::TriangleStripAdjacency:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_WITH_ADJACENCY;
case Maxwell::PrimitiveTopology::Quads:
case Maxwell::PrimitiveTopology::QuadStrip:
// TODO: Use VK_PRIMITIVE_TOPOLOGY_QUAD_LIST_EXT/VK_PRIMITIVE_TOPOLOGY_QUAD_STRIP_EXT
// whenever it releases
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
case Maxwell::PrimitiveTopology::Patches:
return VK_PRIMITIVE_TOPOLOGY_PATCH_LIST;
case Maxwell::PrimitiveTopology::Polygon:
LOG_WARNING(Render_Vulkan, "Draw mode is Polygon with a polygon mode of lines should be a "
"single body and not a bunch of triangles.");
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN;
}
UNIMPLEMENTED_MSG("Unimplemented topology={}", topology);
return {};
Maxwell::PrimitiveTopology topology,
Maxwell::PolygonMode polygon_mode) {
return detail::PrimitiveTopologyNoDevice(topology, polygon_mode);
}
VkFormat VertexFormat(const Device& device, Maxwell::VertexAttribute::Type type,

View File

@@ -1,3 +1,6 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -15,6 +18,52 @@ namespace Vulkan::MaxwellToVK {
using Maxwell = Tegra::Engines::Maxwell3D::Regs;
using PixelFormat = VideoCore::Surface::PixelFormat;
namespace detail {
constexpr VkPrimitiveTopology PrimitiveTopologyNoDevice(Maxwell::PrimitiveTopology topology,
Maxwell::PolygonMode polygon_mode) {
switch (topology) {
case Maxwell::PrimitiveTopology::Points:
return VK_PRIMITIVE_TOPOLOGY_POINT_LIST;
case Maxwell::PrimitiveTopology::Lines:
return VK_PRIMITIVE_TOPOLOGY_LINE_LIST;
case Maxwell::PrimitiveTopology::LineLoop:
return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP;
case Maxwell::PrimitiveTopology::LineStrip:
return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP;
case Maxwell::PrimitiveTopology::Triangles:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
case Maxwell::PrimitiveTopology::TriangleStrip:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP;
case Maxwell::PrimitiveTopology::TriangleFan:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN;
case Maxwell::PrimitiveTopology::LinesAdjacency:
return VK_PRIMITIVE_TOPOLOGY_LINE_LIST_WITH_ADJACENCY;
case Maxwell::PrimitiveTopology::LineStripAdjacency:
return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP_WITH_ADJACENCY;
case Maxwell::PrimitiveTopology::TrianglesAdjacency:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST_WITH_ADJACENCY;
case Maxwell::PrimitiveTopology::TriangleStripAdjacency:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP_WITH_ADJACENCY;
case Maxwell::PrimitiveTopology::Quads:
case Maxwell::PrimitiveTopology::QuadStrip:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
case Maxwell::PrimitiveTopology::Patches:
return VK_PRIMITIVE_TOPOLOGY_PATCH_LIST;
case Maxwell::PrimitiveTopology::Polygon:
switch (polygon_mode) {
case Maxwell::PolygonMode::Fill:
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN;
case Maxwell::PolygonMode::Line:
return VK_PRIMITIVE_TOPOLOGY_LINE_STRIP;
case Maxwell::PolygonMode::Point:
return VK_PRIMITIVE_TOPOLOGY_POINT_LIST;
}
break;
}
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
}
} // namespace detail
namespace Sampler {
VkFilter Filter(Tegra::Texture::TextureFilter filter);
@@ -46,7 +95,8 @@ struct FormatInfo {
VkShaderStageFlagBits ShaderStage(Shader::Stage stage);
VkPrimitiveTopology PrimitiveTopology(const Device& device, Maxwell::PrimitiveTopology topology);
VkPrimitiveTopology PrimitiveTopology(const Device& device, Maxwell::PrimitiveTopology topology,
Maxwell::PolygonMode polygon_mode);
VkFormat VertexFormat(const Device& device, Maxwell::VertexAttribute::Type type,
Maxwell::VertexAttribute::Size size);

View File

@@ -583,9 +583,7 @@ void BufferCacheRuntime::BindVertexBuffer(u32 index, VkBuffer buffer, u32 offset
if (index >= device.GetMaxVertexInputBindings()) {
return;
}
// Use BindVertexBuffers2EXT only if EDS1 is supported AND VIDS is not active
// When VIDS is active, the pipeline doesn't declare VERTEX_INPUT_BINDING_STRIDE as dynamic
if (device.IsExtExtendedDynamicStateSupported() && !device.IsExtVertexInputDynamicStateSupported()) {
if (device.IsExtExtendedDynamicStateSupported()) {
scheduler.Record([index, buffer, offset, size, stride](vk::CommandBuffer cmdbuf) {
const VkDeviceSize vk_offset = buffer != VK_NULL_HANDLE ? offset : 0;
const VkDeviceSize vk_size = buffer != VK_NULL_HANDLE ? size : VK_WHOLE_SIZE;
@@ -625,8 +623,7 @@ void BufferCacheRuntime::BindVertexBuffers(VideoCommon::HostBindings<Buffer>& bi
if (binding_count == 0) {
return;
}
// Use BindVertexBuffers2EXT only if EDS1 is supported AND VIDS is not active
if (device.IsExtExtendedDynamicStateSupported() && !device.IsExtVertexInputDynamicStateSupported()) {
if (device.IsExtExtendedDynamicStateSupported()) {
scheduler.Record([bindings_ = std::move(bindings),
buffer_handles_ = std::move(buffer_handles),
binding_count](vk::CommandBuffer cmdbuf) {

View File

@@ -418,9 +418,6 @@ ConditionalRenderingResolvePass::ConditionalRenderingResolvePass(
void ConditionalRenderingResolvePass::Resolve(VkBuffer dst_buffer, VkBuffer src_buffer,
u32 src_offset, bool compare_to_zero) {
if (!device.IsExtConditionalRendering()) {
return;
}
const size_t compare_size = compare_to_zero ? 8 : 24;
compute_pass_descriptor_queue.Acquire();
@@ -451,7 +448,7 @@ void ConditionalRenderingResolvePass::Resolve(VkBuffer dst_buffer, VkBuffer src_
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *layout, 0, set, {});
cmdbuf.Dispatch(1, 1, 1);
cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, 0, write_barrier);
VK_PIPELINE_STAGE_CONDITIONAL_RENDERING_BIT_EXT, 0, write_barrier);
});
}
@@ -473,14 +470,6 @@ QueriesPrefixScanPass::QueriesPrefixScanPass(
void QueriesPrefixScanPass::Run(VkBuffer accumulation_buffer, VkBuffer dst_buffer,
VkBuffer src_buffer, size_t number_of_sums,
size_t min_accumulation_limit, size_t max_accumulation_limit) {
constexpr VkAccessFlags BASE_DST_ACCESS = VK_ACCESS_SHADER_READ_BIT |
VK_ACCESS_TRANSFER_READ_BIT |
VK_ACCESS_VERTEX_ATTRIBUTE_READ_BIT |
VK_ACCESS_INDIRECT_COMMAND_READ_BIT |
VK_ACCESS_INDEX_READ_BIT |
VK_ACCESS_UNIFORM_READ_BIT;
const VkAccessFlags conditional_access =
device.IsExtConditionalRendering() ? VK_ACCESS_CONDITIONAL_RENDERING_READ_BIT_EXT : 0;
size_t current_runs = number_of_sums;
size_t offset = 0;
while (current_runs != 0) {
@@ -497,18 +486,22 @@ void QueriesPrefixScanPass::Run(VkBuffer accumulation_buffer, VkBuffer dst_buffe
scheduler.RequestOutsideRenderPassOperationContext();
scheduler.Record([this, descriptor_data, min_accumulation_limit, max_accumulation_limit,
runs_to_do, used_offset, conditional_access](vk::CommandBuffer cmdbuf) {
runs_to_do, used_offset](vk::CommandBuffer cmdbuf) {
static constexpr VkMemoryBarrier read_barrier{
.sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER,
.pNext = nullptr,
.srcAccessMask = VK_ACCESS_TRANSFER_WRITE_BIT,
.dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT,
};
const VkMemoryBarrier write_barrier{
static constexpr VkMemoryBarrier write_barrier{
.sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER,
.pNext = nullptr,
.srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
.dstAccessMask = BASE_DST_ACCESS | conditional_access,
.dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_TRANSFER_READ_BIT |
VK_ACCESS_VERTEX_ATTRIBUTE_READ_BIT |
VK_ACCESS_INDIRECT_COMMAND_READ_BIT | VK_ACCESS_INDEX_READ_BIT |
VK_ACCESS_UNIFORM_READ_BIT |
VK_ACCESS_CONDITIONAL_RENDERING_READ_BIT_EXT,
};
const QueriesPrefixScanPushConstants uniforms{
.min_accumulation_base = static_cast<u32>(min_accumulation_limit),
@@ -526,7 +519,8 @@ void QueriesPrefixScanPass::Run(VkBuffer accumulation_buffer, VkBuffer dst_buffe
cmdbuf.PushConstants(*layout, VK_SHADER_STAGE_COMPUTE_BIT, uniforms);
cmdbuf.Dispatch(1, 1, 1);
cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, 0, write_barrier);
VK_PIPELINE_STAGE_CONDITIONAL_RENDERING_BIT_EXT, 0,
write_barrier);
});
}
}

View File

@@ -7,6 +7,7 @@
#include <algorithm>
#include <iostream>
#include <span>
#include <string_view>
#include <boost/container/small_vector.hpp>
#include <boost/container/static_vector.hpp>
@@ -22,6 +23,7 @@
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_texture_cache.h"
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/polygon_mode_utils.h"
#include "video_core/shader_notify.h"
#include "video_core/texture_cache/texture_cache.h"
#include "video_core/vulkan_common/vulkan_device.h"
@@ -263,7 +265,6 @@ GraphicsPipeline::GraphicsPipeline(
std::ranges::copy(info->constant_buffer_used_sizes, uniform_buffer_sizes[stage].begin());
num_textures += Shader::NumDescriptors(info->texture_descriptors);
}
fragment_has_color0_output = stage_infos[NUM_STAGES - 1].stores_frag_color[0];
auto func{[this, shader_notify, &render_pass_cache, &descriptor_pool, pipeline_statistics] {
DescriptorLayoutBuilder builder{MakeBuilder(device, stage_infos)};
uses_push_descriptor = builder.CanUsePushDescriptor();
@@ -615,7 +616,10 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
vertex_input_ci.pNext = &input_divisor_ci;
}
const bool has_tess_stages = spv_modules[1] || spv_modules[2];
auto input_assembly_topology = MaxwellToVK::PrimitiveTopology(device, key.state.topology);
const auto polygon_mode =
FixedPipelineState::UnpackPolygonMode(key.state.polygon_mode.Value());
auto input_assembly_topology =
MaxwellToVK::PrimitiveTopology(device, key.state.topology, polygon_mode);
if (input_assembly_topology == VK_PRIMITIVE_TOPOLOGY_PATCH_LIST) {
if (!has_tess_stages) {
LOG_WARNING(Render_Vulkan, "Patch topology used without tessellation, using points");
@@ -630,6 +634,33 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
input_assembly_topology = VK_PRIMITIVE_TOPOLOGY_PATCH_LIST;
}
}
if (key.state.topology == Maxwell::PrimitiveTopology::Polygon) {
const auto polygon_mode_name = [polygon_mode]() -> std::string_view {
switch (polygon_mode) {
case Maxwell::PolygonMode::Fill:
return "Fill";
case Maxwell::PolygonMode::Line:
return "Line";
case Maxwell::PolygonMode::Point:
return "Point";
}
return "Unknown";
}();
const auto vk_topology_name = [input_assembly_topology]() -> std::string_view {
switch (input_assembly_topology) {
case VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN:
return "TriangleFan";
case VK_PRIMITIVE_TOPOLOGY_LINE_STRIP:
return "LineStrip";
case VK_PRIMITIVE_TOPOLOGY_POINT_LIST:
return "PointList";
default:
return "Unexpected";
}
}();
LOG_DEBUG(Render_Vulkan, "Polygon primitive in {} mode mapped to {}", polygon_mode_name,
vk_topology_name);
}
const VkPipelineInputAssemblyStateCreateInfo input_assembly_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
.pNext = nullptr,
@@ -703,18 +734,13 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
.lineWidth = 1.0f,
// TODO(alekpop): Transfer from regs
};
const bool smooth_lines_supported =
device.IsExtLineRasterizationSupported() && device.SupportsSmoothLines();
const bool stippled_lines_supported =
device.IsExtLineRasterizationSupported() && device.SupportsStippledRectangularLines();
VkPipelineRasterizationLineStateCreateInfoEXT line_state{
.sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_LINE_STATE_CREATE_INFO_EXT,
.pNext = nullptr,
.lineRasterizationMode = key.state.smooth_lines != 0 && smooth_lines_supported
.lineRasterizationMode = key.state.smooth_lines != 0
? VK_LINE_RASTERIZATION_MODE_RECTANGULAR_SMOOTH_EXT
: VK_LINE_RASTERIZATION_MODE_RECTANGULAR_EXT,
.stippledLineEnable =
(dynamic.line_stipple_enable && stippled_lines_supported) ? VK_TRUE : VK_FALSE,
.stippledLineEnable = dynamic.line_stipple_enable ? VK_TRUE : VK_FALSE,
.lineStippleFactor = key.state.line_stipple_factor,
.lineStipplePattern = static_cast<uint16_t>(key.state.line_stipple_pattern),
};
@@ -745,8 +771,6 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
provoking_vertex.pNext = std::exchange(rasterization_ci.pNext, &provoking_vertex);
}
const bool supports_alpha_output = fragment_has_color0_output;
const bool alpha_to_one_supported = device.SupportsAlphaToOne();
const VkPipelineMultisampleStateCreateInfo multisample_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
.pNext = nullptr,
@@ -755,10 +779,8 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
.sampleShadingEnable = Settings::values.sample_shading.GetValue() ? VK_TRUE : VK_FALSE,
.minSampleShading = static_cast<float>(Settings::values.sample_shading_fraction.GetValue()) / 100.0f,
.pSampleMask = nullptr,
.alphaToCoverageEnable =
supports_alpha_output && key.state.alpha_to_coverage_enabled != 0 ? VK_TRUE : VK_FALSE,
.alphaToOneEnable = supports_alpha_output && alpha_to_one_supported &&
key.state.alpha_to_one_enabled != 0 ? VK_TRUE : VK_FALSE,
.alphaToCoverageEnable = key.state.alpha_to_coverage_enabled != 0 ? VK_TRUE : VK_FALSE,
.alphaToOneEnable = key.state.alpha_to_one_enabled != 0 ? VK_TRUE : VK_FALSE,
};
const VkPipelineDepthStencilStateCreateInfo depth_stencil_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO,
@@ -816,25 +838,14 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
.blendConstants = {}
};
static_vector<VkDynamicState, 34> dynamic_states{
VK_DYNAMIC_STATE_VIEWPORT,
VK_DYNAMIC_STATE_SCISSOR,
VK_DYNAMIC_STATE_DEPTH_BIAS,
VK_DYNAMIC_STATE_VIEWPORT, VK_DYNAMIC_STATE_SCISSOR,
VK_DYNAMIC_STATE_DEPTH_BIAS, VK_DYNAMIC_STATE_BLEND_CONSTANTS,
VK_DYNAMIC_STATE_DEPTH_BOUNDS, VK_DYNAMIC_STATE_STENCIL_COMPARE_MASK,
VK_DYNAMIC_STATE_STENCIL_WRITE_MASK, VK_DYNAMIC_STATE_STENCIL_REFERENCE,
VK_DYNAMIC_STATE_LINE_WIDTH,
};
if (device.UsesAdvancedCoreDynamicState()) {
static constexpr std::array core_dynamic_states{
VK_DYNAMIC_STATE_BLEND_CONSTANTS,
VK_DYNAMIC_STATE_DEPTH_BOUNDS,
VK_DYNAMIC_STATE_STENCIL_COMPARE_MASK,
VK_DYNAMIC_STATE_STENCIL_WRITE_MASK,
VK_DYNAMIC_STATE_STENCIL_REFERENCE,
};
dynamic_states.insert(dynamic_states.end(), core_dynamic_states.begin(),
core_dynamic_states.end());
}
if (key.state.extended_dynamic_state) {
static constexpr std::array extended{
std::vector<VkDynamicState> extended{
VK_DYNAMIC_STATE_CULL_MODE_EXT,
VK_DYNAMIC_STATE_FRONT_FACE_EXT,
VK_DYNAMIC_STATE_DEPTH_TEST_ENABLE_EXT,
@@ -844,68 +855,51 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
VK_DYNAMIC_STATE_STENCIL_TEST_ENABLE_EXT,
VK_DYNAMIC_STATE_STENCIL_OP_EXT,
};
if (!device.IsExtVertexInputDynamicStateSupported()) {
extended.push_back(VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT);
}
if (key.state.dynamic_vertex_input) {
dynamic_states.push_back(VK_DYNAMIC_STATE_VERTEX_INPUT_EXT);
}
dynamic_states.insert(dynamic_states.end(), extended.begin(), extended.end());
// VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT is part of EDS1
// Only use it if VIDS is not active (VIDS replaces it with full vertex input control)
if (!key.state.dynamic_vertex_input) {
dynamic_states.push_back(VK_DYNAMIC_STATE_VERTEX_INPUT_BINDING_STRIDE_EXT);
if (key.state.extended_dynamic_state_2) {
static constexpr std::array extended2{
VK_DYNAMIC_STATE_DEPTH_BIAS_ENABLE_EXT,
VK_DYNAMIC_STATE_PRIMITIVE_RESTART_ENABLE_EXT,
VK_DYNAMIC_STATE_RASTERIZER_DISCARD_ENABLE_EXT,
};
dynamic_states.insert(dynamic_states.end(), extended2.begin(), extended2.end());
}
}
// VK_DYNAMIC_STATE_VERTEX_INPUT_EXT (VIDS) - Independent from EDS
// Provides full dynamic vertex input control, replaces VERTEX_INPUT_BINDING_STRIDE
if (key.state.dynamic_vertex_input) {
dynamic_states.push_back(VK_DYNAMIC_STATE_VERTEX_INPUT_EXT);
}
// EDS2 - Core (3 states)
if (key.state.extended_dynamic_state_2) {
static constexpr std::array extended2{
VK_DYNAMIC_STATE_DEPTH_BIAS_ENABLE_EXT,
VK_DYNAMIC_STATE_PRIMITIVE_RESTART_ENABLE_EXT,
VK_DYNAMIC_STATE_RASTERIZER_DISCARD_ENABLE_EXT,
};
dynamic_states.insert(dynamic_states.end(), extended2.begin(), extended2.end());
}
// EDS2 - LogicOp (granular)
if (key.state.extended_dynamic_state_2_logic_op) {
dynamic_states.push_back(VK_DYNAMIC_STATE_LOGIC_OP_EXT);
}
// EDS3 - Blending (composite: 3 states)
if (key.state.extended_dynamic_state_3_blend) {
static constexpr std::array extended3{
VK_DYNAMIC_STATE_COLOR_BLEND_ENABLE_EXT,
VK_DYNAMIC_STATE_COLOR_BLEND_EQUATION_EXT,
VK_DYNAMIC_STATE_COLOR_WRITE_MASK_EXT,
};
dynamic_states.insert(dynamic_states.end(), extended3.begin(), extended3.end());
}
// EDS3 - Enables (composite: per-feature)
if (key.state.extended_dynamic_state_3_enables) {
if (device.SupportsDynamicState3DepthClampEnable()) {
dynamic_states.push_back(VK_DYNAMIC_STATE_DEPTH_CLAMP_ENABLE_EXT);
if (key.state.extended_dynamic_state_2_extra) {
dynamic_states.push_back(VK_DYNAMIC_STATE_LOGIC_OP_EXT);
}
if (device.SupportsDynamicState3LogicOpEnable()) {
dynamic_states.push_back(VK_DYNAMIC_STATE_LOGIC_OP_ENABLE_EXT);
if (key.state.extended_dynamic_state_3_blend) {
static constexpr std::array extended3{
VK_DYNAMIC_STATE_COLOR_BLEND_ENABLE_EXT,
VK_DYNAMIC_STATE_COLOR_BLEND_EQUATION_EXT,
VK_DYNAMIC_STATE_COLOR_WRITE_MASK_EXT,
// VK_DYNAMIC_STATE_COLOR_BLEND_ADVANCED_EXT,
};
dynamic_states.insert(dynamic_states.end(), extended3.begin(), extended3.end());
}
if (device.SupportsDynamicState3LineRasterizationMode()) {
dynamic_states.push_back(VK_DYNAMIC_STATE_LINE_RASTERIZATION_MODE_EXT);
}
if (device.SupportsDynamicState3ConservativeRasterizationMode()) {
dynamic_states.push_back(VK_DYNAMIC_STATE_CONSERVATIVE_RASTERIZATION_MODE_EXT);
}
if (device.SupportsDynamicState3LineStippleEnable()) {
dynamic_states.push_back(VK_DYNAMIC_STATE_LINE_STIPPLE_ENABLE_EXT);
}
if (device.SupportsDynamicState3AlphaToCoverageEnable()) {
dynamic_states.push_back(VK_DYNAMIC_STATE_ALPHA_TO_COVERAGE_ENABLE_EXT);
}
if (device.SupportsDynamicState3AlphaToOneEnable()) {
dynamic_states.push_back(VK_DYNAMIC_STATE_ALPHA_TO_ONE_ENABLE_EXT);
if (key.state.extended_dynamic_state_3_enables) {
static constexpr std::array extended3{
VK_DYNAMIC_STATE_DEPTH_CLAMP_ENABLE_EXT,
VK_DYNAMIC_STATE_LOGIC_OP_ENABLE_EXT,
// additional state3 extensions
VK_DYNAMIC_STATE_LINE_RASTERIZATION_MODE_EXT,
VK_DYNAMIC_STATE_CONSERVATIVE_RASTERIZATION_MODE_EXT,
VK_DYNAMIC_STATE_LINE_STIPPLE_ENABLE_EXT,
VK_DYNAMIC_STATE_ALPHA_TO_COVERAGE_ENABLE_EXT,
VK_DYNAMIC_STATE_ALPHA_TO_ONE_ENABLE_EXT,
VK_DYNAMIC_STATE_DEPTH_CLIP_ENABLE_EXT,
VK_DYNAMIC_STATE_PROVOKING_VERTEX_MODE_EXT,
};
dynamic_states.insert(dynamic_states.end(), extended3.begin(), extended3.end());
}
}

View File

@@ -82,17 +82,6 @@ public:
const std::array<const Shader::Info*, NUM_STAGES>& infos);
bool HasDynamicVertexInput() const noexcept { return key.state.dynamic_vertex_input; }
bool SupportsAlphaToCoverage() const noexcept {
return fragment_has_color0_output;
}
bool SupportsAlphaToOne() const noexcept {
return fragment_has_color0_output;
}
bool UsesExtendedDynamicState() const noexcept {
return key.state.extended_dynamic_state != 0;
}
GraphicsPipeline& operator=(GraphicsPipeline&&) noexcept = delete;
GraphicsPipeline(GraphicsPipeline&&) noexcept = delete;
@@ -160,7 +149,6 @@ private:
std::array<u32, 5> enabled_uniform_buffer_masks{};
VideoCommon::UniformBufferSizes uniform_buffer_sizes{};
u32 num_textures{};
bool fragment_has_color0_output{};
vk::DescriptorSetLayout descriptor_set_layout;
DescriptorAllocator descriptor_allocator;

View File

@@ -146,8 +146,7 @@ Shader::AttributeType AttributeType(const FixedPipelineState& state, size_t inde
Shader::RuntimeInfo MakeRuntimeInfo(std::span<const Shader::IR::Program> programs,
const GraphicsPipelineCacheKey& key,
const Shader::IR::Program& program,
const Shader::IR::Program* previous_program,
const Vulkan::Device& device) {
const Shader::IR::Program* previous_program) {
Shader::RuntimeInfo info;
if (previous_program) {
info.previous_stage_stores = previous_program->info.stores;
@@ -169,14 +168,10 @@ Shader::RuntimeInfo MakeRuntimeInfo(std::span<const Shader::IR::Program> program
info.fixed_state_point_size = point_size;
}
if (key.state.xfb_enabled) {
if (device.IsExtTransformFeedbackSupported()) {
auto [varyings, count] =
VideoCommon::MakeTransformFeedbackVaryings(key.state.xfb_state);
info.xfb_varyings = varyings;
info.xfb_count = count;
} else {
LOG_WARNING(Render_Vulkan, "XFB requested in pipeline key but device lacks VK_EXT_transform_feedback; ignoring XFB decorations");
}
auto [varyings, count] =
VideoCommon::MakeTransformFeedbackVaryings(key.state.xfb_state);
info.xfb_varyings = varyings;
info.xfb_count = count;
}
info.convert_depth_mode = gl_ndc;
}
@@ -223,14 +218,10 @@ Shader::RuntimeInfo MakeRuntimeInfo(std::span<const Shader::IR::Program> program
info.fixed_state_point_size = point_size;
}
if (key.state.xfb_enabled != 0) {
if (device.IsExtTransformFeedbackSupported()) {
auto [varyings, count] =
VideoCommon::MakeTransformFeedbackVaryings(key.state.xfb_state);
info.xfb_varyings = varyings;
info.xfb_count = count;
} else {
LOG_WARNING(Render_Vulkan, "XFB requested in pipeline key but device lacks VK_EXT_transform_feedback; ignoring XFB decorations");
}
auto [varyings, count] =
VideoCommon::MakeTransformFeedbackVaryings(key.state.xfb_state);
info.xfb_varyings = varyings;
info.xfb_count = count;
}
info.convert_depth_mode = gl_ndc;
break;
@@ -413,35 +404,14 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
device.GetMaxVertexInputBindings(), Maxwell::NumVertexArrays);
}
LOG_INFO(Render_Vulkan, "DynamicState setting value: {}", Settings::values.dyna_state.GetValue());
dynamic_features = {};
// User granularity enforced in vulkan_device.cpp switch statement:
// Level 0: Core Dynamic States only
// Level 1: Core + EDS1
// Level 2: Core + EDS1 + EDS2 (accumulative)
// Level 3: Core + EDS1 + EDS2 + EDS3 (accumulative)
// Here we only verify if extensions were successfully loaded by the device
dynamic_features.has_extended_dynamic_state =
device.IsExtExtendedDynamicStateSupported();
dynamic_features.has_extended_dynamic_state_2 =
device.IsExtExtendedDynamicState2Supported();
dynamic_features.has_extended_dynamic_state_2_logic_op =
device.IsExtExtendedDynamicState2ExtrasSupported();
dynamic_features.has_extended_dynamic_state_2_patch_control_points = false;
dynamic_features.has_extended_dynamic_state_3_blend =
device.IsExtExtendedDynamicState3BlendingSupported();
dynamic_features.has_extended_dynamic_state_3_enables =
device.IsExtExtendedDynamicState3EnablesSupported();
// VIDS: Independent toggle (not affected by dyna_state levels)
dynamic_features.has_dynamic_vertex_input =
device.IsExtVertexInputDynamicStateSupported() &&
Settings::values.vertex_input_dynamic_state.GetValue();
dynamic_features = DynamicFeatures{
.has_extended_dynamic_state = device.IsExtExtendedDynamicStateSupported(),
.has_extended_dynamic_state_2 = device.IsExtExtendedDynamicState2Supported(),
.has_extended_dynamic_state_2_extra = device.IsExtExtendedDynamicState2ExtrasSupported(),
.has_extended_dynamic_state_3_blend = device.IsExtExtendedDynamicState3BlendingSupported(),
.has_extended_dynamic_state_3_enables = device.IsExtExtendedDynamicState3EnablesSupported(),
.has_dynamic_vertex_input = device.IsExtVertexInputDynamicStateSupported(),
};
}
PipelineCache::~PipelineCache() {
@@ -546,8 +516,8 @@ void PipelineCache::LoadDiskResources(u64 title_id, std::stop_token stop_loading
dynamic_features.has_extended_dynamic_state ||
(key.state.extended_dynamic_state_2 != 0) !=
dynamic_features.has_extended_dynamic_state_2 ||
(key.state.extended_dynamic_state_2_logic_op != 0) !=
dynamic_features.has_extended_dynamic_state_2_logic_op ||
(key.state.extended_dynamic_state_2_extra != 0) !=
dynamic_features.has_extended_dynamic_state_2_extra ||
(key.state.extended_dynamic_state_3_blend != 0) !=
dynamic_features.has_extended_dynamic_state_3_blend ||
(key.state.extended_dynamic_state_3_enables != 0) !=
@@ -701,7 +671,7 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline(
const size_t stage_index{index - 1};
infos[stage_index] = &program.info;
const auto runtime_info{MakeRuntimeInfo(programs, key, program, previous_stage, device)};
const auto runtime_info{MakeRuntimeInfo(programs, key, program, previous_stage)};
ConvertLegacyToGeneric(program, runtime_info);
const std::vector<u32> code{EmitSPIRV(profile, runtime_info, program, binding, this->optimize_spirv_output)};
device.SaveShader(code);

View File

@@ -872,18 +872,17 @@ private:
return;
}
has_flushed_end_pending = true;
// Refresh buffers state before beginning transform feedback so counters are up-to-date
UpdateBuffers();
if (buffers_count == 0) {
// No counter buffers available: begin without counters
if (!has_started || buffers_count == 0) {
scheduler.Record([](vk::CommandBuffer cmdbuf) {
cmdbuf.BeginTransformFeedbackEXT(0, 0, nullptr, nullptr);
});
UpdateBuffers();
return;
}
scheduler.Record([this, total = static_cast<u32>(buffers_count)](vk::CommandBuffer cmdbuf) {
cmdbuf.BeginTransformFeedbackEXT(0, total, counter_buffers.data(), offsets.data());
});
UpdateBuffers();
}
void FlushEndTFB() {
@@ -893,15 +892,11 @@ private:
}
has_flushed_end_pending = false;
// Refresh buffer state before ending transform feedback to ensure counters_count is up-to-date.
UpdateBuffers();
if (buffers_count == 0) {
LOG_DEBUG(Render_Vulkan, "EndTransformFeedbackEXT called with no counters (buffers_count=0)");
scheduler.Record([](vk::CommandBuffer cmdbuf) {
cmdbuf.EndTransformFeedbackEXT(0, 0, nullptr, nullptr);
});
} else {
LOG_DEBUG(Render_Vulkan, "EndTransformFeedbackEXT called with counters (buffers_count={})", buffers_count);
scheduler.Record([this,
total = static_cast<u32>(buffers_count)](vk::CommandBuffer cmdbuf) {
cmdbuf.EndTransformFeedbackEXT(0, total, counter_buffers.data(), offsets.data());
@@ -912,7 +907,6 @@ private:
void UpdateBuffers() {
last_queries.fill(0);
last_queries_stride.fill(1);
streams_mask = 0; // reset previously recorded streams
runtime.View3DRegs([this](Maxwell3D& maxwell3d) {
buffers_count = 0;
out_topology = maxwell3d.draw_manager->GetDrawState().topology;
@@ -922,10 +916,6 @@ private:
continue;
}
const size_t stream = tf.controls[i].stream;
if (stream >= last_queries_stride.size()) {
LOG_WARNING(Render_Vulkan, "TransformFeedback stream {} out of range", stream);
continue;
}
last_queries_stride[stream] = tf.controls[i].stride;
streams_mask |= 1ULL << stream;
buffers_count = std::max<size_t>(buffers_count, stream + 1);
@@ -1126,21 +1116,16 @@ public:
query->flags |= VideoCommon::QueryFlagBits::IsFinalValueSynced;
u64 num_vertices = 0;
// Protect against stride == 0 (avoid divide-by-zero). Use fallback stride=1 and warn.
u64 safe_stride = query->stride == 0 ? 1 : query->stride;
if (query->stride == 0) {
LOG_WARNING(Render_Vulkan, "TransformFeedback query has stride 0; using 1 to avoid div-by-zero (addr=0x{:x})", query->dependant_address);
}
if (query->dependant_manage) {
auto* dependant_query = tfb_streamer.GetQuery(query->dependant_index);
num_vertices = dependant_query->value / safe_stride;
num_vertices = dependant_query->value / query->stride;
tfb_streamer.Free(query->dependant_index);
} else {
u8* pointer = device_memory.GetPointer<u8>(query->dependant_address);
if (pointer != nullptr) {
u32 result;
std::memcpy(&result, pointer, sizeof(u32));
num_vertices = static_cast<u64>(result) / safe_stride;
num_vertices = static_cast<u64>(result) / query->stride;
}
}
query->value = [&]() -> u64 {

View File

@@ -37,6 +37,7 @@
#include "video_core/renderer_vulkan/vk_update_descriptor.h"
#include "video_core/shader_cache.h"
#include "video_core/texture_cache/texture_cache_base.h"
#include "video_core/polygon_mode_utils.h"
#include "video_core/vulkan_common/vulkan_device.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
@@ -108,7 +109,7 @@ VkViewport GetViewportState(const Device& device, const Maxwell& regs, size_t in
VkRect2D GetScissorState(const Maxwell& regs, size_t index, u32 up_scale = 1, u32 down_shift = 0) {
const auto& src = regs.scissor_test[index];
VkRect2D scissor{};
VkRect2D scissor;
const auto scale_up = [&](s32 value) -> s32 {
if (value == 0) {
return 0U;
@@ -148,7 +149,8 @@ VkRect2D GetScissorState(const Maxwell& regs, size_t index, u32 up_scale = 1, u3
return scissor;
}
DrawParams MakeDrawParams(const MaxwellDrawState& draw_state, u32 num_instances, bool is_indexed) {
DrawParams MakeDrawParams(const MaxwellDrawState& draw_state, u32 num_instances, bool is_indexed,
Maxwell::PolygonMode polygon_mode) {
DrawParams params{
.base_instance = draw_state.base_instance,
.num_instances = num_instances,
@@ -168,6 +170,21 @@ DrawParams MakeDrawParams(const MaxwellDrawState& draw_state, u32 num_instances,
params.base_vertex = 0;
params.is_indexed = true;
}
const bool polygon_line =
draw_state.topology == Maxwell::PrimitiveTopology::Polygon &&
polygon_mode == Maxwell::PolygonMode::Line;
if (polygon_line) {
if (params.is_indexed) {
if (draw_state.index_buffer.count > 1) {
params.num_vertices = draw_state.index_buffer.count + 1;
}
} else if (draw_state.vertex_buffer.count > 1) {
params.num_vertices = draw_state.vertex_buffer.count + 1;
params.is_indexed = true;
params.first_index = 0;
params.base_vertex = draw_state.vertex_buffer.first;
}
}
return params;
}
} // Anonymous namespace
@@ -197,11 +214,6 @@ RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra
fence_manager(*this, gpu, texture_cache, buffer_cache, query_cache, device, scheduler),
wfi_event(device.GetLogical().CreateEvent()) {
scheduler.SetQueryCache(query_cache);
// Log multi-draw support
if (device.IsExtMultiDrawSupported()) {
LOG_INFO(Render_Vulkan, "VK_EXT_multi_draw is enabled for optimized draw calls");
}
}
RasterizerVulkan::~RasterizerVulkan() = default;
@@ -238,45 +250,18 @@ void RasterizerVulkan::Draw(bool is_indexed, u32 instance_count) {
PrepareDraw(is_indexed, [this, is_indexed, instance_count] {
const auto& draw_state = maxwell3d->draw_manager->GetDrawState();
const u32 num_instances{instance_count};
const DrawParams draw_params{MakeDrawParams(draw_state, num_instances, is_indexed)};
// Use VK_EXT_multi_draw if available (single draw becomes multi-draw with count=1)
if (device.IsExtMultiDrawSupported()) {
scheduler.Record([draw_params](vk::CommandBuffer cmdbuf) {
if (draw_params.is_indexed) {
// Use multi-draw indexed with single draw
const VkMultiDrawIndexedInfoEXT multi_draw_info{
.firstIndex = draw_params.first_index,
.indexCount = draw_params.num_vertices,
};
const int32_t vertex_offset = static_cast<int32_t>(draw_params.base_vertex);
cmdbuf.DrawMultiIndexedEXT(1, &multi_draw_info, draw_params.num_instances,
draw_params.base_instance,
sizeof(VkMultiDrawIndexedInfoEXT), &vertex_offset);
} else {
// Use multi-draw with single draw
const VkMultiDrawInfoEXT multi_draw_info{
.firstVertex = draw_params.base_vertex,
.vertexCount = draw_params.num_vertices,
};
cmdbuf.DrawMultiEXT(1, &multi_draw_info, draw_params.num_instances,
draw_params.base_instance,
sizeof(VkMultiDrawInfoEXT));
}
});
} else {
// Fallback to standard draw calls
scheduler.Record([draw_params](vk::CommandBuffer cmdbuf) {
if (draw_params.is_indexed) {
cmdbuf.DrawIndexed(draw_params.num_vertices, draw_params.num_instances,
draw_params.first_index, draw_params.base_vertex,
draw_params.base_instance);
} else {
cmdbuf.Draw(draw_params.num_vertices, draw_params.num_instances,
draw_params.base_vertex, draw_params.base_instance);
}
});
}
const auto polygon_mode = VideoCore::EffectivePolygonMode(maxwell3d->regs);
const DrawParams draw_params{MakeDrawParams(draw_state, num_instances, is_indexed, polygon_mode)};
scheduler.Record([draw_params](vk::CommandBuffer cmdbuf) {
if (draw_params.is_indexed) {
cmdbuf.DrawIndexed(draw_params.num_vertices, draw_params.num_instances,
draw_params.first_index, draw_params.base_vertex,
draw_params.base_instance);
} else {
cmdbuf.Draw(draw_params.num_vertices, draw_params.num_instances,
draw_params.base_vertex, draw_params.base_instance);
}
});
});
}
@@ -407,7 +392,7 @@ void RasterizerVulkan::Clear(u32 layer_count) {
}
UpdateViewportsState(regs);
VkRect2D default_scissor{};
VkRect2D default_scissor;
default_scissor.offset.x = 0;
default_scissor.offset.y = 0;
default_scissor.extent.width = (std::numeric_limits<s32>::max)();
@@ -419,48 +404,13 @@ void RasterizerVulkan::Clear(u32 layer_count) {
.baseArrayLayer = regs.clear_surface.layer,
.layerCount = layer_count,
};
const auto clamp_rect_to_render_area = [render_area](VkRect2D& rect) -> bool {
const auto clamp_axis = [](s32& offset, u32& extent, u32 limit) {
auto clamp_offset = [&offset, limit]() {
if (limit == 0) {
offset = 0;
return;
}
offset = std::clamp(offset, 0, static_cast<s32>(limit));
};
if (extent == 0) {
clamp_offset();
return;
}
if (offset < 0) {
const u32 shrink = (std::min)(extent, static_cast<u32>(-offset));
extent -= shrink;
offset = 0;
}
if (limit == 0) {
extent = 0;
offset = 0;
return;
}
if (offset >= static_cast<s32>(limit)) {
offset = static_cast<s32>(limit);
extent = 0;
return;
}
const u64 end_coord = static_cast<u64>(offset) + extent;
if (end_coord > limit) {
extent = limit - static_cast<u32>(offset);
}
};
clamp_axis(rect.offset.x, rect.extent.width, render_area.width);
clamp_axis(rect.offset.y, rect.extent.height, render_area.height);
return rect.extent.width != 0 && rect.extent.height != 0;
};
if (!clamp_rect_to_render_area(clear_rect.rect)) {
if (clear_rect.rect.extent.width == 0 || clear_rect.rect.extent.height == 0) {
return;
}
clear_rect.rect.extent = VkExtent2D{
.width = (std::min)(clear_rect.rect.extent.width, render_area.width),
.height = (std::min)(clear_rect.rect.extent.height, render_area.height),
};
const u32 color_attachment = regs.clear_surface.RT;
if (use_color && framebuffer->HasAspectColorBit(color_attachment)) {
@@ -907,21 +857,23 @@ void RasterizerVulkan::LoadDiskResources(u64 title_id, std::stop_token stop_load
void RasterizerVulkan::FlushWork() {
#ifdef ANDROID
static constexpr u32 DRAWS_TO_DISPATCH = 512;
static constexpr u32 CHECK_MASK = 3;
static constexpr u32 DRAWS_TO_DISPATCH = 1024;
#else
static constexpr u32 DRAWS_TO_DISPATCH = 4096;
static constexpr u32 CHECK_MASK = 7;
#endif // ANDROID
static_assert(DRAWS_TO_DISPATCH % (CHECK_MASK + 1) == 0);
if ((++draw_counter & CHECK_MASK) != CHECK_MASK) {
// Only check multiples of 8 draws
static_assert(DRAWS_TO_DISPATCH % 8 == 0);
if ((++draw_counter & 7) != 7) {
return;
}
if (draw_counter < DRAWS_TO_DISPATCH) {
// Send recorded tasks to the worker thread
scheduler.DispatchWork();
return;
}
// Otherwise (every certain number of draws) flush execution.
// This submits commands to the Vulkan driver.
scheduler.Flush();
draw_counter = 0;
}
@@ -987,8 +939,6 @@ bool AccelerateDMA::BufferToImage(const Tegra::DMA::ImageCopy& copy_info,
void RasterizerVulkan::UpdateDynamicStates() {
auto& regs = maxwell3d->regs;
// Core Dynamic States (Vulkan 1.0) - Always active regardless of dyna_state setting
UpdateViewportsState(regs);
UpdateScissorsState(regs);
UpdateDepthBias(regs);
@@ -996,8 +946,6 @@ void RasterizerVulkan::UpdateDynamicStates() {
UpdateDepthBounds(regs);
UpdateStencilFaces(regs);
UpdateLineWidth(regs);
// EDS1: CullMode, DepthCompare, FrontFace, StencilOp, DepthBoundsTest, DepthTest, DepthWrite, StencilTest
if (device.IsExtExtendedDynamicStateSupported()) {
UpdateCullMode(regs);
UpdateDepthCompareOp(regs);
@@ -1008,52 +956,40 @@ void RasterizerVulkan::UpdateDynamicStates() {
UpdateDepthTestEnable(regs);
UpdateDepthWriteEnable(regs);
UpdateStencilTestEnable(regs);
}
}
// EDS2: PrimitiveRestart, RasterizerDiscard, DepthBias enable/disable
if (device.IsExtExtendedDynamicState2Supported()) {
UpdatePrimitiveRestartEnable(regs);
UpdateRasterizerDiscardEnable(regs);
UpdateDepthBiasEnable(regs);
}
// EDS2 Extras: LogicOp operation selection
if (device.IsExtExtendedDynamicState2ExtrasSupported()) {
UpdateLogicOp(regs);
}
// EDS3 Enables: LogicOpEnable, DepthClamp, LineStipple, ConservativeRaster
if (device.IsExtExtendedDynamicState3EnablesSupported()) {
using namespace Tegra::Engines;
// AMD Workaround: LogicOp incompatible with float render targets
if (device.GetDriverID() == VkDriverIdKHR::VK_DRIVER_ID_AMD_OPEN_SOURCE ||
device.GetDriverID() == VkDriverIdKHR::VK_DRIVER_ID_AMD_PROPRIETARY) {
const auto has_float = std::any_of(
regs.vertex_attrib_format.begin(), regs.vertex_attrib_format.end(),
[](const auto& attrib) {
return attrib.type == Maxwell3D::Regs::VertexAttribute::Type::Float;
if (device.IsExtExtendedDynamicState2Supported()) {
UpdatePrimitiveRestartEnable(regs);
UpdateRasterizerDiscardEnable(regs);
UpdateDepthBiasEnable(regs);
}
if (device.IsExtExtendedDynamicState3EnablesSupported()) {
using namespace Tegra::Engines;
if (device.GetDriverID() == VkDriverIdKHR::VK_DRIVER_ID_AMD_OPEN_SOURCE || device.GetDriverID() == VkDriverIdKHR::VK_DRIVER_ID_AMD_PROPRIETARY) {
const auto has_float = std::any_of(
regs.vertex_attrib_format.begin(),
regs.vertex_attrib_format.end(),
[](const auto& attrib) {
return attrib.type == Maxwell3D::Regs::VertexAttribute::Type::Float;
}
);
if (regs.logic_op.enable) {
regs.logic_op.enable = static_cast<u32>(!has_float);
}
}
);
if (regs.logic_op.enable) {
regs.logic_op.enable = static_cast<u32>(!has_float);
UpdateLogicOpEnable(regs);
UpdateDepthClampEnable(regs);
}
}
UpdateLogicOpEnable(regs);
UpdateDepthClampEnable(regs);
UpdateLineRasterizationMode(regs);
UpdateLineStippleEnable(regs);
UpdateConservativeRasterizationMode(regs);
UpdateAlphaToCoverageEnable(regs);
UpdateAlphaToOneEnable(regs);
if (device.IsExtExtendedDynamicState2ExtrasSupported()) {
UpdateLogicOp(regs);
}
if (device.IsExtExtendedDynamicState3BlendingSupported()) {
UpdateBlending(regs);
}
if (device.IsExtExtendedDynamicState3EnablesSupported()) {
UpdateLineStippleEnable(regs);
UpdateConservativeRasterizationMode(regs);
}
}
// EDS3 Blending: ColorBlendEnable, ColorBlendEquation, ColorWriteMask
if (device.IsExtExtendedDynamicState3BlendingSupported()) {
UpdateBlending(regs);
}
// Vertex Input Dynamic State: Independent from EDS levels
if (device.IsExtVertexInputDynamicStateSupported()) {
if (auto* gp = pipeline_cache.CurrentGraphicsPipeline(); gp && gp->HasDynamicVertexInput()) {
UpdateVertexInput(regs);
@@ -1066,16 +1002,9 @@ void RasterizerVulkan::HandleTransformFeedback() {
const auto& regs = maxwell3d->regs;
if (!device.IsExtTransformFeedbackSupported()) {
// If the guest enabled transform feedback, warn once that the device lacks support.
if (regs.transform_feedback_enabled != 0) {
std::call_once(warn_unsupported, [&] {
LOG_WARNING(Render_Vulkan, "Transform feedback requested by guest but VK_EXT_transform_feedback is unavailable; queries disabled");
});
} else {
std::call_once(warn_unsupported, [&] {
LOG_INFO(Render_Vulkan, "VK_EXT_transform_feedback not available on device");
});
}
std::call_once(warn_unsupported, [&] {
LOG_ERROR(Render_Vulkan, "Transform feedbacks used but not supported");
});
return;
}
query_cache.CounterEnable(VideoCommon::QueryType::StreamingByteCount,
@@ -1084,7 +1013,7 @@ void RasterizerVulkan::HandleTransformFeedback() {
UNIMPLEMENTED_IF(regs.IsShaderConfigEnabled(Maxwell::ShaderType::TessellationInit) ||
regs.IsShaderConfigEnabled(Maxwell::ShaderType::Tessellation));
}
}
}
void RasterizerVulkan::UpdateViewportsState(Tegra::Engines::Maxwell3D::Regs& regs) {
if (!state_tracker.TouchViewports()) {
@@ -1233,9 +1162,6 @@ void RasterizerVulkan::UpdateBlendConstants(Tegra::Engines::Maxwell3D::Regs& reg
if (!state_tracker.TouchBlendConstants()) {
return;
}
if (!device.UsesAdvancedCoreDynamicState()) {
return;
}
const std::array blend_color = {regs.blend_color.r, regs.blend_color.g, regs.blend_color.b,
regs.blend_color.a};
scheduler.Record(
@@ -1246,9 +1172,6 @@ void RasterizerVulkan::UpdateDepthBounds(Tegra::Engines::Maxwell3D::Regs& regs)
if (!state_tracker.TouchDepthBounds()) {
return;
}
if (!device.UsesAdvancedCoreDynamicState() || !device.IsDepthBoundsSupported()) {
return;
}
scheduler.Record([min = regs.depth_bounds[0], max = regs.depth_bounds[1]](
vk::CommandBuffer cmdbuf) { cmdbuf.SetDepthBounds(min, max); });
}
@@ -1257,10 +1180,6 @@ void RasterizerVulkan::UpdateStencilFaces(Tegra::Engines::Maxwell3D::Regs& regs)
if (!state_tracker.TouchStencilProperties()) {
return;
}
if (!device.UsesAdvancedCoreDynamicState()) {
state_tracker.ClearStencilReset();
return;
}
bool update_references = state_tracker.TouchStencilReference();
bool update_write_mask = state_tracker.TouchStencilWriteMask();
bool update_compare_masks = state_tracker.TouchStencilCompare();
@@ -1423,10 +1342,6 @@ void RasterizerVulkan::UpdateConservativeRasterizationMode(Tegra::Engines::Maxwe
return;
}
if (!device.SupportsDynamicState3ConservativeRasterizationMode()) {
return;
}
scheduler.Record([enable = regs.conservative_raster_enable](vk::CommandBuffer cmdbuf) {
cmdbuf.SetConservativeRasterizationModeEXT(
enable ? VK_CONSERVATIVE_RASTERIZATION_MODE_OVERESTIMATE_EXT
@@ -1439,50 +1354,23 @@ void RasterizerVulkan::UpdateLineStippleEnable(Tegra::Engines::Maxwell3D::Regs&
return;
}
if (!device.SupportsDynamicState3LineStippleEnable()) {
return;
}
scheduler.Record([enable = regs.line_stipple_enable](vk::CommandBuffer cmdbuf) {
cmdbuf.SetLineStippleEnableEXT(enable);
});
}
void RasterizerVulkan::UpdateLineRasterizationMode(Tegra::Engines::Maxwell3D::Regs& regs) {
if (!device.IsExtLineRasterizationSupported()) {
return;
}
if (!state_tracker.TouchLineRasterizationMode()) {
return;
}
// if (!state_tracker.TouchLi()) {
// return;
// }
if (!device.SupportsDynamicState3LineRasterizationMode()) {
static std::once_flag warn_missing_rect;
std::call_once(warn_missing_rect, [] {
LOG_WARNING(Render_Vulkan,
"Driver lacks rectangular line rasterization support; skipping dynamic "
"line state updates");
});
return;
}
// TODO: The maxwell emulator does not capture line rasters
const bool wants_smooth = regs.line_anti_alias_enable != 0;
VkLineRasterizationModeEXT mode = VK_LINE_RASTERIZATION_MODE_RECTANGULAR_EXT;
if (wants_smooth) {
if (device.SupportsSmoothLines()) {
mode = VK_LINE_RASTERIZATION_MODE_RECTANGULAR_SMOOTH_EXT;
} else {
static std::once_flag warn_missing_smooth;
std::call_once(warn_missing_smooth, [] {
LOG_WARNING(Render_Vulkan,
"Line anti-aliasing requested but smoothLines feature unavailable; "
"using rectangular rasterization");
});
}
}
scheduler.Record([mode](vk::CommandBuffer cmdbuf) {
cmdbuf.SetLineRasterizationModeEXT(mode);
});
// scheduler.Record([enable = regs.line](vk::CommandBuffer cmdbuf) {
// cmdbuf.SetConservativeRasterizationModeEXT(
// enable ? VK_CONSERVATIVE_RASTERIZATION_MODE_UNDERESTIMATE_EXT
// : VK_CONSERVATIVE_RASTERIZATION_MODE_DISABLED_EXT);
// });
}
void RasterizerVulkan::UpdateDepthBiasEnable(Tegra::Engines::Maxwell3D::Regs& regs) {
@@ -1524,9 +1412,6 @@ void RasterizerVulkan::UpdateLogicOpEnable(Tegra::Engines::Maxwell3D::Regs& regs
if (!state_tracker.TouchLogicOpEnable()) {
return;
}
if (!device.SupportsDynamicState3LogicOpEnable()) {
return;
}
scheduler.Record([enable = regs.logic_op.enable](vk::CommandBuffer cmdbuf) {
cmdbuf.SetLogicOpEnableEXT(enable != 0);
});
@@ -1536,9 +1421,6 @@ void RasterizerVulkan::UpdateDepthClampEnable(Tegra::Engines::Maxwell3D::Regs& r
if (!state_tracker.TouchDepthClampEnable()) {
return;
}
if (!device.SupportsDynamicState3DepthClampEnable()) {
return;
}
bool is_enabled = !(regs.viewport_clip_control.geometry_clip ==
Maxwell::ViewportClipControl::GeometryClip::Passthrough ||
regs.viewport_clip_control.geometry_clip ==
@@ -1549,41 +1431,6 @@ void RasterizerVulkan::UpdateDepthClampEnable(Tegra::Engines::Maxwell3D::Regs& r
[is_enabled](vk::CommandBuffer cmdbuf) { cmdbuf.SetDepthClampEnableEXT(is_enabled); });
}
void RasterizerVulkan::UpdateAlphaToCoverageEnable(Tegra::Engines::Maxwell3D::Regs& regs) {
if (!state_tracker.TouchAlphaToCoverageEnable()) {
return;
}
if (!device.SupportsDynamicState3AlphaToCoverageEnable()) {
return;
}
GraphicsPipeline* const pipeline = pipeline_cache.CurrentGraphicsPipeline();
const bool enable = pipeline != nullptr && pipeline->SupportsAlphaToCoverage() &&
regs.anti_alias_alpha_control.alpha_to_coverage != 0;
scheduler.Record([enable](vk::CommandBuffer cmdbuf) {
cmdbuf.SetAlphaToCoverageEnableEXT(enable ? VK_TRUE : VK_FALSE);
});
}
void RasterizerVulkan::UpdateAlphaToOneEnable(Tegra::Engines::Maxwell3D::Regs& regs) {
if (!state_tracker.TouchAlphaToOneEnable()) {
return;
}
if (!device.SupportsDynamicState3AlphaToOneEnable()) {
static std::once_flag warn_alpha_to_one;
std::call_once(warn_alpha_to_one, [] {
LOG_WARNING(Render_Vulkan,
"Alpha-to-one is not supported on this device; forcing it disabled");
});
return;
}
GraphicsPipeline* const pipeline = pipeline_cache.CurrentGraphicsPipeline();
const bool enable = pipeline != nullptr && pipeline->SupportsAlphaToOne() &&
regs.anti_alias_alpha_control.alpha_to_one != 0;
scheduler.Record([enable](vk::CommandBuffer cmdbuf) {
cmdbuf.SetAlphaToOneEnableEXT(enable ? VK_TRUE : VK_FALSE);
});
}
void RasterizerVulkan::UpdateDepthCompareOp(Tegra::Engines::Maxwell3D::Regs& regs) {
if (!state_tracker.TouchDepthCompareOp()) {
return;

View File

@@ -183,8 +183,6 @@ private:
void UpdateDepthBiasEnable(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateLogicOpEnable(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateDepthClampEnable(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateAlphaToCoverageEnable(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateAlphaToOneEnable(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateFrontFace(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateStencilOp(Tegra::Engines::Maxwell3D::Regs& regs);
void UpdateStencilTestEnable(Tegra::Engines::Maxwell3D::Regs& regs);

View File

@@ -13,7 +13,6 @@
#include "common/thread.h"
#include "video_core/renderer_vulkan/vk_command_pool.h"
#include "video_core/renderer_vulkan/vk_graphics_pipeline.h"
#include "video_core/renderer_vulkan/vk_master_semaphore.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_state_tracker.h"
@@ -131,27 +130,9 @@ void Scheduler::RequestOutsideRenderPassOperationContext() {
bool Scheduler::UpdateGraphicsPipeline(GraphicsPipeline* pipeline) {
if (state.graphics_pipeline == pipeline) {
if (pipeline && pipeline->UsesExtendedDynamicState() &&
state.needs_state_enable_refresh) {
state_tracker.InvalidateStateEnableFlag();
state.needs_state_enable_refresh = false;
}
return false;
}
state.graphics_pipeline = pipeline;
if (!pipeline) {
return true;
}
if (!pipeline->UsesExtendedDynamicState()) {
state.needs_state_enable_refresh = true;
} else if (state.needs_state_enable_refresh) {
state_tracker.InvalidateStateEnableFlag();
state.needs_state_enable_refresh = false;
}
return true;
}
@@ -296,7 +277,6 @@ void Scheduler::EndRenderPass()
}
query_cache->CounterEnable(VideoCommon::QueryType::ZPassPixelCount64, false);
query_cache->CounterEnable(VideoCommon::QueryType::StreamingByteCount, false);
query_cache->NotifySegment(false);
Record([num_images = num_renderpass_images,

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -217,7 +214,6 @@ private:
GraphicsPipeline* graphics_pipeline = nullptr;
bool is_rescaling = false;
bool rescaling_defined = false;
bool needs_state_enable_refresh = false;
};
void WorkerThread(std::stop_token stop_token);

View File

@@ -48,7 +48,6 @@ Flags MakeInvalidationFlags() {
FrontFace,
StencilOp,
StencilTestEnable,
RasterizerDiscardEnable,
VertexBuffers,
VertexInput,
StateEnable,
@@ -56,9 +55,6 @@ Flags MakeInvalidationFlags() {
DepthBiasEnable,
LogicOpEnable,
DepthClampEnable,
AlphaToCoverageEnable,
AlphaToOneEnable,
LineRasterizationMode,
LogicOp,
Blending,
ColorMask,
@@ -152,8 +148,6 @@ void SetupDirtyStateEnable(Tables& tables) {
setup(OFF(logic_op.enable), LogicOpEnable);
setup(OFF(viewport_clip_control.geometry_clip), DepthClampEnable);
setup(OFF(line_stipple_enable), LineStippleEnable);
setup(OFF(anti_alias_alpha_control.alpha_to_coverage), AlphaToCoverageEnable);
setup(OFF(anti_alias_alpha_control.alpha_to_one), AlphaToOneEnable);
}
void SetupDirtyDepthCompareOp(Tables& tables) {
@@ -232,7 +226,6 @@ void SetupRasterModes(Tables &tables) {
table[OFF(line_stipple_params)] = LineStippleParams;
table[OFF(conservative_raster_enable)] = ConservativeRasterizationMode;
table[OFF(line_anti_alias_enable)] = LineRasterizationMode;
}
} // Anonymous namespace

View File

@@ -54,7 +54,6 @@ enum : u8 {
PrimitiveRestartEnable,
RasterizerDiscardEnable,
ConservativeRasterizationMode,
LineRasterizationMode,
LineStippleEnable,
LineStippleParams,
DepthBiasEnable,
@@ -62,8 +61,6 @@ enum : u8 {
LogicOp,
LogicOpEnable,
DepthClampEnable,
AlphaToCoverageEnable,
AlphaToOneEnable,
Blending,
BlendEnable,
@@ -97,10 +94,6 @@ public:
(*flags)[Dirty::Scissors] = true;
}
void InvalidateStateEnableFlag() {
(*flags)[Dirty::StateEnable] = true;
}
bool TouchViewports() {
const bool dirty_viewports = Exchange(Dirty::Viewports, false);
const bool rescale_viewports = Exchange(VideoCommon::Dirty::RescaleViewports, false);
@@ -232,14 +225,6 @@ public:
return Exchange(Dirty::DepthClampEnable, false);
}
bool TouchAlphaToCoverageEnable() {
return Exchange(Dirty::AlphaToCoverageEnable, false);
}
bool TouchAlphaToOneEnable() {
return Exchange(Dirty::AlphaToOneEnable, false);
}
bool TouchDepthCompareOp() {
return Exchange(Dirty::DepthCompareOp, false);
}
@@ -276,10 +261,6 @@ public:
return Exchange(Dirty::LogicOp, false);
}
bool TouchLineRasterizationMode() {
return Exchange(Dirty::LineRasterizationMode, false);
}
bool ChangePrimitiveTopology(Maxwell::PrimitiveTopology new_topology) {
const bool has_changed = current_topology != new_topology;
current_topology = new_topology;

View File

@@ -306,17 +306,7 @@ void Swapchain::CreateSwapchain(const VkSurfaceCapabilitiesKHR& capabilities) {
swapchain_ci.queueFamilyIndexCount = static_cast<u32>(queue_indices.size());
swapchain_ci.pQueueFamilyIndices = queue_indices.data();
}
// According to Vulkan spec, when using VK_SWAPCHAIN_CREATE_MUTABLE_FORMAT_BIT_KHR,
// the base format (imageFormat) MUST be included in pViewFormats
const std::array view_formats{
swapchain_ci.imageFormat, // Base format MUST be first
VK_FORMAT_B8G8R8A8_UNORM,
VK_FORMAT_B8G8R8A8_SRGB,
#ifdef ANDROID
VK_FORMAT_R8G8B8A8_UNORM, // Android may use RGBA
VK_FORMAT_R8G8B8A8_SRGB,
#endif
};
static constexpr std::array view_formats{VK_FORMAT_B8G8R8A8_UNORM, VK_FORMAT_B8G8R8A8_SRGB};
VkImageFormatListCreateInfo format_list{
.sType = VK_STRUCTURE_TYPE_IMAGE_FORMAT_LIST_CREATE_INFO_KHR,
.pNext = nullptr,

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -104,7 +101,7 @@ public:
}
VkSemaphore CurrentRenderSemaphore() const {
return *render_semaphores[image_index];
return *render_semaphores[frame_index];
}
u32 GetWidth() const {

View File

@@ -1104,8 +1104,6 @@ void TextureCacheRuntime::BlitImage(Framebuffer* dst_framebuffer, ImageView& dst
UNREACHABLE();
}
}();
// Use shader-based depth/stencil blits if hardware doesn't support the format
// Note: MSAA resolves (MSAA->single) use vkCmdResolveImage which works fine
if (!can_blit_depth_stencil) {
UNIMPLEMENTED_IF(is_src_msaa || is_dst_msaa);
blit_image_helper.BlitDepthStencil(dst_framebuffer, src, dst_region, src_region,
@@ -1120,15 +1118,6 @@ void TextureCacheRuntime::BlitImage(Framebuffer* dst_framebuffer, ImageView& dst
const VkImage src_image = src.ImageHandle();
const VkImageSubresourceLayers dst_layers = MakeSubresourceLayers(&dst);
const VkImageSubresourceLayers src_layers = MakeSubresourceLayers(&src);
const bool is_msaa_to_msaa = is_src_msaa && is_dst_msaa;
// NVIDIA 510+ and Intel crash on MSAA->MSAA blits (scaling operations)
// Fall back to 3D helpers for MSAA scaling
if (is_msaa_to_msaa && device.CantBlitMSAA()) {
// This should be handled by NeedsScaleHelper() and use 3D helpers instead
UNIMPLEMENTED_MSG("MSAA to MSAA blit not supported on this driver");
return;
}
const bool is_resolve = is_src_msaa && !is_dst_msaa;
scheduler.RequestOutsideRenderPassOperationContext();
scheduler.Record([filter, dst_region, src_region, dst_image, src_image, dst_layers, src_layers,
@@ -2233,26 +2222,18 @@ vk::ImageView ImageView::MakeView(VkFormat vk_format, VkImageAspectFlags aspect_
Sampler::Sampler(TextureCacheRuntime& runtime, const Tegra::Texture::TSCEntry& tsc) {
const auto& device = runtime.device;
// Check if custom border colors are supported
const bool has_custom_border_colors = runtime.device.IsCustomBorderColorsSupported();
const bool has_format_undefined = runtime.device.IsCustomBorderColorWithoutFormatSupported();
const bool arbitrary_borders = runtime.device.IsExtCustomBorderColorSupported();
const auto color = tsc.BorderColor();
// Determine border format based on available features:
// - If customBorderColorWithoutFormat is available: use VK_FORMAT_UNDEFINED (most flexible)
// - If only customBorderColors is available: use concrete format (R8G8B8A8_UNORM)
// - If neither is available: use standard border colors (handled by ConvertBorderColor)
const VkFormat border_format = has_format_undefined ? VK_FORMAT_UNDEFINED
: VK_FORMAT_R8G8B8A8_UNORM;
const VkSamplerCustomBorderColorCreateInfoEXT border_ci{
.sType = VK_STRUCTURE_TYPE_SAMPLER_CUSTOM_BORDER_COLOR_CREATE_INFO_EXT,
.pNext = nullptr,
// TODO: Make use of std::bit_cast once libc++ supports it.
.customBorderColor = std::bit_cast<VkClearColorValue>(color),
.format = border_format,
.format = VK_FORMAT_UNDEFINED,
};
const void* pnext = nullptr;
if (has_custom_border_colors) {
if (arbitrary_borders) {
pnext = &border_ci;
}
const VkSamplerReductionModeCreateInfoEXT reduction_ci{
@@ -2286,8 +2267,8 @@ Sampler::Sampler(TextureCacheRuntime& runtime, const Tegra::Texture::TSCEntry& t
.compareOp = MaxwellToVK::Sampler::DepthCompareFunction(tsc.depth_compare_func),
.minLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.0f : tsc.MinLod(),
.maxLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.25f : tsc.MaxLod(),
.borderColor = has_custom_border_colors ? VK_BORDER_COLOR_FLOAT_CUSTOM_EXT
: ConvertBorderColor(color),
.borderColor =
arbitrary_borders ? VK_BORDER_COLOR_FLOAT_CUSTOM_EXT : ConvertBorderColor(color),
.unnormalizedCoordinates = VK_FALSE,
});
};

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
@@ -244,6 +241,7 @@ public:
[[nodiscard]] VkImageView Handle(Shader::TextureType texture_type) const noexcept {
return *image_views[static_cast<size_t>(texture_type)];
}
[[nodiscard]] VkImage ImageHandle() const noexcept {
return image_handle;
}

View File

@@ -33,7 +33,6 @@ ImageInfo::ImageInfo(const TICEntry& config) noexcept {
dma_downloaded = forced_flushed;
format = PixelFormatFromTextureInfo(config.format, config.r_type, config.g_type, config.b_type,
config.a_type, config.srgb_conversion);
num_samples = NumSamples(config.msaa_mode);
resources.levels = config.max_mip_level + 1;
if (config.IsPitchLinear()) {

View File

@@ -22,17 +22,6 @@
#include <vulkan/vulkan.h>
// Define maintenance 7-9 extension names (not yet in official Vulkan headers)
#ifndef VK_KHR_MAINTENANCE_7_EXTENSION_NAME
#define VK_KHR_MAINTENANCE_7_EXTENSION_NAME "VK_KHR_maintenance7"
#endif
#ifndef VK_KHR_MAINTENANCE_8_EXTENSION_NAME
#define VK_KHR_MAINTENANCE_8_EXTENSION_NAME "VK_KHR_maintenance8"
#endif
#ifndef VK_KHR_MAINTENANCE_9_EXTENSION_NAME
#define VK_KHR_MAINTENANCE_9_EXTENSION_NAME "VK_KHR_maintenance9"
#endif
// Sanitize macros
#undef CreateEvent
#undef CreateSemaphore

View File

@@ -292,10 +292,9 @@ std::unordered_map<VkFormat, VkFormatProperties> GetFormatProperties(vk::Physica
#if defined(ANDROID) && defined(ARCHITECTURE_arm64)
void OverrideBcnFormats(std::unordered_map<VkFormat, VkFormatProperties>& format_properties) {
// These properties are extracted from Adreno driver 512.687.0
constexpr VkFormatFeatureFlags tiling_features{VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT |
VK_FORMAT_FEATURE_BLIT_SRC_BIT |
VK_FORMAT_FEATURE_SAMPLED_IMAGE_FILTER_LINEAR_BIT |
VK_FORMAT_FEATURE_TRANSFER_SRC_BIT |
constexpr VkFormatFeatureFlags tiling_features{
VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT | VK_FORMAT_FEATURE_BLIT_SRC_BIT |
VK_FORMAT_FEATURE_SAMPLED_IMAGE_FILTER_LINEAR_BIT | VK_FORMAT_FEATURE_TRANSFER_SRC_BIT |
VK_FORMAT_FEATURE_TRANSFER_DST_BIT};
constexpr VkFormatFeatureFlags buffer_features{VK_FORMAT_FEATURE_UNIFORM_TEXEL_BUFFER_BIT};
@@ -417,6 +416,7 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
const bool is_suitable = GetSuitability(surface != nullptr);
const VkDriverId driver_id = properties.driver.driverID;
const auto device_id = properties.properties.deviceID;
const bool is_radv = driver_id == VK_DRIVER_ID_MESA_RADV;
const bool is_amd_driver =
driver_id == VK_DRIVER_ID_AMD_PROPRIETARY || driver_id == VK_DRIVER_ID_AMD_OPEN_SOURCE;
@@ -427,13 +427,11 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
const bool is_mvk = driver_id == VK_DRIVER_ID_MOLTENVK;
const bool is_qualcomm = driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY;
const bool is_turnip = driver_id == VK_DRIVER_ID_MESA_TURNIP;
const bool is_arm = driver_id == VK_DRIVER_ID_ARM_PROPRIETARY;
const bool is_s8gen2 = device_id == 0x43050a01;
//const bool is_arm = driver_id == VK_DRIVER_ID_ARM_PROPRIETARY;
if ((is_mvk || is_qualcomm || is_turnip || is_arm) && !is_suitable) {
LOG_WARNING(Render_Vulkan, "Unsuitable driver, continuing anyway");
} else if (!is_suitable) {
throw vk::Exception(VK_ERROR_INCOMPATIBLE_DRIVER);
}
if (!is_suitable)
LOG_WARNING(Render_Vulkan, "Unsuitable driver - continuing anyways");
if (is_nvidia) {
nvidia_arch = GetNvidiaArchitecture(physical, supported_extensions);
@@ -494,11 +492,6 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
CollectToolingInfo();
if (is_qualcomm) {
// Qualcomm Adreno GPUs doesn't handle scaled vertex attributes; keep emulation enabled
must_emulate_scaled_formats = true;
LOG_WARNING(Render_Vulkan,
"Qualcomm drivers require scaled vertex format emulation; forcing fallback");
LOG_WARNING(Render_Vulkan,
"Disabling shader float controls and 64-bit integer features on Qualcomm proprietary drivers");
RemoveExtension(extensions.shader_float_controls, VK_KHR_SHADER_FLOAT_CONTROLS_EXTENSION_NAME);
@@ -540,31 +533,35 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
}
if (nv_major_version >= 510) {
LOG_WARNING(Render_Vulkan,
"NVIDIA Drivers >= 510 do not support MSAA->MSAA image blits. "
"MSAA scaling will use 3D helpers. MSAA resolves work normally.");
LOG_WARNING(Render_Vulkan, "NVIDIA Drivers >= 510 do not support MSAA image blits");
cant_blit_msaa = true;
}
}
// Mali/ NVIDIA proprietary drivers: Shader stencil export not supported
// Use hardware depth/stencil blits instead when available
if (!extensions.shader_stencil_export) {
LOG_INFO(Render_Vulkan,
"NVIDIA: VK_EXT_shader_stencil_export not supported, using hardware blits "
"for depth/stencil operations");
LOG_INFO(Render_Vulkan, " D24S8 hardware blit support: {}",
is_blit_depth24_stencil8_supported);
LOG_INFO(Render_Vulkan, " D32S8 hardware blit support: {}",
is_blit_depth32_stencil8_supported);
if (!is_blit_depth24_stencil8_supported && !is_blit_depth32_stencil8_supported) {
LOG_WARNING(Render_Vulkan,
"NVIDIA: Neither shader export nor hardware blits available for "
"depth/stencil. Performance may be degraded.");
}
if (extensions.extended_dynamic_state3 && is_radv) {
LOG_WARNING(Render_Vulkan, "RADV has broken extendedDynamicState3ColorBlendEquation");
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEnable = false;
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEquation = false;
dynamic_state3_blending = false;
const u32 version = (properties.properties.driverVersion << 3) >> 3;
if (version < VK_MAKE_API_VERSION(0, 23, 1, 0)) {
LOG_WARNING(Render_Vulkan,
"RADV versions older than 23.1.0 have broken depth clamp dynamic state");
features.extended_dynamic_state3.extendedDynamicState3DepthClampEnable = false;
dynamic_state3_enables = false;
}
}
if (extensions.extended_dynamic_state3 && (is_amd_driver || driver_id == VK_DRIVER_ID_SAMSUNG_PROPRIETARY)) {
// AMD and Samsung drivers have broken extendedDynamicState3ColorBlendEquation
LOG_WARNING(Render_Vulkan,
"AMD and Samsung drivers have broken extendedDynamicState3ColorBlendEquation");
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEnable = false;
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEquation = false;
dynamic_state3_blending = false;
}
sets_per_pool = 64;
if (is_amd_driver) {
// AMD drivers need a higher amount of Sets per Pool in certain circumstances like in XC2.
@@ -601,27 +598,15 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
}
if (is_intel_windows) {
LOG_WARNING(Render_Vulkan,
"Intel proprietary drivers do not support MSAA->MSAA image blits. "
"MSAA scaling will use 3D helpers. MSAA resolves work normally.");
LOG_WARNING(Render_Vulkan, "Intel proprietary drivers do not support MSAA image blits");
cant_blit_msaa = true;
}
has_broken_compute =
CheckBrokenCompute(properties.driver.driverID, properties.properties.driverVersion) &&
!Settings::values.enable_compute_pipelines.GetValue();
must_emulate_bgr565 = false; // Default: assume emulation isn't required
if (is_intel_anv) {
LOG_WARNING(Render_Vulkan, "Intel ANV driver does not support native BGR format");
must_emulate_bgr565 = true;
} else if (is_qualcomm) {
LOG_WARNING(Render_Vulkan,
"Qualcomm driver mishandles BGR5 formats even with VK_KHR_maintenance5, forcing emulation");
must_emulate_bgr565 = true;
} else if (is_arm) {
LOG_WARNING(Render_Vulkan,
"ARM Mali driver mishandles BGR5 formats even with VK_KHR_maintenance5, forcing emulation");
if (is_intel_anv || (is_qualcomm && !is_s8gen2)) {
LOG_WARNING(Render_Vulkan, "Driver does not support native BGR format");
must_emulate_bgr565 = true;
}
@@ -634,55 +619,51 @@ Device::Device(VkInstance instance_, vk::PhysicalDevice physical_, VkSurfaceKHR
(std::min)(properties.properties.limits.maxVertexInputBindings, 16U);
}
if (is_turnip || is_qualcomm) {
LOG_WARNING(Render_Vulkan, "Driver requires higher-than-reported binding limits");
if (is_turnip) {
LOG_WARNING(Render_Vulkan, "Turnip requires higher-than-reported binding limits");
properties.properties.limits.maxVertexInputBindings = 32;
}
// Base dynamic states (VIEWPORT, SCISSOR, DEPTH_BIAS, etc.) are ALWAYS active in vk_graphics_pipeline.cpp
// This slider controls EXTENDED dynamic states with accumulative levels per Vulkan specs:
// Level 0 = Core Dynamic States only (Vulkan 1.0)
// Level 1 = Core + VK_EXT_extended_dynamic_state
// Level 2 = Core + VK_EXT_extended_dynamic_state + VK_EXT_extended_dynamic_state2
// Level 3 = Core + VK_EXT_extended_dynamic_state + VK_EXT_extended_dynamic_state2 + VK_EXT_extended_dynamic_state3
if (!extensions.extended_dynamic_state && extensions.extended_dynamic_state2) {
LOG_INFO(Render_Vulkan,
"Removing extendedDynamicState2 due to missing extendedDynamicState");
RemoveExtensionFeature(extensions.extended_dynamic_state2, features.extended_dynamic_state2,
VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME);
}
if (!extensions.extended_dynamic_state2 && extensions.extended_dynamic_state3) {
LOG_INFO(Render_Vulkan,
"Removing extendedDynamicState3 due to missing extendedDynamicState2");
RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3,
VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME);
dynamic_state3_blending = false;
dynamic_state3_enables = false;
}
// Mesa Intel drivers on UHD 620 have broken EDS causing extreme flickering - unknown if it affects other iGPUs
// ALSO affects ALL versions of UHD drivers on Windows 10+, seems to cause even worse issues like straight up crashing
// So... Yeah, UHD drivers fucking suck -- maybe one day we can work past this, maybe; some driver hacking?
// And then we can rest in peace by doing `< VK_MAKE_API_VERSION(26, 0, 0)` for our beloved mesa drivers... one day
if ((is_mvk || (is_integrated && is_intel_anv) || (is_integrated && is_intel_windows)) && Settings::values.dyna_state.GetValue() != 0) {
LOG_WARNING(Render_Vulkan, "Driver has broken dynamic state, forcing to 0 to prevent graphical issues");
Settings::values.dyna_state.SetValue(0);
}
switch (Settings::values.dyna_state.GetValue()) {
case 0:
// Level 0: Disable all extended dynamic state extensions
RemoveExtensionFeature(extensions.extended_dynamic_state, features.extended_dynamic_state,
VK_EXT_EXTENDED_DYNAMIC_STATE_EXTENSION_NAME);
RemoveExtensionFeature(extensions.extended_dynamic_state2, features.extended_dynamic_state2,
VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME);
RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3,
VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME);
dynamic_state3_blending = false;
dynamic_state3_enables = false;
break;
RemoveExtensionFeature(extensions.extended_dynamic_state, features.extended_dynamic_state, VK_EXT_EXTENDED_DYNAMIC_STATE_EXTENSION_NAME);
[[fallthrough]];
case 1:
// Level 1: Enable EDS1, disable EDS2 and EDS3
RemoveExtensionFeature(extensions.extended_dynamic_state2, features.extended_dynamic_state2,
VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME);
RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3,
VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME);
dynamic_state3_blending = false;
dynamic_state3_enables = false;
break;
RemoveExtensionFeature(extensions.extended_dynamic_state2, features.extended_dynamic_state2, VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME);
[[fallthrough]];
case 2:
// Level 2: Enable EDS1 + EDS2, disable EDS3
RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3,
VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME);
RemoveExtensionFeature(extensions.extended_dynamic_state3, features.extended_dynamic_state3, VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME);
dynamic_state3_blending = false;
dynamic_state3_enables = false;
break;
case 3:
default:
// Level 3: Enable all (EDS1 + EDS2 + EDS3)
break;
}
// VK_EXT_vertex_input_dynamic_state is independent from EDS
// It can be enabled even without extended_dynamic_state
if (!Settings::values.vertex_input_dynamic_state.GetValue()) {
if (!Settings::values.vertex_input_dynamic_state.GetValue() || !extensions.extended_dynamic_state) {
RemoveExtensionFeature(extensions.vertex_input_dynamic_state, features.vertex_input_dynamic_state, VK_EXT_VERTEX_INPUT_DYNAMIC_STATE_EXTENSION_NAME);
}
@@ -770,8 +751,8 @@ void Device::SaveShader(std::span<const u32> spirv) const {
}
bool Device::ComputeIsOptimalAstcSupported() const {
// Verify hardware supports all ASTC formats with optimal tiling to avoid software conversion
static constexpr std::array<VkFormat, 28> astc_formats = {
// Disable for now to avoid converting ASTC twice.
static constexpr std::array astc_formats = {
VK_FORMAT_ASTC_4x4_UNORM_BLOCK, VK_FORMAT_ASTC_4x4_SRGB_BLOCK,
VK_FORMAT_ASTC_5x4_UNORM_BLOCK, VK_FORMAT_ASTC_5x4_SRGB_BLOCK,
VK_FORMAT_ASTC_5x5_UNORM_BLOCK, VK_FORMAT_ASTC_5x5_SRGB_BLOCK,
@@ -790,10 +771,9 @@ bool Device::ComputeIsOptimalAstcSupported() const {
if (!features.features.textureCompressionASTC_LDR) {
return false;
}
const auto format_feature_usage{VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT |
VK_FORMAT_FEATURE_BLIT_SRC_BIT |
VK_FORMAT_FEATURE_BLIT_DST_BIT |
VK_FORMAT_FEATURE_TRANSFER_SRC_BIT |
const auto format_feature_usage{
VK_FORMAT_FEATURE_SAMPLED_IMAGE_BIT | VK_FORMAT_FEATURE_BLIT_SRC_BIT |
VK_FORMAT_FEATURE_BLIT_DST_BIT | VK_FORMAT_FEATURE_TRANSFER_SRC_BIT |
VK_FORMAT_FEATURE_TRANSFER_DST_BIT};
for (const auto format : astc_formats) {
const auto physical_format_properties{physical.GetFormatProperties(format)};
@@ -990,7 +970,7 @@ bool Device::GetSuitability(bool requires_swapchain) {
// Set next pointer.
void** next = &features2.pNext;
// Vulkan 1.2 and 1.3 features
// Vulkan 1.2, 1.3 and 1.4 features
if (instance_version >= VK_API_VERSION_1_2) {
features_1_2.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VULKAN_1_2_FEATURES;
features_1_3.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_VULKAN_1_3_FEATURES;
@@ -1096,16 +1076,6 @@ bool Device::GetSuitability(bool requires_swapchain) {
VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_TRANSFORM_FEEDBACK_PROPERTIES_EXT;
SetNext(next, properties.transform_feedback);
}
if (extensions.maintenance5) {
properties.maintenance5.sType =
VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MAINTENANCE_5_PROPERTIES_KHR;
SetNext(next, properties.maintenance5);
}
if (extensions.multi_draw) {
properties.multi_draw.sType =
VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MULTI_DRAW_PROPERTIES_EXT;
SetNext(next, properties.multi_draw);
}
// Perform the property fetch.
physical.GetProperties2(properties2);
@@ -1138,75 +1108,14 @@ bool Device::GetSuitability(bool requires_swapchain) {
}
}
// VK_DYNAMIC_STATE
// Driver detection variables for workarounds in GetSuitability
const VkDriverId driver_id = properties.driver.driverID;
[[maybe_unused]] const bool is_amd_driver =
driver_id == VK_DRIVER_ID_AMD_PROPRIETARY || driver_id == VK_DRIVER_ID_AMD_OPEN_SOURCE;
const bool is_intel_windows = driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS;
[[maybe_unused]] const bool is_qualcomm = driver_id == VK_DRIVER_ID_QUALCOMM_PROPRIETARY;
// VK_EXT_extended_dynamic_state2 below this will appear drivers that need workarounds.
// VK_EXT_extended_dynamic_state3 below this will appear drivers that need workarounds.
[[maybe_unused]] const auto device_id = properties.properties.deviceID;
// Samsung: Broken extendedDynamicState3ColorBlendEquation
// Disable blend equation dynamic state, force static pipeline state
if (extensions.extended_dynamic_state3 &&
(driver_id == VK_DRIVER_ID_SAMSUNG_PROPRIETARY)) {
LOG_WARNING(Render_Vulkan,
"Samsung: Disabling broken extendedDynamicState3ColorBlendEquation");
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEnable = false;
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEquation = false;
}
// Intel Windows < 27.20.100.0: Broken VertexInputDynamicState
// Disable VertexInputDynamicState on old Intel Windows drivers
if (extensions.vertex_input_dynamic_state && is_intel_windows) {
const u32 version = (properties.properties.driverVersion << 3) >> 3;
if (version < VK_MAKE_API_VERSION(27, 20, 100, 0)) {
LOG_WARNING(Render_Vulkan,
"Intel Windows < 27.20.100.0: Disabling broken VK_EXT_vertex_input_dynamic_state");
RemoveExtensionFeature(extensions.vertex_input_dynamic_state,
features.vertex_input_dynamic_state,
VK_EXT_VERTEX_INPUT_DYNAMIC_STATE_EXTENSION_NAME);
}
}
if (Settings::values.dyna_state.GetValue() == 0) {
LOG_INFO(Render_Vulkan, "Extended Dynamic State disabled by user setting, clearing all EDS features");
features.custom_border_color.customBorderColors = false;
features.custom_border_color.customBorderColorWithoutFormat = false;
features.extended_dynamic_state.extendedDynamicState = false;
features.extended_dynamic_state2.extendedDynamicState2 = false;
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEnable = false;
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEquation = false;
features.extended_dynamic_state3.extendedDynamicState3ColorWriteMask = false;
features.extended_dynamic_state3.extendedDynamicState3DepthClampEnable = false;
features.extended_dynamic_state3.extendedDynamicState3LogicOpEnable = false;
}
// Return whether we were suitable.
return suitable;
}
void Device::RemoveUnsuitableExtensions() {
// VK_EXT_custom_border_color
// Enable extension if driver supports it, then check individual features
// - customBorderColors: Required to use VK_BORDER_COLOR_FLOAT_CUSTOM_EXT
// - customBorderColorWithoutFormat: Optional, allows VK_FORMAT_UNDEFINED
// If only customBorderColors is available, we must provide a specific format
if (extensions.custom_border_color) {
// Verify that at least customBorderColors is available
if (!features.custom_border_color.customBorderColors) {
LOG_WARNING(Render_Vulkan,
"VK_EXT_custom_border_color reported but customBorderColors feature not available, disabling");
extensions.custom_border_color = false;
}
}
extensions.custom_border_color = features.custom_border_color.customBorderColors &&
features.custom_border_color.customBorderColorWithoutFormat;
RemoveExtensionFeatureIfUnsuitable(extensions.custom_border_color, features.custom_border_color,
VK_EXT_CUSTOM_BORDER_COLOR_EXTENSION_NAME);
@@ -1235,79 +1144,21 @@ void Device::RemoveUnsuitableExtensions() {
VK_EXT_EXTENDED_DYNAMIC_STATE_2_EXTENSION_NAME);
// VK_EXT_extended_dynamic_state3
const bool supports_color_blend_enable =
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEnable;
const bool supports_color_blend_equation =
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEquation;
const bool supports_color_write_mask =
dynamic_state3_blending =
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEnable &&
features.extended_dynamic_state3.extendedDynamicState3ColorBlendEquation &&
features.extended_dynamic_state3.extendedDynamicState3ColorWriteMask;
dynamic_state3_blending = supports_color_blend_enable && supports_color_blend_equation &&
supports_color_write_mask;
const bool supports_depth_clamp_enable =
features.extended_dynamic_state3.extendedDynamicState3DepthClampEnable;
const bool supports_logic_op_enable =
dynamic_state3_enables =
features.extended_dynamic_state3.extendedDynamicState3DepthClampEnable &&
features.extended_dynamic_state3.extendedDynamicState3LogicOpEnable;
const bool supports_line_raster_mode =
features.extended_dynamic_state3.extendedDynamicState3LineRasterizationMode &&
extensions.line_rasterization && features.line_rasterization.rectangularLines;
const bool supports_conservative_raster_mode =
features.extended_dynamic_state3.extendedDynamicState3ConservativeRasterizationMode &&
extensions.conservative_rasterization;
const bool supports_line_stipple_enable =
features.extended_dynamic_state3.extendedDynamicState3LineStippleEnable &&
extensions.line_rasterization && features.line_rasterization.stippledRectangularLines;
const bool supports_alpha_to_coverage =
features.extended_dynamic_state3.extendedDynamicState3AlphaToCoverageEnable;
const bool supports_alpha_to_one =
features.extended_dynamic_state3.extendedDynamicState3AlphaToOneEnable &&
features.features.alphaToOne;
dynamic_state3_depth_clamp_enable = supports_depth_clamp_enable;
dynamic_state3_logic_op_enable = supports_logic_op_enable;
dynamic_state3_line_raster_mode = supports_line_raster_mode;
dynamic_state3_conservative_raster_mode = supports_conservative_raster_mode;
dynamic_state3_line_stipple_enable = supports_line_stipple_enable;
dynamic_state3_alpha_to_coverage = supports_alpha_to_coverage;
dynamic_state3_alpha_to_one = supports_alpha_to_one;
dynamic_state3_enables = dynamic_state3_depth_clamp_enable || dynamic_state3_logic_op_enable ||
dynamic_state3_line_raster_mode ||
dynamic_state3_conservative_raster_mode ||
dynamic_state3_line_stipple_enable ||
dynamic_state3_alpha_to_coverage || dynamic_state3_alpha_to_one;
extensions.extended_dynamic_state3 = dynamic_state3_blending || dynamic_state3_enables;
if (!extensions.extended_dynamic_state3) {
dynamic_state3_blending = false;
dynamic_state3_enables = false;
dynamic_state3_depth_clamp_enable = false;
dynamic_state3_logic_op_enable = false;
dynamic_state3_line_raster_mode = false;
dynamic_state3_conservative_raster_mode = false;
dynamic_state3_line_stipple_enable = false;
dynamic_state3_alpha_to_coverage = false;
dynamic_state3_alpha_to_one = false;
}
dynamic_state3_blending = dynamic_state3_blending && extensions.extended_dynamic_state3;
dynamic_state3_enables = dynamic_state3_enables && extensions.extended_dynamic_state3;
RemoveExtensionFeatureIfUnsuitable(extensions.extended_dynamic_state3,
features.extended_dynamic_state3,
VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME);
// VK_EXT_robustness2
// Enable if at least one robustness2 feature is available
extensions.robustness_2 = features.robustness2.robustBufferAccess2 ||
features.robustness2.robustImageAccess2 ||
features.robustness2.nullDescriptor;
RemoveExtensionFeatureIfUnsuitable(extensions.robustness_2, features.robustness2,
VK_EXT_ROBUSTNESS_2_EXTENSION_NAME);
// VK_EXT_image_robustness
// Enable if robustImageAccess is available
extensions.image_robustness = features.image_robustness.robustImageAccess;
RemoveExtensionFeatureIfUnsuitable(extensions.image_robustness, features.image_robustness,
VK_EXT_IMAGE_ROBUSTNESS_EXTENSION_NAME);
// VK_EXT_provoking_vertex
if (Settings::values.provoking_vertex.GetValue()) {
extensions.provoking_vertex = features.provoking_vertex.provokingVertexLast
@@ -1345,21 +1196,15 @@ void Device::RemoveUnsuitableExtensions() {
VK_EXT_SUBGROUP_SIZE_CONTROL_EXTENSION_NAME);
// VK_EXT_transform_feedback
// We only require the basic transformFeedback feature and at least
// one transform feedback buffer. We keep transformFeedbackQueries as it's used by
// the streaming byte count implementation. GeometryStreams and multiple streams
// are not strictly required since we currently support only stream 0.
extensions.transform_feedback =
features.transform_feedback.transformFeedback &&
features.transform_feedback.geometryStreams &&
properties.transform_feedback.maxTransformFeedbackStreams >= 4 &&
properties.transform_feedback.maxTransformFeedbackBuffers > 0 &&
properties.transform_feedback.transformFeedbackQueries;
properties.transform_feedback.transformFeedbackQueries &&
properties.transform_feedback.transformFeedbackDraw;
RemoveExtensionFeatureIfUnsuitable(extensions.transform_feedback, features.transform_feedback,
VK_EXT_TRANSFORM_FEEDBACK_EXTENSION_NAME);
if (extensions.transform_feedback) {
LOG_INFO(Render_Vulkan, "VK_EXT_transform_feedback enabled (buffers={}, queries={})",
properties.transform_feedback.maxTransformFeedbackBuffers,
properties.transform_feedback.transformFeedbackQueries);
}
// VK_EXT_vertex_input_dynamic_state
extensions.vertex_input_dynamic_state =
@@ -1368,17 +1213,6 @@ void Device::RemoveUnsuitableExtensions() {
features.vertex_input_dynamic_state,
VK_EXT_VERTEX_INPUT_DYNAMIC_STATE_EXTENSION_NAME);
// VK_EXT_multi_draw
extensions.multi_draw = features.multi_draw.multiDraw;
if (extensions.multi_draw) {
LOG_INFO(Render_Vulkan, "VK_EXT_multi_draw: maxMultiDrawCount={}",
properties.multi_draw.maxMultiDrawCount);
}
RemoveExtensionFeatureIfUnsuitable(extensions.multi_draw, features.multi_draw,
VK_EXT_MULTI_DRAW_EXTENSION_NAME);
// VK_KHR_pipeline_executable_properties
if (Settings::values.renderer_shader_feedback.GetValue()) {
extensions.pipeline_executable_properties =
@@ -1402,76 +1236,6 @@ void Device::RemoveUnsuitableExtensions() {
RemoveExtensionFeatureIfUnsuitable(extensions.workgroup_memory_explicit_layout,
features.workgroup_memory_explicit_layout,
VK_KHR_WORKGROUP_MEMORY_EXPLICIT_LAYOUT_EXTENSION_NAME);
// VK_EXT_swapchain_maintenance1 (extension only, has features)
// Requires VK_EXT_surface_maintenance1 instance extension
extensions.swapchain_maintenance1 = features.swapchain_maintenance1.swapchainMaintenance1;
if (extensions.swapchain_maintenance1) {
// Check if VK_EXT_surface_maintenance1 instance extension is available
const auto instance_extensions = vk::EnumerateInstanceExtensionProperties(dld);
const bool has_surface_maintenance1 = instance_extensions && std::ranges::any_of(*instance_extensions,
[](const VkExtensionProperties& prop) {
return std::strcmp(prop.extensionName, VK_EXT_SURFACE_MAINTENANCE_1_EXTENSION_NAME) == 0;
});
if (!has_surface_maintenance1) {
LOG_WARNING(Render_Vulkan,
"VK_EXT_swapchain_maintenance1 requires VK_EXT_surface_maintenance1, disabling");
extensions.swapchain_maintenance1 = false;
features.swapchain_maintenance1.swapchainMaintenance1 = false;
}
}
RemoveExtensionFeatureIfUnsuitable(extensions.swapchain_maintenance1, features.swapchain_maintenance1,
VK_EXT_SWAPCHAIN_MAINTENANCE_1_EXTENSION_NAME);
// VK_KHR_maintenance1 (core in Vulkan 1.1, no features)
extensions.maintenance1 = loaded_extensions.contains(VK_KHR_MAINTENANCE_1_EXTENSION_NAME);
RemoveExtensionIfUnsuitable(extensions.maintenance1, VK_KHR_MAINTENANCE_1_EXTENSION_NAME);
// VK_KHR_maintenance2 (core in Vulkan 1.1, no features)
extensions.maintenance2 = loaded_extensions.contains(VK_KHR_MAINTENANCE_2_EXTENSION_NAME);
RemoveExtensionIfUnsuitable(extensions.maintenance2, VK_KHR_MAINTENANCE_2_EXTENSION_NAME);
// VK_KHR_maintenance3 (core in Vulkan 1.1, no features)
extensions.maintenance3 = loaded_extensions.contains(VK_KHR_MAINTENANCE_3_EXTENSION_NAME);
RemoveExtensionIfUnsuitable(extensions.maintenance3, VK_KHR_MAINTENANCE_3_EXTENSION_NAME);
// VK_KHR_maintenance4
extensions.maintenance4 = features.maintenance4.maintenance4;
RemoveExtensionFeatureIfUnsuitable(extensions.maintenance4, features.maintenance4,
VK_KHR_MAINTENANCE_4_EXTENSION_NAME);
// VK_KHR_maintenance5
extensions.maintenance5 = features.maintenance5.maintenance5;
if (extensions.maintenance5) {
LOG_INFO(Render_Vulkan, "VK_KHR_maintenance5 properties: polygonModePointSize={} "
"depthStencilSwizzleOne={} earlyFragmentTests={} nonStrictWideLines={}",
properties.maintenance5.polygonModePointSize,
properties.maintenance5.depthStencilSwizzleOneSupport,
properties.maintenance5.earlyFragmentMultisampleCoverageAfterSampleCounting &&
properties.maintenance5.earlyFragmentSampleMaskTestBeforeSampleCounting,
properties.maintenance5.nonStrictWideLinesUseParallelogram);
}
RemoveExtensionFeatureIfUnsuitable(extensions.maintenance5, features.maintenance5,
VK_KHR_MAINTENANCE_5_EXTENSION_NAME);
// VK_KHR_maintenance6
extensions.maintenance6 = features.maintenance6.maintenance6;
RemoveExtensionFeatureIfUnsuitable(extensions.maintenance6, features.maintenance6,
VK_KHR_MAINTENANCE_6_EXTENSION_NAME);
// VK_KHR_maintenance7 (proposed for Vulkan 1.4, no features)
extensions.maintenance7 = loaded_extensions.contains(VK_KHR_MAINTENANCE_7_EXTENSION_NAME);
RemoveExtensionIfUnsuitable(extensions.maintenance7, VK_KHR_MAINTENANCE_7_EXTENSION_NAME);
// VK_KHR_maintenance8 (proposed for Vulkan 1.4, no features)
extensions.maintenance8 = loaded_extensions.contains(VK_KHR_MAINTENANCE_8_EXTENSION_NAME);
RemoveExtensionIfUnsuitable(extensions.maintenance8, VK_KHR_MAINTENANCE_8_EXTENSION_NAME);
// VK_KHR_maintenance9 (proposed for Vulkan 1.4, no features)
extensions.maintenance9 = loaded_extensions.contains(VK_KHR_MAINTENANCE_9_EXTENSION_NAME);
RemoveExtensionIfUnsuitable(extensions.maintenance9, VK_KHR_MAINTENANCE_9_EXTENSION_NAME);
}
void Device::SetupFamilies(VkSurfaceKHR surface) {

View File

@@ -37,13 +37,9 @@ VK_DEFINE_HANDLE(VmaAllocator)
FEATURE(KHR, TimelineSemaphore, TIMELINE_SEMAPHORE, timeline_semaphore)
#define FOR_EACH_VK_FEATURE_1_3(FEATURE) \
FEATURE(EXT, ImageRobustness, IMAGE_ROBUSTNESS, image_robustness) \
FEATURE(EXT, ShaderDemoteToHelperInvocation, SHADER_DEMOTE_TO_HELPER_INVOCATION, \
shader_demote_to_helper_invocation) \
FEATURE(EXT, SubgroupSizeControl, SUBGROUP_SIZE_CONTROL, subgroup_size_control) \
FEATURE(KHR, Maintenance4, MAINTENANCE_4, maintenance4)
#define FOR_EACH_VK_FEATURE_1_4(FEATURE)
FEATURE(EXT, SubgroupSizeControl, SUBGROUP_SIZE_CONTROL, subgroup_size_control)
// Define all features which may be used by the implementation and require an extension here.
#define FOR_EACH_VK_FEATURE_EXT(FEATURE) \
@@ -56,16 +52,12 @@ VK_DEFINE_HANDLE(VmaAllocator)
FEATURE(EXT, 4444Formats, 4444_FORMATS, format_a4b4g4r4) \
FEATURE(EXT, IndexTypeUint8, INDEX_TYPE_UINT8, index_type_uint8) \
FEATURE(EXT, LineRasterization, LINE_RASTERIZATION, line_rasterization) \
FEATURE(EXT, MultiDraw, MULTI_DRAW, multi_draw) \
FEATURE(EXT, PrimitiveTopologyListRestart, PRIMITIVE_TOPOLOGY_LIST_RESTART, \
primitive_topology_list_restart) \
FEATURE(EXT, ProvokingVertex, PROVOKING_VERTEX, provoking_vertex) \
FEATURE(EXT, Robustness2, ROBUSTNESS_2, robustness2) \
FEATURE(EXT, TransformFeedback, TRANSFORM_FEEDBACK, transform_feedback) \
FEATURE(EXT, VertexInputDynamicState, VERTEX_INPUT_DYNAMIC_STATE, vertex_input_dynamic_state) \
FEATURE(EXT, SwapchainMaintenance1, SWAPCHAIN_MAINTENANCE_1, swapchain_maintenance1) \
FEATURE(KHR, Maintenance5, MAINTENANCE_5, maintenance5) \
FEATURE(KHR, Maintenance6, MAINTENANCE_6, maintenance6) \
FEATURE(KHR, PipelineExecutableProperties, PIPELINE_EXECUTABLE_PROPERTIES, \
pipeline_executable_properties) \
FEATURE(KHR, WorkgroupMemoryExplicitLayout, WORKGROUP_MEMORY_EXPLICIT_LAYOUT, \
@@ -92,12 +84,6 @@ VK_DEFINE_HANDLE(VmaAllocator)
EXTENSION(KHR, SWAPCHAIN, swapchain) \
EXTENSION(KHR, SWAPCHAIN_MUTABLE_FORMAT, swapchain_mutable_format) \
EXTENSION(KHR, IMAGE_FORMAT_LIST, image_format_list) \
EXTENSION(KHR, MAINTENANCE_1, maintenance1) \
EXTENSION(KHR, MAINTENANCE_2, maintenance2) \
EXTENSION(KHR, MAINTENANCE_3, maintenance3) \
EXTENSION(KHR, MAINTENANCE_7, maintenance7) \
EXTENSION(KHR, MAINTENANCE_8, maintenance8) \
EXTENSION(KHR, MAINTENANCE_9, maintenance9) \
EXTENSION(NV, DEVICE_DIAGNOSTICS_CONFIG, device_diagnostics_config) \
EXTENSION(NV, GEOMETRY_SHADER_PASSTHROUGH, geometry_shader_passthrough) \
EXTENSION(NV, VIEWPORT_ARRAY2, viewport_array2) \
@@ -124,7 +110,6 @@ VK_DEFINE_HANDLE(VmaAllocator)
EXTENSION_NAME(VK_EXT_EXTENDED_DYNAMIC_STATE_3_EXTENSION_NAME) \
EXTENSION_NAME(VK_EXT_EXTERNAL_MEMORY_HOST_EXTENSION_NAME) \
EXTENSION_NAME(VK_EXT_4444_FORMATS_EXTENSION_NAME) \
EXTENSION_NAME(VK_EXT_IMAGE_ROBUSTNESS_EXTENSION_NAME) \
EXTENSION_NAME(VK_EXT_LINE_RASTERIZATION_EXTENSION_NAME) \
EXTENSION_NAME(VK_EXT_ROBUSTNESS_2_EXTENSION_NAME) \
EXTENSION_NAME(VK_EXT_VERTEX_INPUT_DYNAMIC_STATE_EXTENSION_NAME) \
@@ -176,7 +161,6 @@ VK_DEFINE_HANDLE(VmaAllocator)
FEATURE_NAME(depth_bias_control, depthBiasExact) \
FEATURE_NAME(extended_dynamic_state, extendedDynamicState) \
FEATURE_NAME(format_a4b4g4r4, formatA4B4G4R4) \
FEATURE_NAME(image_robustness, robustImageAccess) \
FEATURE_NAME(index_type_uint8, indexTypeUint8) \
FEATURE_NAME(primitive_topology_list_restart, primitiveTopologyListRestart) \
FEATURE_NAME(provoking_vertex, provokingVertexLast) \
@@ -456,11 +440,6 @@ public:
return extensions.swapchain_mutable_format;
}
/// Returns true if VK_EXT_swapchain_maintenance1 is enabled.
bool IsExtSwapchainMaintenance1Enabled() const {
return extensions.swapchain_maintenance1;
}
/// Returns true if VK_KHR_shader_float_controls is enabled.
bool IsKhrShaderFloatControlsSupported() const {
return extensions.shader_float_controls;
@@ -497,18 +476,10 @@ public:
}
/// Returns true if the device supports VK_EXT_shader_stencil_export.
/// Note: Most Mali/NVIDIA drivers don't support this. Use hardware blits as fallback.
bool IsExtShaderStencilExportSupported() const {
return extensions.shader_stencil_export;
}
/// Returns true if depth/stencil operations can be performed efficiently.
/// Either through shader export or hardware blits.
bool CanPerformDepthStencilOperations() const {
return extensions.shader_stencil_export || is_blit_depth24_stencil8_supported ||
is_blit_depth32_stencil8_supported;
}
/// Returns true if the device supports VK_EXT_depth_range_unrestricted.
bool IsExtDepthRangeUnrestrictedSupported() const {
return extensions.depth_range_unrestricted;
@@ -549,46 +520,6 @@ public:
return extensions.custom_border_color;
}
/// Returns true if the device supports VK_EXT_image_robustness.
bool IsExtImageRobustnessSupported() const {
return extensions.image_robustness;
}
/// Returns true if robustImageAccess is supported.
bool IsRobustImageAccessSupported() const {
return features.image_robustness.robustImageAccess;
}
/// Returns true if the device supports VK_EXT_robustness2.
bool IsExtRobustness2Supported() const {
return extensions.robustness_2;
}
/// Returns true if robustBufferAccess2 is supported.
bool IsRobustBufferAccess2Supported() const {
return features.robustness2.robustBufferAccess2;
}
/// Returns true if robustImageAccess2 is supported.
bool IsRobustImageAccess2Supported() const {
return features.robustness2.robustImageAccess2;
}
/// Returns true if nullDescriptor is supported.
bool IsNullDescriptorSupported() const {
return features.robustness2.nullDescriptor;
}
/// Returns true if customBorderColors feature is available.
bool IsCustomBorderColorsSupported() const {
return features.custom_border_color.customBorderColors;
}
/// Returns true if customBorderColorWithoutFormat feature is available.
bool IsCustomBorderColorWithoutFormatSupported() const {
return features.custom_border_color.customBorderColorWithoutFormat;
}
/// Returns true if the device supports VK_EXT_extended_dynamic_state.
bool IsExtExtendedDynamicStateSupported() const {
return extensions.extended_dynamic_state;
@@ -638,55 +569,6 @@ public:
return extensions.line_rasterization;
}
bool SupportsRectangularLines() const {
return features.line_rasterization.rectangularLines != VK_FALSE;
}
bool SupportsSmoothLines() const {
return features.line_rasterization.smoothLines != VK_FALSE;
}
bool SupportsStippledRectangularLines() const {
return features.line_rasterization.stippledRectangularLines != VK_FALSE;
}
bool SupportsAlphaToOne() const {
return features.features.alphaToOne != VK_FALSE;
}
bool SupportsDynamicState3DepthClampEnable() const {
return dynamic_state3_depth_clamp_enable;
}
bool SupportsDynamicState3LogicOpEnable() const {
return dynamic_state3_logic_op_enable;
}
bool SupportsDynamicState3LineRasterizationMode() const {
return dynamic_state3_line_raster_mode;
}
bool SupportsDynamicState3ConservativeRasterizationMode() const {
return dynamic_state3_conservative_raster_mode;
}
bool SupportsDynamicState3LineStippleEnable() const {
return dynamic_state3_line_stipple_enable;
}
bool SupportsDynamicState3AlphaToCoverageEnable() const {
return dynamic_state3_alpha_to_coverage;
}
bool SupportsDynamicState3AlphaToOneEnable() const {
return dynamic_state3_alpha_to_one;
}
/// Returns true when the user enabled extended core dynamic states (level > 0).
bool UsesAdvancedCoreDynamicState() const {
return Settings::values.dyna_state.GetValue() > 0;
}
/// Returns true if the device supports VK_EXT_vertex_input_dynamic_state.
bool IsExtVertexInputDynamicStateSupported() const {
return extensions.vertex_input_dynamic_state;
@@ -821,73 +703,6 @@ public:
return features2.features.multiViewport;
}
/// Returns true if the device supports VK_KHR_maintenance1.
bool IsKhrMaintenance1Supported() const {
return extensions.maintenance1;
}
/// Returns true if the device supports VK_KHR_maintenance2.
bool IsKhrMaintenance2Supported() const {
return extensions.maintenance2;
}
/// Returns true if the device supports VK_KHR_maintenance3.
bool IsKhrMaintenance3Supported() const {
return extensions.maintenance3;
}
/// Returns true if the device supports VK_KHR_maintenance4.
bool IsKhrMaintenance4Supported() const {
return extensions.maintenance4;
}
/// Returns true if the device supports VK_KHR_maintenance5.
bool IsKhrMaintenance5Supported() const {
return extensions.maintenance5;
}
/// Returns true if polygon mode POINT supports gl_PointSize.
bool SupportsPolygonModePointSize() const {
return extensions.maintenance5 && properties.maintenance5.polygonModePointSize;
}
/// Returns true if depth/stencil swizzle ONE is supported.
bool SupportsDepthStencilSwizzleOne() const {
return extensions.maintenance5 && properties.maintenance5.depthStencilSwizzleOneSupport;
}
/// Returns true if early fragment tests optimizations are available.
bool SupportsEarlyFragmentTests() const {
return extensions.maintenance5 &&
properties.maintenance5.earlyFragmentMultisampleCoverageAfterSampleCounting &&
properties.maintenance5.earlyFragmentSampleMaskTestBeforeSampleCounting;
}
/// Returns true if the device supports VK_KHR_maintenance6.
bool IsKhrMaintenance6Supported() const {
return extensions.maintenance6;
}
/// Returns true if the device supports VK_EXT_multi_draw.
bool IsExtMultiDrawSupported() const {
return extensions.multi_draw;
}
/// Returns true if the device supports VK_KHR_maintenance7.
bool IsKhrMaintenance7Supported() const {
return extensions.maintenance7;
}
/// Returns true if the device supports VK_KHR_maintenance8.
bool IsKhrMaintenance8Supported() const {
return extensions.maintenance8;
}
/// Returns true if the device supports VK_KHR_maintenance9.
bool IsKhrMaintenance9Supported() const {
return extensions.maintenance9;
}
[[nodiscard]] static constexpr bool CheckBrokenCompute(VkDriverId driver_id,
u32 driver_version) {
if (driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS) {
@@ -971,7 +786,6 @@ private:
FOR_EACH_VK_FEATURE_1_1(FEATURE);
FOR_EACH_VK_FEATURE_1_2(FEATURE);
FOR_EACH_VK_FEATURE_1_3(FEATURE);
FOR_EACH_VK_FEATURE_1_4(FEATURE);
FOR_EACH_VK_FEATURE_EXT(FEATURE);
FOR_EACH_VK_EXTENSION(EXTENSION);
@@ -988,7 +802,6 @@ private:
FOR_EACH_VK_FEATURE_1_1(FEATURE_CORE);
FOR_EACH_VK_FEATURE_1_2(FEATURE_CORE);
FOR_EACH_VK_FEATURE_1_3(FEATURE_CORE);
FOR_EACH_VK_FEATURE_1_4(FEATURE_CORE);
FOR_EACH_VK_FEATURE_EXT(FEATURE_EXT);
#undef FEATURE_CORE
@@ -1004,8 +817,6 @@ private:
VkPhysicalDevicePushDescriptorPropertiesKHR push_descriptor{};
VkPhysicalDeviceSubgroupSizeControlProperties subgroup_size_control{};
VkPhysicalDeviceTransformFeedbackPropertiesEXT transform_feedback{};
VkPhysicalDeviceMaintenance5PropertiesKHR maintenance5{};
VkPhysicalDeviceMultiDrawPropertiesEXT multi_draw{};
VkPhysicalDeviceProperties properties{};
};
@@ -1035,15 +846,8 @@ private:
bool cant_blit_msaa{}; ///< Does not support MSAA<->MSAA blitting.
bool must_emulate_scaled_formats{}; ///< Requires scaled vertex format emulation
bool must_emulate_bgr565{}; ///< Emulates BGR565 by swizzling RGB565 format.
bool dynamic_state3_blending{}; ///< Has blending features of dynamic_state3.
bool dynamic_state3_enables{}; ///< Has at least one enable feature of dynamic_state3.
bool dynamic_state3_depth_clamp_enable{};
bool dynamic_state3_logic_op_enable{};
bool dynamic_state3_line_raster_mode{};
bool dynamic_state3_conservative_raster_mode{};
bool dynamic_state3_line_stipple_enable{};
bool dynamic_state3_alpha_to_coverage{};
bool dynamic_state3_alpha_to_one{};
bool dynamic_state3_blending{}; ///< Has all blending features of dynamic_state3.
bool dynamic_state3_enables{}; ///< Has all enables features of dynamic_state3.
bool supports_conditional_barriers{}; ///< Allows barriers in conditional control flow.
u64 device_access_memory{}; ///< Total size of device local memory in bytes.
u32 sets_per_pool{}; ///< Sets per Description Pool

View File

@@ -116,8 +116,6 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept {
X(vkCmdDrawIndirectCount);
X(vkCmdDrawIndexedIndirectCount);
X(vkCmdDrawIndirectByteCountEXT);
X(vkCmdDrawMultiEXT);
X(vkCmdDrawMultiIndexedEXT);
X(vkCmdEndConditionalRenderingEXT);
X(vkCmdEndQuery);
X(vkCmdEndRenderPass);
@@ -147,8 +145,6 @@ void Load(VkDevice device, DeviceDispatch& dld) noexcept {
X(vkCmdSetDepthWriteEnableEXT);
X(vkCmdSetPrimitiveRestartEnableEXT);
X(vkCmdSetRasterizerDiscardEnableEXT);
X(vkCmdSetAlphaToCoverageEnableEXT);
X(vkCmdSetAlphaToOneEnableEXT);
X(vkCmdSetConservativeRasterizationModeEXT);
X(vkCmdSetLineRasterizationModeEXT);
X(vkCmdSetLineStippleEnableEXT);

View File

@@ -216,8 +216,6 @@ struct DeviceDispatch : InstanceDispatch {
PFN_vkCmdDrawIndirectCount vkCmdDrawIndirectCount{};
PFN_vkCmdDrawIndexedIndirectCount vkCmdDrawIndexedIndirectCount{};
PFN_vkCmdDrawIndirectByteCountEXT vkCmdDrawIndirectByteCountEXT{};
PFN_vkCmdDrawMultiEXT vkCmdDrawMultiEXT{};
PFN_vkCmdDrawMultiIndexedEXT vkCmdDrawMultiIndexedEXT{};
PFN_vkCmdEndConditionalRenderingEXT vkCmdEndConditionalRenderingEXT{};
PFN_vkCmdEndDebugUtilsLabelEXT vkCmdEndDebugUtilsLabelEXT{};
PFN_vkCmdEndQuery vkCmdEndQuery{};
@@ -240,8 +238,6 @@ struct DeviceDispatch : InstanceDispatch {
PFN_vkCmdSetDepthWriteEnableEXT vkCmdSetDepthWriteEnableEXT{};
PFN_vkCmdSetPrimitiveRestartEnableEXT vkCmdSetPrimitiveRestartEnableEXT{};
PFN_vkCmdSetRasterizerDiscardEnableEXT vkCmdSetRasterizerDiscardEnableEXT{};
PFN_vkCmdSetAlphaToCoverageEnableEXT vkCmdSetAlphaToCoverageEnableEXT{};
PFN_vkCmdSetAlphaToOneEnableEXT vkCmdSetAlphaToOneEnableEXT{};
PFN_vkCmdSetConservativeRasterizationModeEXT vkCmdSetConservativeRasterizationModeEXT{};
PFN_vkCmdSetLineRasterizationModeEXT vkCmdSetLineRasterizationModeEXT{};
PFN_vkCmdSetLineStippleEnableEXT vkCmdSetLineStippleEnableEXT{};
@@ -1243,19 +1239,6 @@ public:
counter_buffer_offset, counter_offset, stride);
}
void DrawMultiEXT(u32 draw_count, const VkMultiDrawInfoEXT* vertex_info,
u32 instance_count, u32 first_instance, u32 stride) const noexcept {
dld->vkCmdDrawMultiEXT(handle, draw_count, vertex_info, instance_count, first_instance,
stride);
}
void DrawMultiIndexedEXT(u32 draw_count, const VkMultiDrawIndexedInfoEXT* index_info,
u32 instance_count, u32 first_instance, u32 stride,
const int32_t* vertex_offset) const noexcept {
dld->vkCmdDrawMultiIndexedEXT(handle, draw_count, index_info, instance_count,
first_instance, stride, vertex_offset);
}
void ClearAttachments(Span<VkClearAttachment> attachments,
Span<VkClearRect> rects) const noexcept {
dld->vkCmdClearAttachments(handle, attachments.size(), attachments.data(), rects.size(),
@@ -1488,14 +1471,6 @@ public:
dld->vkCmdSetLogicOpEnableEXT(handle, enable ? VK_TRUE : VK_FALSE);
}
void SetAlphaToCoverageEnableEXT(bool enable) const noexcept {
dld->vkCmdSetAlphaToCoverageEnableEXT(handle, enable ? VK_TRUE : VK_FALSE);
}
void SetAlphaToOneEnableEXT(bool enable) const noexcept {
dld->vkCmdSetAlphaToOneEnableEXT(handle, enable ? VK_TRUE : VK_FALSE);
}
void SetDepthClampEnableEXT(bool enable) const noexcept {
dld->vkCmdSetDepthClampEnableEXT(handle, enable ? VK_TRUE : VK_FALSE);
}

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: 2016 Citra Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -74,11 +71,11 @@ private:
std::unique_ptr<InputProfiles> profiles;
std::array<ConfigureInputPlayer*, 8> player_controllers{};
std::array<QWidget*, 8> player_tabs{};
std::array<ConfigureInputPlayer*, 8> player_controllers;
std::array<QWidget*, 8> player_tabs;
// Checkboxes representing the "Connected Controllers".
std::array<QCheckBox*, 8> connected_controller_checkboxes{};
ConfigureInputAdvanced* advanced = nullptr;
std::array<QCheckBox*, 8> connected_controller_checkboxes;
ConfigureInputAdvanced* advanced;
Core::System& system;
};

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -692,10 +689,10 @@ void PlayerControlPreview::DrawHandheldController(QPainter& p, const QPointF cen
{
// Draw joysticks
using namespace Settings::NativeAnalog;
const auto& l_stick = QPointF(stick_values[LStick].x.value, stick_values[LStick].y.value);
const auto& l_button = button_values[Settings::NativeButton::LStick];
const auto& r_stick = QPointF(stick_values[RStick].x.value, stick_values[RStick].y.value);
const auto& r_button = button_values[Settings::NativeButton::RStick];
const auto l_stick = QPointF(stick_values[LStick].x.value, stick_values[LStick].y.value);
const auto l_button = button_values[Settings::NativeButton::LStick];
const auto r_stick = QPointF(stick_values[RStick].x.value, stick_values[RStick].y.value);
const auto r_button = button_values[Settings::NativeButton::RStick];
DrawJoystick(p, center + QPointF(-171, -41) + (l_stick * 4), 1.0f, l_button);
DrawJoystick(p, center + QPointF(171, 8) + (r_stick * 4), 1.0f, r_button);

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -217,7 +214,7 @@ private:
bool mapping_active{};
int blink_counter{};
int callback_key{};
int callback_key;
QColor button_color{};
ColorMapping colors{};
Core::HID::LedPattern led_pattern{0, 0, 0, 0};

View File

@@ -1495,7 +1495,7 @@ void MainWindow::OnAppFocusStateChanged(Qt::ApplicationState state) {
(state & (Qt::ApplicationHidden | Qt::ApplicationInactive))) {
auto_paused = true;
OnPauseGame();
} else if (!emu_thread->IsRunning() && auto_paused && (state & Qt::ApplicationActive)) {
} else if (!emu_thread->IsRunning() && auto_paused && state == Qt::ApplicationActive) {
auto_paused = false;
OnStartGame();
}
@@ -1505,7 +1505,7 @@ void MainWindow::OnAppFocusStateChanged(Qt::ApplicationState state) {
(state & (Qt::ApplicationHidden | Qt::ApplicationInactive))) {
Settings::values.audio_muted = true;
auto_muted = true;
} else if (auto_muted && (state & Qt::ApplicationActive)) {
} else if (auto_muted && state == Qt::ApplicationActive) {
Settings::values.audio_muted = false;
auto_muted = false;
}

View File

@@ -482,13 +482,13 @@ private:
MultiplayerState* multiplayer_state = nullptr;
GRenderWindow* render_window = nullptr;
GameList* game_list = nullptr;
LoadingScreen* loading_screen = nullptr;
GRenderWindow* render_window;
GameList* game_list;
LoadingScreen* loading_screen;
QTimer shutdown_timer;
OverlayDialog* shutdown_dialog{};
GameListPlaceholder* game_list_placeholder = nullptr;
GameListPlaceholder* game_list_placeholder;
std::vector<VkDeviceInfo::Record> vk_device_records;
@@ -531,7 +531,7 @@ private:
QString startup_icon_theme;
// Debugger panes
ControllerDialog* controller_dialog = nullptr;
ControllerDialog* controller_dialog;
QAction* actions_recent_files[max_recent_files_item];
@@ -543,7 +543,7 @@ private:
QTranslator translator;
// Install progress dialog
QProgressDialog* install_progress = nullptr;
QProgressDialog* install_progress;
// Last game booted, used for multi-process apps
QString last_filename_booted;

View File

@@ -1,6 +1,3 @@
// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
// SPDX-License-Identifier: GPL-3.0-or-later
// SPDX-FileCopyrightText: Copyright 2017 Citra Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
@@ -178,7 +175,7 @@ public:
private:
QString username;
QString nickname;
u64 title_id{};
u64 title_id;
QString game_name;
};

View File

@@ -14,7 +14,7 @@ try {
Exit 1
}
$VulkanSDKVer = "1.4.335.0"
$VulkanSDKVer = "1.4.328.1"
$VULKAN_SDK = "C:/VulkanSDK/$VulkanSDKVer"
$ExeFile = "vulkansdk-windows-X64-$VulkanSDKVer.exe"
$Uri = "https://sdk.lunarg.com/sdk/download/$VulkanSDKVer/windows/$ExeFile"

View File

@@ -2,7 +2,7 @@
# SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
# SPDX-License-Identifier: GPL-3.0-or-later
: "${VULKAN_SDK_VER:=1.4.335.0}"
: "${VULKAN_SDK_VER:=1.4.328.1}"
: "${VULKAN_ROOT:=C:/VulkanSDK/$VULKAN_SDK_VER}"
EXE_FILE="vulkansdk-windows-X64-$VULKAN_SDK_VER.exe"
URI="https://sdk.lunarg.com/sdk/download/$VULKAN_SDK_VER/windows/$EXE_FILE"