diff --git a/core/libs/prx/libSceAgcDriver/Graphics/src/Pipeline.cpp b/core/libs/prx/libSceAgcDriver/Graphics/src/Pipeline.cpp index 31cd07f34..66a0d5504 100644 --- a/core/libs/prx/libSceAgcDriver/Graphics/src/Pipeline.cpp +++ b/core/libs/prx/libSceAgcDriver/Graphics/src/Pipeline.cpp @@ -95,7 +95,13 @@ void ValidateDepthBounds(const Context& context, const State& state) { void ValidateProvokingVertex(const Context& context, const State& state, std::span shaders) { Require(state.provokingVertexMode == VK_PROVOKING_VERTEX_MODE_FIRST_VERTEX_EXT || state.provokingVertexMode == VK_PROVOKING_VERTEX_MODE_LAST_VERTEX_EXT, "invalid provoking vertex mode"); - if (state.provokingVertexMode == VK_PROVOKING_VERTEX_MODE_FIRST_VERTEX_EXT) return; + if (state.provokingVertexMode == VK_PROVOKING_VERTEX_MODE_FIRST_VERTEX_EXT) { + for (const auto& shader : shaders) { + if (shader.stage != ShaderRecompiler::ShaderStage::Fragment || shader.program == nullptr) continue; + Require(std::none_of(shader.program->fragmentParameters.begin(), shader.program->fragmentParameters.end(), [](const ShaderRecompiler::FragmentParameter& parameter) { return parameter.flatP10; }), "v_interp_mov p10 of a flat input with the first provoking vertex is not measured"); + } + return; + } Require(context.provokingVertexLast, "last provoking vertex requires VK_EXT_provoking_vertex with provokingVertexLast enabled"); Require(!state.rectList && state.stages.path == ShaderPath::Vertex && !state.stages.mesh && !state.stages.tessellation, "last provoking vertex is unsupported for generated primitive pipelines"); Require(state.topology == VK_PRIMITIVE_TOPOLOGY_POINT_LIST || state.topology == VK_PRIMITIVE_TOPOLOGY_LINE_LIST || state.topology == VK_PRIMITIVE_TOPOLOGY_LINE_STRIP || state.topology == VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST || state.topology == VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP || state.topology == VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN, "last provoking vertex is unsupported for this primitive topology"); diff --git a/core/libs/prx/libSceAgcDriver/tests/Graphics.cpp b/core/libs/prx/libSceAgcDriver/tests/Graphics.cpp index 66bbdde1a..f1d04fa21 100644 --- a/core/libs/prx/libSceAgcDriver/tests/Graphics.cpp +++ b/core/libs/prx/libSceAgcDriver/tests/Graphics.cpp @@ -3965,6 +3965,14 @@ void ProvokingVertexTests() { Require(flatPixel.fragmentParameters.size() == 1 && flatPixel.fragmentParameters[0].flat && !flatPixel.fragmentParameters[0].perVertex, "flat v_interp_mov P0 has explicit per-vertex metadata"); const std::array flatShaders{{{ShaderRecompiler::ShaderStage::Fragment, &flatPixel, 0}}}; ValidateProvokingVertex(supported, last, flatShaders); + const std::array flatP10Color{0xc8120000u, 0xc8160100u, 0xc81a0200u, 0xc81e0300u, 0xf800180fu, 0x07060504u, 0xbf810000u}; + const auto flatP10Pixel = recompilePixel({0x400u}, flatP10Color); + const auto p10Inputs = locatedInputs(flatP10Pixel.spirv.Words()); + Require(p10Inputs.size() == 1 && p10Inputs[0].location == 0 && p10Inputs[0].flat && !p10Inputs[0].perVertex, "flat v_interp_mov P10 did not compile to a flat input"); + Require(flatP10Pixel.fragmentParameters.size() == 1 && flatP10Pixel.fragmentParameters[0].flatP10 && !flatP10Pixel.fragmentParameters[0].perVertex && !flatPixel.fragmentParameters[0].flatP10, "flat v_interp_mov P10 was not recorded apart from P0"); + const std::array flatP10Shaders{{{ShaderRecompiler::ShaderStage::Fragment, &flatP10Pixel, 0}}}; + ValidateProvokingVertex(supported, last, flatP10Shaders); + expectFailure([&] { ValidateProvokingVertex(supported, first, flatP10Shaders); }, "v_interp_mov p10 of a flat input with the first provoking vertex is not measured"); ShaderRecompiler::RecompileResult fragment{}; const std::array shaders{{{ShaderRecompiler::ShaderStage::Fragment, &fragment, 0}}}; for (const auto topology : {VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST, VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP, VK_PRIMITIVE_TOPOLOGY_TRIANGLE_FAN}) { diff --git a/core/shader/recompiler/Optimization/src/ShaderInfoCollector.cpp b/core/shader/recompiler/Optimization/src/ShaderInfoCollector.cpp index 7ea25b14f..e9481485b 100644 --- a/core/shader/recompiler/Optimization/src/ShaderInfoCollector.cpp +++ b/core/shader/recompiler/Optimization/src/ShaderInfoCollector.cpp @@ -257,7 +257,7 @@ void CollectPixelInputs(const IrProgram& program, const ShaderPixelInputInfo* pi const auto input = inst->Argument(0)->Resolve()->ImmediateU32(); const auto mode = inst->Argument(2)->Resolve()->ImmediateU32(); read[input] = true; - perVertex[input] = perVertex[input] || mode < 2u || !IsPixelParameterFlat(*pixel, input); + perVertex[input] = perVertex[input] || mode == 1u || !IsPixelParameterFlat(*pixel, input); } else if (inst->Opcode() == IrOpcode::GetInterpolationParameterF16) { const auto input = inst->Argument(0)->Resolve()->ImmediateU32(); const bool high = inst->Argument(3)->Resolve()->ImmediateU32() != 0u; diff --git a/core/shader/recompiler/Recompiler.hpp b/core/shader/recompiler/Recompiler.hpp index 8c7cf526f..121ce4fce 100644 --- a/core/shader/recompiler/Recompiler.hpp +++ b/core/shader/recompiler/Recompiler.hpp @@ -391,6 +391,7 @@ struct FragmentParameter { bool flat; bool perVertex; bool custom = false; + bool flatP10 = false; }; struct BarycentricEmulation { diff --git a/core/shader/recompiler/ShaderDiskCache.cpp b/core/shader/recompiler/ShaderDiskCache.cpp index 24b6a9e37..1d85b6668 100644 --- a/core/shader/recompiler/ShaderDiskCache.cpp +++ b/core/shader/recompiler/ShaderDiskCache.cpp @@ -323,6 +323,7 @@ void encodeArtifact(Writer& writer, const CompiledShaderArtifact& result) { out.Value(parameter.flat); out.Value(parameter.perVertex); out.Value(parameter.custom); + out.Value(parameter.flatP10); }); writer.Value(result.barycentricEmulation.active); writer.Value(result.barycentricEmulation.smooth); @@ -366,6 +367,7 @@ void decodeArtifact(Reader& reader, CompiledShaderArtifact& result) { in.Value(parameter.flat); in.Value(parameter.perVertex); in.Value(parameter.custom); + in.Value(parameter.flatP10); }); reader.Value(result.barycentricEmulation.active); reader.Value(result.barycentricEmulation.smooth); diff --git a/core/shader/recompiler/SpirvBackend/src/SpirvMemory/SpirvInputOutput.cpp b/core/shader/recompiler/SpirvBackend/src/SpirvMemory/SpirvInputOutput.cpp index 2c3eb07f5..a7a465d86 100644 --- a/core/shader/recompiler/SpirvBackend/src/SpirvMemory/SpirvInputOutput.cpp +++ b/core/shader/recompiler/SpirvBackend/src/SpirvMemory/SpirvInputOutput.cpp @@ -48,11 +48,25 @@ namespace ShaderRecompiler if (inputInfo.pixel == nullptr) FailEmit("pixel input info is missing"); const auto& pixel = *inputInfo.pixel; std::vector result; + std::array readsP10{}; + for (const auto& block : program.Blocks()) { + for (const auto* instruction : block->Instructions()) { + if (instruction->Opcode() != IrOpcode::GetInterpolationParameter || instruction->Argument(2)->Resolve()->ImmediateU32() != 0u) continue; + const auto attribute = instruction->Argument(0)->Resolve()->ImmediateU32(); + if (attribute < readsP10.size()) readsP10[attribute] = true; + } + } for (const auto& input : program.Info().inputs) { if (input.kind != StageInputKind::Parameter) continue; const auto location = PixelInputLocation(pixel, input.location); - if (std::any_of(result.begin(), result.end(), [&](const auto& parameter) { return parameter.location == location; })) continue; - result.push_back({location, location, PixelInputIsFlat(pixel, input.location), input.perVertex, PixelInputIsCustom(pixel, input.location)}); + const bool flat = PixelInputIsFlat(pixel, input.location); + const bool flatP10 = flat && !input.perVertex && input.location < readsP10.size() && readsP10[input.location]; + const auto existing = std::find_if(result.begin(), result.end(), [&](const auto& parameter) { return parameter.location == location; }); + if (existing != result.end()) { + existing->flatP10 = existing->flatP10 || flatP10; + continue; + } + result.push_back({location, location, flat, input.perVertex, PixelInputIsCustom(pixel, input.location), flatP10}); } return result; } diff --git a/docs/dev/TechnicalDebt.md b/docs/dev/TechnicalDebt.md index 3c866d60a..8c372b2af 100644 --- a/docs/dev/TechnicalDebt.md +++ b/docs/dev/TechnicalDebt.md @@ -571,7 +571,7 @@ Throughout the project, every function at every stage either **does exactly what - [Overlay output bus](../../core/libs/prx/libSceVideoOut/src/VideoOutDriver.cpp) (libSceVideoOut) - flips on a port opened with `SCE_VIDEO_OUT_BUS_TYPE_OVERLAY` complete and signal their events like main-bus flips, but their buffers are not shown: the overlay is not composited over the main output. Presenting them as whole frames made the window alternate between the two buses - [PA_CL_CLIP_CNTL](../../core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp) (libSceAgcDriver) - `DX_LINEAR_ATTR_CLIP_ENA` (bit 24) clear is drawn like bit 24 set, with Vulkan's clipping of `NoPerspective` outputs; what a clear bit changes on the console is undocumented. - [10_11_11 unorm color targets](../../core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp) (libSceAgcDriver) render into R32_UINT attachments: the pixel shader packs each export, rounded to nearest even like format-30 image stores, which is assumed to match the CB's conversion. -- [PA_SU_SC_MODE_CNTL.PROVOKING_VTX_LAST](../../core/libs/prx/libSceAgcDriver/Graphics/src/Pipeline.cpp) (libSceAgcDriver) uses Vulkan's last-vertex mode for native primitives. Bit 19 follows [opengnm's `flatshade_first` mapping](https://github.com/blackbearreloaded/ps5-opengl/blob/db752da4c89380af228497b5b6852163a4e61069/src/gallium/ps5/ps5_screen.c#L2244); vertex selection follows [Vulkan](https://docs.vulkan.org/spec/latest/chapters/drawing.html#drawing-primitive-topologies), without console measurements. Last mode throws without `VK_EXT_provoking_vertex`, for generated rectangle, mesh or tessellation primitives, explicit per-vertex flat inputs, and triangle strips with per-vertex inputs, whose [ordering depends on the device](https://docs.vulkan.org/refpages/latest/refpages/source/VkPhysicalDeviceFragmentShaderBarycentricPropertiesKHR.html). +- [PA_SU_SC_MODE_CNTL.PROVOKING_VTX_LAST](../../core/libs/prx/libSceAgcDriver/Graphics/src/Pipeline.cpp) (libSceAgcDriver) uses Vulkan's last-vertex mode for native primitives. Bit 19 follows [opengnm's `flatshade_first` mapping](https://github.com/blackbearreloaded/ps5-opengl/blob/db752da4c89380af228497b5b6852163a4e61069/src/gallium/ps5/ps5_screen.c#L2244); vertex selection follows [Vulkan](https://docs.vulkan.org/spec/latest/chapters/drawing.html#drawing-primitive-topologies), without console measurements. Last mode throws without `VK_EXT_provoking_vertex`, for generated rectangle, mesh or tessellation primitives, explicit per-vertex flat inputs, and triangle strips with per-vertex inputs, whose [ordering depends on the device](https://docs.vulkan.org/refpages/latest/refpages/source/VkPhysicalDeviceFragmentShaderBarycentricPropertiesKHR.html). A FLAT_SHADE input without pass-through that is read through `v_interp_mov_f32` P10 and not P20 [returns the provoking vertex's value](../../core/shader/recompiler/Optimization/src/ShaderInfoCollector.cpp), like P0. ps5-opengl reads every flat input that way, and its driver passes the GL 4.6 CTS on a PS5 (#1413). The lane itself was not measured on the console, so with the first provoking vertex such a shader throws. - [v_rcp, v_rsq, v_sqrt, v_log, v_exp, v_sin and v_cos](../../core/shader/recompiler/Translation/src/FloatInstructions.cpp) (f32 and f16) use the driver's approximation for finite results, not the console's: the low bits can differ. Only NaN, invalid, zero and infinite results and the exact sin/cos zeros are chosen on the bits. Measured on RDNA2 (gfx1036) over every f16 input and 32,832 f32 inputs: RADV returns the hardware's bits for every input of rcp, rsq, sqrt and log and all but one of exp (f16 0x11c5, 1 ULP); NVIDIA (615.71.09) stays within 2 ULP except `v_log_f32` near 1.0 (next entry). [v_sin and v_cos](../../core/shader/recompiler/SpirvBackend/src/SpirvAlu/SpirvAluEmitterMath.cpp) take the driver's sine or cosine of the radians of the exact cycle, except within 2^-12 of a turn from a zero of the function, where the result is the linear term of the exact distance to the zero: the hardware's constant 6.2831826 (0x40c90fd5) at the integers, which the hardware applies exactly to inputs below 2^-12 of a turn, and the f32 2π at the half and quarter turns, where the hardware is within 7 ULP of it. Outside those bands, a driver result more than 4 ULP from a Taylor series in turns of the exact cycle is replaced by the series, which is within 2 ULP of the true sine and cosine; the hardware is up to 9 ULP from them (gfx1035, 243,092 f32 inputs), so results stay within 13 ULP of the hardware on any driver. RADV returns the hardware's bits for every f16 input and for 32,422 sine and 32,486 cosine results of the 32,832 f32 inputs; the rest are 1 ULP off, except up to 14 ULP on either side of non-zero integers and up to 15 ULP beyond ±256 cycles, where the hardware's own reduction differs from the exact fraction, and up to 6 ULP inside the half- and quarter-turn bands. NVIDIA's own sine is within 1 ULP on 93% of the f32 inputs and up to 1,166 ULP off just outside the bands, its cosine within 163 ULP; the series replaces those results. A denormal f32 input with f32 denormals kept gives the hardware the input times 6.2831826; both hosts flush it. - [v_log_f32 and v_log_f16](../../core/shader/recompiler/SpirvBackend/src/SpirvAlu/SpirvAluEmitterMath.cpp) keep the driver's log2 unless, for inputs within 1/8 of 1.0, it is more than 4 ULP from an eight-term series in x - 1, which then replaces it. `v_log_f16` goes through the same f32 lowering and is rounded to half. The series is evaluated for every `v_log` and chosen with `OpSelect` rather than a branch, so its seven fused multiply-adds and two multiplies are an unconditional cost. Measured on RDNA2 over all 3,145,728 f32 inputs within 1/8 of 1.0 (gfx1035, IEEE 0, f32 denormals flushed), the hardware is within 1 ULP of the correctly rounded log2 and the series within 2 ULP of it and within 3 ULP of the hardware (2 except at 0x3f752e47, 0x3f7baab0 and 0x3f7bc4fb), so a kept driver value is within 7 ULP of the hardware and a driver that returns the hardware's bits never triggers the gate; RADV (Mesa 26.2.4) returns the hardware's bits on the 9,280 such inputs measured on gfx1036; NVIDIA (615.71.09) was up to 23,947,836 ULP off at 1 - 2^-24 and is within 6 ULP inside the band after the change; outside it its own log2 stays up to 9 ULP off (measured up to 1.375). Over every f16 input, the change moves one `v_log_f16` result on NVIDIA (0x3bed, 1 ULP, onto the hardware's bits) and none on RADV. - [Dual-source blending](../../core/libs/prx/libSceAgcDriver/Graphics/src/State.cpp) (libSceAgcDriver) takes the MRT1 export as the second source of the MRT0 blend, as RADV and radeonsi program it; what the CB does with other export or target layouts is unknown, so they throw.