shader: add missing RDNA2 ALU opcodes

Add missing scalar and vector opcodes with compute tests.

Original work by @Pouare514 in #176.

Co-authored-by: Pouare514 <julienaulycee@gmail.com>

src/graphics/host_gpu/renderer/debug.cpp

@@ -490,7 +490,7 @@ static void ClipCheck(const HW::ClipControl& c) {
 	// dx_linear_attr_clip_enable preserves linear (noperspective) attributes at clip-generated
 	// vertices, which Vulkan provides as part of clipping and interpolation.
 	EXIT_NOT_IMPLEMENTED(c.user_clip_planes != 0 || c.user_clip_plane_mode != 0 ||
-	                     c.vertex_kill_any || c.user_clip_plane_negate_y || c.clip_disable ||
+	                     c.vertex_kill_any || c.user_clip_plane_negate_y ||
 	                     c.user_clip_plane_cull_only || c.cull_on_clipping_error_disable ||
 	                     c.force_viewport_index_from_vs_enable);
 }

src/graphics/host_gpu/renderer/image/image.cpp

@@ -59,16 +59,8 @@ namespace {
 	if (HasFormatFeature(properties, vk::FormatFeatureFlagBits::eColorAttachment)) {
 		usage |= vk::ImageUsageFlagBits::eColorAttachment;
 	}
-	if (info.samples == 1 &&
-	    HasFormatFeature(properties, vk::FormatFeatureFlagBits::eStorageImage)) {
+	if (info.samples == 1) {
 		usage |= vk::ImageUsageFlagBits::eStorage;
-	} else if (info.samples == 1) {
-		const auto compatible = SrgbStorageViewFormat(info.pixel_format);
-		if (compatible != vk::Format::eUndefined &&
-		    HasFormatFeature(graphics.GetFormatProperties(compatible),
-		                     vk::FormatFeatureFlagBits::eStorageImage)) {
-			usage |= vk::ImageUsageFlagBits::eStorage;
-		}
 	}
 	return usage;
 }

src/graphics/host_gpu/renderer/pipeline/pipelineCache.cpp

@@ -116,7 +116,7 @@ struct PipelineCache::ProgramCache {
 	}
 
 	static constexpr std::size_t MaxStaticKeyWords =
-	    5 + ShaderVertexInputInfo::RES_MAX * 17;
+	    12 + ShaderVertexInputInfo::RES_MAX * 17;
 
 	template <typename InputInfo>
 	ShaderProgram Get(const ShaderParams& params, InputInfo& input_info) {
@@ -340,8 +340,23 @@ void PipelineCache::Save() {
 
 ShaderProgram PipelineCache::GetVertexProgram(const HW::VertexShaderInfo& regs,
                                               const HW::ShaderRegisters&  sh,
+                                              const HW::Context&          context,
                                               ShaderVertexInputInfo&      input_info) {
 	const auto params = PrepareProgram(regs, sh, input_info);
+	if (context.GetClipControl().clip_disable) {
+		const auto& viewport = context.GetScreenViewport().viewports[0];
+		const auto& limits   = m_graphics.GetPhysicalDeviceProperties().limits;
+		auto&       clip     = input_info.clip_space;
+		clip.scale[0]        = viewport.xscale;
+		clip.scale[1]        = viewport.yscale;
+		clip.offset[0]       = viewport.xoffset;
+		clip.offset[1]       = viewport.yoffset;
+		clip.half_extent[0] =
+		    static_cast<float>(std::min(limits.maxViewportDimensions[0], 16384u)) * 0.5f;
+		clip.half_extent[1] =
+		    static_cast<float>(std::min(limits.maxViewportDimensions[1], 16384u)) * 0.5f;
+		clip.enabled = true;
+	}
 	Common::LockGuard lock(m_mutex);
 	return m_program_cache->Get(params, input_info);
 }

src/graphics/host_gpu/renderer/pipeline/pipelineCache.h

@@ -130,6 +130,7 @@ public:
 
 	ShaderProgram GetVertexProgram(const HW::VertexShaderInfo& regs,
 	                               const HW::ShaderRegisters&  sh,
+	                               const HW::Context&          context,
 	                               ShaderVertexInputInfo&      input_info);
 	ShaderProgram
 	GetPixelProgram(const HW::PixelShaderInfo& regs, const HW::ShaderRegisters& sh,

src/graphics/host_gpu/renderer/renderDraw.cpp

@@ -331,10 +331,18 @@ static void SetGraphicsDynamicParams(const CommandBuffer& buffer, vk::CommandBuf
 	const auto final_scissor = calc_final_scissor(vp, ctx.GetScanModeControl(), framebuffer_extent);
 
 	vk::Viewport viewport {};
-	viewport.x        = vp.viewports[0].xoffset - vp.viewports[0].xscale;
-	viewport.y        = vp.viewports[0].yoffset - vp.viewports[0].yscale;
-	viewport.width    = vp.viewports[0].xscale * 2.0f;
-	viewport.height   = vp.viewports[0].yscale * 2.0f;
+	if (ctx.GetClipControl().clip_disable) {
+		const auto& limits = buffer.GetGraphics().GetPhysicalDeviceProperties().limits;
+		viewport.x         = 0.0f;
+		viewport.y         = 0.0f;
+		viewport.width     = static_cast<float>(std::min(limits.maxViewportDimensions[0], 16384u));
+		viewport.height    = static_cast<float>(std::min(limits.maxViewportDimensions[1], 16384u));
+	} else {
+		viewport.x      = vp.viewports[0].xoffset - vp.viewports[0].xscale;
+		viewport.y      = vp.viewports[0].yoffset - vp.viewports[0].yscale;
+		viewport.width  = vp.viewports[0].xscale * 2.0f;
+		viewport.height = vp.viewports[0].yscale * 2.0f;
+	}
 	viewport.minDepth = vp.viewports[0].zoffset;
 	viewport.maxDepth = vp.viewports[0].zscale + vp.viewports[0].zoffset;
 	vk_buffer.setViewport(0, 1, &viewport);
@@ -1013,8 +1021,8 @@ static void RefreshShaders(CommandBuffer& buffer, const DrawCallInfo& draw, bool
 		LogDrawPhase(draw.name, "GetVertexProgram");
 	}
 	auto& pipeline_cache = buffer.GetContext().GetPipelineCache();
-	state.vertex_program =
-	    pipeline_cache.GetVertexProgram(vertex_shader_info, shader_regs, state.vs_input_info);
+	state.vertex_program = pipeline_cache.GetVertexProgram(vertex_shader_info, shader_regs, ctx,
+	                                                         state.vs_input_info);
 
 	if (!state.ps_active) {
 		return;

src/graphics/presentation/window/window.cpp

@@ -814,6 +814,7 @@ static void WindowCreate(WindowContext& context) {
 	int width  = static_cast<int>(context.graphic_ctx.screen_width);
 	int height = static_cast<int>(context.graphic_ctx.screen_height);
 
+	SDL_SetHint(SDL_HINT_JOYSTICK_HIDAPI_PS5_RUMBLE, "1");
 #if KYTY_PLATFORM == KYTY_PLATFORM_WINDOWS
 	SDL_SetHint(SDL_HINT_WINDOWS_DPI_SCALING, "0");
 #endif

src/graphics/shader/recompiler/backend/spirv/spirvEmitterFlow.cpp

@@ -408,6 +408,41 @@ void EmitAuxPositionExport(ValueEmitContext& ctx, uint32_t data, const IR::Expor
 	}
 }
 
+uint32_t ConvertClipCoordinate(EmitterState& state, uint32_t coordinate, float scale,
+                               float offset, float half_extent) {
+	const auto window  = state.builder.AllocateId();
+	const auto biased  = state.builder.AllocateId();
+	const auto divided = state.builder.AllocateId();
+	const auto ndc     = state.builder.AllocateId();
+	state.builder.AddFunction(
+	    {OpFMul, TypeF32(state), window, coordinate, ConstantF32Value(state, scale)});
+	state.builder.AddFunction(
+	    {OpFAdd, TypeF32(state), biased, window, ConstantF32Value(state, offset)});
+	state.builder.AddFunction(
+	    {OpFDiv, TypeF32(state), divided, biased, ConstantF32Value(state, half_extent)});
+	state.builder.AddFunction(
+	    {OpFSub, TypeF32(state), ndc, divided, ConstantF32Value(state, 1.0f)});
+	return ndc;
+}
+
+uint32_t ConvertPositionToClipSpace(EmitterState& state, uint32_t position) {
+	const auto& transform = state.input_info.vertex->clip_space;
+	uint32_t    components[4] {};
+	for (uint32_t i = 0; i < 4; i++) {
+		components[i] = state.builder.AllocateId();
+		state.builder.AddFunction(
+		    {OpCompositeExtract, TypeF32(state), components[i], position, i});
+	}
+	components[0] = ConvertClipCoordinate(state, components[0], transform.scale[0],
+	                                      transform.offset[0], transform.half_extent[0]);
+	components[1] = ConvertClipCoordinate(state, components[1], transform.scale[1],
+	                                      transform.offset[1], transform.half_extent[1]);
+	const auto converted = state.builder.AllocateId();
+	state.builder.AddFunction({OpCompositeConstruct, TypeF32Vector(state, 4), converted,
+	                           components[0], components[1], components[2], components[3]});
+	return converted;
+}
+
 void EmitExport(ValueEmitContext& ctx, const IR::Inst& inst) {
 	auto&       state = ctx.state;
 	const auto& exp   = ctx.Export(inst);
@@ -467,6 +502,9 @@ void EmitExport(ValueEmitContext& ctx, const IR::Inst& inst) {
 			}
 		}
 		if (exp.kind == IR::ExportTargetKind::Position) {
+			if (state.stage == ShaderType::Vertex && state.input_info.vertex->clip_space.enabled) {
+				value = ConvertPositionToClipSpace(state, value);
+			}
 			const auto pointer = state.builder.AllocateId();
 			state.builder.AddFunction(
 			    {OpAccessChain, TypePointer(state, StorageClassOutput, TypeF32Vector(state, 4)),

src/graphics/shader/recompiler/frontend/decode/ScalarAluOps.cpp

@@ -27,7 +27,8 @@ constexpr OpcodeMap SOP2_OPCODE_LIST[] = {
     {0x22u, Opcode::S_ASHR_I32},        {0x24u, Opcode::S_BFM_B32},
     {0x25u, Opcode::S_BFM_B64},         {0x26u, Opcode::S_MUL_I32},
     {0x27u, Opcode::S_BFE_U32},         {0x28u, Opcode::S_BFE_I32},
-    {0x29u, Opcode::S_BFE_U64},         {0x2eu, Opcode::S_LSHL1_ADD_U32},
+    {0x29u, Opcode::S_BFE_U64},         {0x2cu, Opcode::S_ABSDIFF_I32},
+    {0x2eu, Opcode::S_LSHL1_ADD_U32},
     {0x2fu, Opcode::S_LSHL2_ADD_U32},   {0x30u, Opcode::S_LSHL3_ADD_U32},
     {0x31u, Opcode::S_LSHL4_ADD_U32},   {0x32u, Opcode::S_PACK_LL_B32_B16},
     {0x33u, Opcode::S_PACK_LH_B32_B16}, {0x34u, Opcode::S_PACK_HH_B32_B16},
@@ -48,7 +49,9 @@ constexpr OpcodeMap SOP1_OPCODE_LIST[] = {
     {0x15u, Opcode::S_FLBIT_I32_B32},
     {0x16u, Opcode::S_FLBIT_I32_B64},
     {0x1bu, Opcode::S_BITSET0_B32},
+    {0x1cu, Opcode::S_BITSET0_B64},
     {0x1du, Opcode::S_BITSET1_B32},
+    {0x1eu, Opcode::S_BITSET1_B64},
     {0x1fu, Opcode::S_GETPC_B64},
     {0x20u, Opcode::S_SETPC_B64},
     {0x24u, Opcode::S_AND_SAVEEXEC_B64},
@@ -93,6 +96,7 @@ constexpr OpcodeMap SOPP_OPCODE_LIST[] = {
     {0x0au, Opcode::S_BARRIER},
     {0x0cu, Opcode::S_WAITCNT},
     {0x0eu, Opcode::S_SLEEP},
+    {0x0fu, Opcode::S_SETPRIO},
     {0x10u, Opcode::S_SENDMSG},
     {0x12u, Opcode::S_TRAP},
     {0x16u, Opcode::S_TTRACEDATA},
@@ -276,6 +280,7 @@ void DecodeSopp(uint32_t pc, std::span<const uint32_t> code, uint32_t word_index
 	                           : static_cast<int32_t>(static_cast<int16_t>(simm));
 	inst.src_count = (inst.opcode == Opcode::S_NOP || inst.opcode == Opcode::S_WAITCNT ||
 	                  inst.opcode == Opcode::S_WAITCNT_DEPCTR || inst.opcode == Opcode::S_SLEEP ||
+	                  inst.opcode == Opcode::S_SETPRIO ||
 	                  inst.opcode == Opcode::S_SENDMSG || inst.opcode == Opcode::S_TRAP ||
 	                  inst.opcode == Opcode::S_TTRACEDATA || inst.opcode == Opcode::S_INST_PREFETCH)
 	                     ? 1

src/graphics/shader/recompiler/frontend/decode/ShaderDecoder.cpp

@@ -479,6 +479,8 @@ std::string InstructionToString(const Instruction& inst) {
 		case Opcode::S_FLBIT_I32_B32:
 		case Opcode::S_FF1_I32_B32:
 		case Opcode::S_FF1_I32_B64:
+		case Opcode::S_BITSET0_B64:
+		case Opcode::S_BITSET1_B64:
 		case Opcode::S_NOT_B64:
 		case Opcode::S_WQM_B64:
 		case Opcode::S_QUADMASK_B64:
@@ -505,6 +507,7 @@ std::string InstructionToString(const Instruction& inst) {
 		case Opcode::S_WAITCNT:
 		case Opcode::S_WAITCNT_DEPCTR:
 		case Opcode::S_SLEEP:
+		case Opcode::S_SETPRIO:
 		case Opcode::S_TRAP:
 		case Opcode::S_SENDMSG:
 		case Opcode::S_TTRACEDATA:

src/graphics/shader/recompiler/frontend/decode/ShaderDecoder.h

@@ -40,6 +40,7 @@ enum class Opcode {
 	S_MOV_B64,
 	S_MOVK_I32,
 	S_ABS_I32,
+	S_ABSDIFF_I32,
 	S_BREV_B32,
 	S_BCNT1_I32_B32,
 	S_BCNT1_I32_B64,
@@ -69,6 +70,8 @@ enum class Opcode {
 	S_BITCMP1_B32,
 	S_BITSET0_B32,
 	S_BITSET1_B32,
+	S_BITSET0_B64,
+	S_BITSET1_B64,
 	S_MIN_I32,
 	S_MAX_I32,
 	S_MIN_U32,
@@ -108,6 +111,7 @@ enum class Opcode {
 	S_BFM_B64,
 	S_CSELECT_B32,
 	S_CSELECT_B64,
+	S_SETPRIO,
 	S_PACK_LL_B32_B16,
 	S_PACK_LH_B32_B16,
 	S_PACK_HH_B32_B16,
@@ -282,6 +286,7 @@ enum class Opcode {
 	V_MAD_MIXHI_F16,
 	V_ADD_NC_U32,
 	V_ADDC_U32,
+	V_SUB_CO_CI_U32,
 	V_SUBREV_CO_CI_U32,
 	V_SUB_NC_U32,
 	V_SUBREV_NC_U32,

src/graphics/shader/recompiler/frontend/decode/VectorAluOps.cpp

@@ -61,6 +61,7 @@ constexpr Vop2OpcodeInfo VOP2_OPCODE_LIST[] = {
     {0x24u, Opcode::V_MBCNT_HI_U32_B32},
     {0x25u, Opcode::V_ADD_NC_U32, Vop2SdwaProfile::IntegerPartialDestination},
     {0x28u, Opcode::V_ADDC_U32, Vop2SdwaProfile::IntegerFullDestination},
+    {0x29u, Opcode::V_SUB_CO_CI_U32},
     {0x2au, Opcode::V_SUBREV_CO_CI_U32},
     {0x26u, Opcode::V_SUB_NC_U32, Vop2SdwaProfile::IntegerPartialDestination},
     {0x27u, Opcode::V_SUBREV_NC_U32, Vop2SdwaProfile::IntegerFullDestination},
@@ -1033,6 +1034,7 @@ void FinalizeVop2Instruction(std::span<const uint32_t> code, uint32_t word_index
 			inst.src_count = 3;
 			break;
 		case Opcode::V_ADDC_U32:
+		case Opcode::V_SUB_CO_CI_U32:
 		case Opcode::V_SUBREV_CO_CI_U32:
 			inst.dst2.kind = OperandKind::VccLo;
 			inst.src2.kind = OperandKind::VccLo;
@@ -1643,6 +1645,7 @@ void DecodeVop3(uint32_t pc, std::span<const uint32_t> code, uint32_t word_index
 	SetRawWords(inst, code, word_index, 2);
 
 	const bool carry_in_out    = (inst.opcode == Opcode::V_ADDC_U32 && opcode == 0x128u) ||
+	                             (inst.opcode == Opcode::V_SUB_CO_CI_U32 && opcode == 0x129u) ||
 	                             (inst.opcode == Opcode::V_SUBREV_CO_CI_U32 && opcode == 0x12au);
 	const bool vop3b_carry_out = IsVop3BCarryOutOpcode(inst.opcode);
 	const bool vop3b_mad_u64   = IsVop3BMadU64Opcode(inst.opcode);

src/graphics/shader/recompiler/frontend/translate/Control.cpp

@@ -76,9 +76,9 @@ void Translator::SUB_U32(const Decoder::Instruction& inst, bool vector, bool rev
 	}
 }
 
-void Translator::SUBB_U32(const Decoder::Instruction& inst, bool vector) {
-	const auto lhs       = ReadU32(vector ? inst.src1 : inst.src0);
-	const auto rhs       = ReadU32(vector ? inst.src0 : inst.src1);
+void Translator::SUBB_U32(const Decoder::Instruction& inst, bool vector, bool reverse) {
+	const auto lhs       = ReadU32(reverse ? inst.src1 : inst.src0);
+	const auto rhs       = ReadU32(reverse ? inst.src0 : inst.src1);
 	const auto borrow_in = vector ? ConditionBit(inst.src2)
 	                              : ConditionBit(ConditionOperand(Decoder::OperandKind::Scc));
 	const auto partial   = ir.ISub(lhs, rhs);
@@ -94,6 +94,13 @@ void Translator::SUBB_U32(const Decoder::Instruction& inst, bool vector) {
 	}
 }
 
+void Translator::S_ABSDIFF_I32(const Decoder::Instruction& inst) {
+	const auto difference = ir.ISub(ReadU32(inst.src0), ReadU32(inst.src1));
+	const auto result     = IR::U32(ir.Emit(IR::ValueOpcode::IAbs32, {difference}));
+	WriteOperand(DestinationOperand(inst), result);
+	ir.SetScc(ir.INotEqual(result, IR::U32(IR::Value(0u))));
+}
+
 void Translator::S_ADD_SUB_I32(const Decoder::Instruction& inst, bool subtract) {
 	const auto lhs      = ReadU32(inst.src0);
 	const auto rhs      = ReadU32(inst.src1);

src/graphics/shader/recompiler/frontend/translate/Integer.cpp

@@ -312,6 +312,22 @@ bool Translator::S_BITSET_B32(const Decoder::Instruction& inst, bool set) {
 	return true;
 }
 
+bool Translator::S_BITSET_B64(const Decoder::Instruction& inst, bool set) {
+	const auto offset    = ir.BitwiseAnd(ReadU32(inst.src0), IR::U32(IR::Value(63u)));
+	const auto word_bit  = ir.BitwiseAnd(offset, IR::U32(IR::Value(31u)));
+	const auto bit       = ir.ShiftLeftLogical(IR::U32(IR::Value(1u)), word_bit);
+	const auto old       = ReadU32Pair(inst.dst);
+	const auto low_value = set ? ir.BitwiseOr(old[0], bit)
+	                           : ir.BitwiseAnd(old[0], ir.BitwiseNot(bit));
+	const auto high_value = set ? ir.BitwiseOr(old[1], bit)
+	                            : ir.BitwiseAnd(old[1], ir.BitwiseNot(bit));
+	const auto high = IR::U1(ir.Emit(IR::ValueOpcode::UGreaterThanEqual32,
+	                                {offset, IR::Value(32u)}));
+	WriteU32Pair(inst.dst,
+	             {ir.Select(high, old[0], low_value), ir.Select(high, high_value, old[1])});
+	return true;
+}
+
 bool Translator::V_BCNT_U32_B32(const Decoder::Instruction& inst) {
 	const auto count = IR::U32(ir.Emit(IR::ValueOpcode::BitCount32, {ReadU32(inst.src0)}));
 	WriteOperand(DestinationOperand(inst), ir.IAdd(count, ReadU32(inst.src1)));

src/graphics/shader/recompiler/frontend/translate/Scalar.cpp

@@ -34,7 +34,8 @@ bool Translator::EmitScalar(const Decoder::Instruction& inst) {
 		case O::S_ADD_U32: ADD_U32(inst, false, false); return true;
 		case O::S_ADDC_U32: ADD_U32(inst, false, true); return true;
 		case O::S_SUB_U32: SUB_U32(inst, false, false); return true;
-		case O::S_SUBB_U32: SUBB_U32(inst, false); return true;
+		case O::S_SUBB_U32: SUBB_U32(inst, false, false); return true;
+		case O::S_ABSDIFF_I32: S_ABSDIFF_I32(inst); return true;
 		case O::S_ADD_I32: S_ADD_SUB_I32(inst, false); return true;
 		case O::S_SUB_I32: S_ADD_SUB_I32(inst, true); return true;
 		case O::S_LSHL1_ADD_U32: S_LSHL_ADD_U32(inst, 1u); return true;
@@ -186,6 +187,8 @@ bool Translator::EmitScalar(const Decoder::Instruction& inst) {
 
 		case O::S_BITSET0_B32: return S_BITSET_B32(inst, false);
 		case O::S_BITSET1_B32: return S_BITSET_B32(inst, true);
+		case O::S_BITSET0_B64: return S_BITSET_B64(inst, false);
+		case O::S_BITSET1_B64: return S_BITSET_B64(inst, true);
 		case O::S_BITREPLICATE_B64_B32: return S_BITREPLICATE_B64_B32(inst);
 		case O::S_QUADMASK_B64: return S_QUADMASK_B64(inst);
 		case O::S_BFM_B32: return BFM_B32(inst);
@@ -201,6 +204,7 @@ bool Translator::EmitScalar(const Decoder::Instruction& inst) {
 
 		case O::S_NOP:
 		case O::S_SLEEP:
+		case O::S_SETPRIO:
 		case O::S_TRAP: EmitControlNop(); return true;
 		case O::S_WAITCNT_DEPCTR: EmitWaitcnt(); return true;
 		case O::S_BARRIER: S_BARRIER(); return true;

src/graphics/shader/recompiler/frontend/translate/Translator.h

@@ -192,6 +192,7 @@ private:
 	bool V_SAD_U32(const Decoder::Instruction& inst);
 	bool V_ADD3_U32(const Decoder::Instruction& inst);
 	bool S_BITSET_B32(const Decoder::Instruction& inst, bool set);
+	bool S_BITSET_B64(const Decoder::Instruction& inst, bool set);
 	bool V_BCNT_U32_B32(const Decoder::Instruction& inst);
 	bool V_MBCNT_U32_B32(const Decoder::Instruction& inst, bool low);
 	bool S_BITREPLICATE_B64_B32(const Decoder::Instruction& inst);
@@ -218,7 +219,8 @@ private:
 	                bool negate_source, bool write_64);
 	void ADD_U32(const Decoder::Instruction& inst, bool vector, bool use_carry_in);
 	void SUB_U32(const Decoder::Instruction& inst, bool vector, bool reverse);
-	void SUBB_U32(const Decoder::Instruction& inst, bool vector);
+	void SUBB_U32(const Decoder::Instruction& inst, bool vector, bool reverse);
+	void S_ABSDIFF_I32(const Decoder::Instruction& inst);
 	void S_ADD_SUB_I32(const Decoder::Instruction& inst, bool subtract);
 	void S_LSHL_ADD_U32(const Decoder::Instruction& inst, uint32_t shift_amount);
 	void ScalarMinMax32(const Decoder::Instruction& inst, IR::ValueOpcode value_opcode,

src/graphics/shader/recompiler/frontend/translate/Vector.cpp

@@ -10,7 +10,8 @@ bool Translator::EmitVector(const Decoder::Instruction& inst) {
 		case O::V_ADDC_U32: ADD_U32(inst, true, true); return true;
 		case O::V_SUB_I32: SUB_U32(inst, true, false); return true;
 		case O::V_SUBREV_I32: SUB_U32(inst, true, true); return true;
-		case O::V_SUBREV_CO_CI_U32: SUBB_U32(inst, true); return true;
+		case O::V_SUB_CO_CI_U32: SUBB_U32(inst, true, false); return true;
+		case O::V_SUBREV_CO_CI_U32: SUBB_U32(inst, true, true); return true;
 
 		case O::V_MOV_B32: MOV_B32(inst, true); return true;
 		case O::V_MOVRELS_B32: V_MOVRELS_B32(inst); return true;

src/graphics/shader/shader.cpp

@@ -25,6 +25,7 @@
 
 #include <algorithm>
 #include <atomic>
+#include <bit>
 #include <cstdio>
 #include <filesystem>
 #include <fmt/format.h>
@@ -897,6 +898,18 @@ void BuildStageStaticKey(const ShaderVertexInputInfo& info, std::vector<uint32_t
 	key.push_back(info.resources_num);
 	key.push_back(info.scratch_size_dwords);
 	key.push_back(info.pa_cl_vs_out_cntl);
+	key.push_back(static_cast<uint32_t>(info.clip_space.enabled));
+	if (info.clip_space.enabled) {
+		for (const float value: info.clip_space.scale) {
+			key.push_back(std::bit_cast<uint32_t>(value));
+		}
+		for (const float value: info.clip_space.offset) {
+			key.push_back(std::bit_cast<uint32_t>(value));
+		}
+		for (const float value: info.clip_space.half_extent) {
+			key.push_back(std::bit_cast<uint32_t>(value));
+		}
+	}
 
 	for (int i = 0; i < info.resources_num; i++) {
 		const auto& resource    = info.resources[i];

src/graphics/shader/shader.h

@@ -68,6 +68,13 @@ inline uint8_t GetDstSel(uint32_t swizzle, uint32_t channel) {
 	return (swizzle >> (channel * 3u)) & 0x7u;
 }
 
+struct ShaderClipSpaceTransform {
+	float scale[2]       = {};
+	float offset[2]      = {};
+	float half_extent[2] = {};
+	bool  enabled        = false;
+};
+
 struct ShaderVertexInputInfo {
 	static constexpr int RES_MAX = 32;
 
@@ -84,6 +91,7 @@ struct ShaderVertexInputInfo {
 	uint32_t                scratch_size_dwords = 0;
 	uint32_t                param_export_mask   = 0;
 	uint32_t                pa_cl_vs_out_cntl    = 0;
+	ShaderClipSpaceTransform clip_space;
 	bool                    fetch_external      = false;
 	bool                    fetch_embedded      = false;
 };

src/libs/controller.cpp

@@ -39,11 +39,42 @@ struct PadControllerInformation {
 	uint8_t  reserve[8];
 };
 
+struct PadLightBarParam {
+	uint8_t r;
+	uint8_t g;
+	uint8_t b;
+	uint8_t reserve;
+};
+
 struct PadVibrationParam {
 	uint8_t large_motor;
 	uint8_t small_motor;
 };
 
+struct PadTriggerEffectCommand {
+	uint32_t mode;
+	uint8_t  reserve[4];
+	uint8_t  data[48];
+};
+
+struct PadTriggerEffectParam {
+	uint8_t                 trigger_mask;
+	uint8_t                 reserve[7];
+	PadTriggerEffectCommand command[2];
+};
+
+static_assert(sizeof(PadTriggerEffectCommand) == 56);
+static_assert(sizeof(PadTriggerEffectParam) == 120);
+
+struct DualSenseEffects {
+	uint8_t enable_bits;
+	uint8_t reserve[9];
+	uint8_t right_trigger[11];
+	uint8_t left_trigger[11];
+};
+
+static_assert(sizeof(DualSenseEffects) == 32);
+
 struct ControllerState {
 	struct Touch {
 		uint8_t  id   = 0;
@@ -73,6 +104,8 @@ public:
 	void ResetInputState();
 	void GetConnectionInfo(bool* flag, int* count);
 	void SetVibration(uint8_t large_motor, uint8_t small_motor);
+	void SetLightBar(uint8_t r, uint8_t g, uint8_t b);
+	bool SetTriggerEffect(const PadTriggerEffectParam& param);
 	void ReadState(ControllerState* state, bool* flag, int* count);
 	int  ReadStates(ControllerState* states, int states_num, bool* flag, int* count);
 
@@ -125,6 +158,92 @@ static void pad_fill_data(PadData* data, const ControllerState& state, bool conn
 	data->device_unique_data_len = 0;
 }
 
+static bool trigger_effect_zones(uint8_t* effect, const uint8_t* strengths, uint8_t type,
+                                 uint8_t frequency = 0) {
+	uint16_t active = 0;
+	uint32_t packed = 0;
+	for (int i = 0; i < 10; i++) {
+		if (strengths[i] > 8) {
+			return false;
+		}
+		if (strengths[i] != 0) {
+			active |= static_cast<uint16_t>(1u << i);
+			packed |= static_cast<uint32_t>(strengths[i] - 1u) << (3 * i);
+		}
+	}
+
+	effect[0] = active != 0 && (type != 0x26 || frequency != 0) ? type : 0x05;
+	effect[1] = static_cast<uint8_t>(active);
+	effect[2] = static_cast<uint8_t>(active >> 8u);
+	effect[3] = static_cast<uint8_t>(packed);
+	effect[4] = static_cast<uint8_t>(packed >> 8u);
+	effect[5] = static_cast<uint8_t>(packed >> 16u);
+	effect[6] = static_cast<uint8_t>(packed >> 24u);
+	effect[9] = frequency;
+	return true;
+}
+
+static bool trigger_effect_to_dualsense(const PadTriggerEffectCommand& command, uint8_t* effect) {
+	std::memset(effect, 0, 11);
+	effect[0] = 0x05;
+
+	uint8_t strengths[10] = {};
+	switch (command.mode) {
+		case 0: return true;
+		case 1:
+			if (command.data[0] > 9 || command.data[1] > 8) {
+				return false;
+			}
+			std::fill(strengths + command.data[0], strengths + 10, command.data[1]);
+			return trigger_effect_zones(effect, strengths, 0x21);
+		case 2: {
+			const auto start    = command.data[0];
+			const auto end      = command.data[1];
+			const auto strength = command.data[2];
+			if (start < 2 || start > 7 || end <= start || end > 8 || strength > 8) {
+				return false;
+			}
+			if (strength == 0) {
+				return true;
+			}
+			const uint16_t zones = static_cast<uint16_t>((1u << start) | (1u << end));
+			effect[0]            = 0x25;
+			effect[1]            = static_cast<uint8_t>(zones);
+			effect[2]            = static_cast<uint8_t>(zones >> 8u);
+			effect[3]            = strength - 1u;
+			return true;
+		}
+		case 3:
+			if (command.data[0] > 9 || command.data[1] > 8) {
+				return false;
+			}
+			std::fill(strengths + command.data[0], strengths + 10, command.data[1]);
+			return trigger_effect_zones(effect, strengths, 0x26, command.data[2]);
+		case 4: return trigger_effect_zones(effect, command.data, 0x21);
+		case 5: {
+			const int start          = command.data[0];
+			const int end            = command.data[1];
+			const int start_strength = command.data[2];
+			const int end_strength   = command.data[3];
+			if (start > 8 || end <= start || end > 9 || start_strength < 1 || start_strength > 8 ||
+			    end_strength < 1 || end_strength > 8) {
+				return false;
+			}
+			for (int i = start; i < 10; i++) {
+				const int delta  = (end_strength - start_strength) * (i - start);
+				const int length = end - start;
+				strengths[i]     = static_cast<uint8_t>(
+				    i >= end ? end_strength
+				             : start_strength +
+				                   (delta + (delta < 0 ? -length / 2 : length / 2)) / length);
+			}
+			return trigger_effect_zones(effect, strengths, 0x21);
+		}
+		case 6: return trigger_effect_zones(effect, command.data + 1, 0x26, command.data[0]);
+		default: return false;
+	}
+}
+
 void Initialize() {
 	EXIT_IF(g_controller != nullptr);
 
@@ -304,6 +423,44 @@ void GameController::SetVibration(uint8_t large_motor, uint8_t small_motor) {
 	}
 }
 
+void GameController::SetLightBar(uint8_t r, uint8_t g, uint8_t b) {
+	Common::LockGuard lock(m_mutex);
+	if (auto* pad = SDL_GameControllerFromInstanceID(static_cast<SDL_JoystickID>(m_active_id));
+	    pad != nullptr) {
+		(void)SDL_GameControllerSetLED(pad, r, g, b);
+	}
+}
+
+bool GameController::SetTriggerEffect(const PadTriggerEffectParam& param) {
+	if ((param.trigger_mask & ~0x03u) != 0) {
+		return false;
+	}
+
+	DualSenseEffects effect {};
+	if ((param.trigger_mask & 0x01u) != 0) {
+		effect.enable_bits |= 0x08;
+		if (!trigger_effect_to_dualsense(param.command[0], effect.left_trigger)) {
+			return false;
+		}
+	}
+	if ((param.trigger_mask & 0x02u) != 0) {
+		effect.enable_bits |= 0x04;
+		if (!trigger_effect_to_dualsense(param.command[1], effect.right_trigger)) {
+			return false;
+		}
+	}
+	if (effect.enable_bits == 0) {
+		return true;
+	}
+
+	Common::LockGuard lock(m_mutex);
+	auto* pad = SDL_GameControllerFromInstanceID(static_cast<SDL_JoystickID>(m_active_id));
+	if (pad != nullptr && SDL_GameControllerGetType(pad) == SDL_CONTROLLER_TYPE_PS5) {
+		(void)SDL_GameControllerSendEffect(pad, &effect, sizeof(effect));
+	}
+	return true;
+}
+
 void GameController::GetConnectionInfo(bool* flag, int* count) {
 	EXIT_IF(flag == nullptr);
 	EXIT_IF(count == nullptr);
@@ -617,7 +774,24 @@ int KYTY_SYSV_ABI PadSetLightBar(int handle, const PadLightBarParam* param) {
 		return PAD_ERROR_INVALID_ARG;
 	}
 
+	EXIT_IF(g_controller == nullptr);
+	g_controller->SetLightBar(param->r, param->g, param->b);
+
 	return OK;
 }
 
+int KYTY_SYSV_ABI PadSetTriggerEffect(int handle, const PadTriggerEffectParam* param) {
+	PRINT_NAME();
+
+	if (handle != 1) {
+		return PAD_ERROR_INVALID_HANDLE;
+	}
+	if (param == nullptr) {
+		return PAD_ERROR_INVALID_ARG;
+	}
+
+	EXIT_IF(g_controller == nullptr);
+	return g_controller->SetTriggerEffect(*param) ? OK : PAD_ERROR_INVALID_ARG;
+}
+
 } // namespace Libs::Controller

src/libs/controller.h

@@ -48,6 +48,7 @@ struct PadControllerInformation;
 struct PadData;
 struct PadVibrationParam;
 struct PadLightBarParam;
+struct PadTriggerEffectParam;
 
 inline int controller_get_axis(int min, int max, int value) {
 	int v = (255 * (value - min)) / (max - min);
@@ -78,6 +79,7 @@ int KYTY_SYSV_ABI PadGetControllerInformation(int handle, PadControllerInformati
 int KYTY_SYSV_ABI PadReadState(int handle, PadData* data);
 int KYTY_SYSV_ABI PadRead(int handle, PadData* data, int num);
 int KYTY_SYSV_ABI PadSetVibration(int handle, const PadVibrationParam* param);
+int KYTY_SYSV_ABI PadSetTriggerEffect(int handle, const PadTriggerEffectParam* param);
 int KYTY_SYSV_ABI PadResetLightBar(int handle);
 int KYTY_SYSV_ABI PadSetLightBar(int handle, const PadLightBarParam* param);
 

src/libs/libPad.cpp

@@ -42,23 +42,6 @@ static int KYTY_SYSV_ABI PadGetTriggerEffectState(int
 	return 0;
 }
 
-static int KYTY_SYSV_ABI PadSetTriggerEffect(int handle, const void* param) {
-	PRINT_NAME();
-
-	LOGF("\t handle = %d\n"
-	     "\t param  = 0x%016" PRIx64 "\n",
-	     handle, reinterpret_cast<uint64_t>(param));
-
-	if (handle != 1) {
-		return -2137915389; /* 0x80920003 */
-	}
-	if (param == nullptr) {
-		return -2137915391; /* 0x80920001 */
-	}
-
-	return OK;
-}
-
 static int KYTY_SYSV_ABI PadSetVibrationTriggerEffectWeakWhileEmbeddedMicInUse(bool enabled) {
 	PRINT_NAME();
 
@@ -198,7 +181,7 @@ LIB_DEFINE(InitPad_1) {
 	LIB_FUNC("rIZnR6eSpvk", Controller::PadResetOrientation);
 	LIB_FUNC("vDLMoJLde8I", PadSetTiltCorrectionState);
 	LIB_FUNC("gjP9-KQzoUk", Controller::PadGetControllerInformation);
-	LIB_FUNC("2JgFB2n9oUM", PadSetTriggerEffect);
+	LIB_FUNC("2JgFB2n9oUM", Controller::PadSetTriggerEffect);
 	LIB_FUNC("YndgXqQVV7c", Controller::PadReadState);
 	LIB_FUNC("q1cHNfGycLI", Controller::PadRead);
 	LIB_FUNC("yFVnOdGxvZY", Controller::PadSetVibration);

tests/ShaderRecompilerComputeTests.cpp

@@ -2950,6 +2950,28 @@ public:
             "dynamic view identity omitted mapping, format, mip, layer, type, "
             "or storage usage");
 
+    auto video_out_info = color_info;
+    video_out_info.pixel_format = vk::Format::eA2R10G10B10UnormPack32;
+    video_out_info.guest_format = Prospero::BufferFormat::k10_10_10_2UNorm;
+    Libs::Graphics::Image video_out(m_runtime_context, scheduler,
+                                    video_out_info);
+    auto video_out_view_info = sampled;
+    video_out_view_info.format = video_out_info.pixel_format;
+    video_out_view_info.usage = vk::ImageUsageFlagBits::eTransferSrc;
+    const auto video_out_view = video_out.FindView(video_out_view_info);
+    auto storage_alias_info = video_out_view_info;
+    storage_alias_info.format = vk::Format::eA2B10G10R10UnormPack32;
+    storage_alias_info.usage = vk::ImageUsageFlagBits::eStorage;
+    const auto storage_alias_view = video_out.FindView(storage_alias_info);
+    Require(
+        name, "video-out storage alias usage",
+        ImageViewOps::FormatsCompatible(video_out_info.pixel_format,
+                                        storage_alias_info.format) &&
+            static_cast<bool>(video_out.backing.usage &
+                              vk::ImageUsageFlagBits::eStorage) &&
+            video_out_view != nullptr && storage_alias_view != nullptr,
+        "compatible video-out backing was created without storage usage");
+
     ImageInfo depth_info{};
     depth_info.pixel_format = vk::Format::eD32Sfloat;
     depth_info.guest_format = Prospero::BufferFormat::k32Float;
@@ -12655,6 +12677,76 @@ TestCase ScalarAbsI32UpdatesScc() {
            O::V_MOV_B32, O::BUFFER_STORE_DWORD, O::S_ENDPGM}};
 }
 
+TestCase ScalarAbsdiffBitset64AndSetprio() {
+  using O = ShaderOpcode;
+
+  std::vector<u32> code;
+  auto capture_scc = [&](u32 dst_sgpr) {
+    code.push_back(
+        EncodeSop2(0x0a, dst_sgpr, InlineU32(1), InlineU32(0)));
+  };
+
+  code.push_back(EncodeSMovB32(0, InlineU32(2)));
+  code.push_back(EncodeSMovB32(1, InlineU32(5)));
+  code.push_back(EncodeSop2(0x2c, 10, 0, 1));
+  capture_scc(20);
+  AppendSMovLiteral(&code, 0, 0xffffffffu);
+  code.push_back(EncodeSMovB32(1, InlineU32(0)));
+  code.push_back(EncodeSop2(0x2c, 11, 0, 1));
+  capture_scc(21);
+  AppendSMovLiteral(&code, 0, 0x80000000u);
+  code.push_back(EncodeSop2(0x2c, 12, 0, 1));
+  capture_scc(22);
+  code.push_back(EncodeSMovB32(1, InlineU32(1)));
+  code.push_back(EncodeSop2(0x2c, 13, 0, 1));
+  capture_scc(23);
+  code.push_back(EncodeSop2(0x2c, 14, 0, 0));
+  capture_scc(24);
+
+  AppendSMovLiteral(&code, 2, 65u);
+  AppendSMovLiteral(&code, 3, 101u);
+  code.push_back(EncodeSMovB32(4, InlineU32(0)));
+  code.push_back(EncodeSMovB32(5, InlineU32(0)));
+  code.push_back(EncodeSop1(0x1e, 4, 2));
+  AppendStoreSgprPair(&code, 4, 10);
+  code.push_back(EncodeSop1(0x1e, 4, 3));
+  AppendStoreSgprPair(&code, 4, 12);
+  code.push_back(EncodeSop1(0x1c, 4, 2));
+  AppendStoreSgprPair(&code, 4, 14);
+  code.push_back(EncodeSop1(0x1c, 4, 3));
+  AppendStoreSgprPair(&code, 4, 16);
+  code.push_back(EncodeSopp(0x0f, 3));
+
+  AppendStoreSgpr(&code, 10, 0);
+  AppendStoreSgpr(&code, 11, 1);
+  AppendStoreSgpr(&code, 12, 2);
+  AppendStoreSgpr(&code, 13, 3);
+  AppendStoreSgpr(&code, 14, 4);
+  AppendStoreSgpr(&code, 20, 5);
+  AppendStoreSgpr(&code, 21, 6);
+  AppendStoreSgpr(&code, 22, 7);
+  AppendStoreSgpr(&code, 23, 8);
+  AppendStoreSgpr(&code, 24, 9);
+  AppendEnd(&code);
+
+  TestCase test;
+  test.name = "ScalarAbsdiffBitset64AndSetprio";
+  test.code = std::move(code);
+  test.expected = {3u, 1u, 0x80000000u, 0x7fffffffu, 0u,
+                   1u, 1u, 1u,          1u,          0u,
+                   2u, 0u, 2u,          32u,         0u,
+                   32u, 0u, 0u};
+  test.opcodes = {O::S_MOV_B32,     O::S_ABSDIFF_I32, O::S_CSELECT_B32,
+                  O::S_BITSET1_B64, O::S_BITSET0_B64, O::S_SETPRIO,
+                  O::V_MOV_B32,
+                  O::BUFFER_STORE_DWORD, O::S_ENDPGM};
+  test.decoded_counts = {{"S_ABSDIFF_I32", 5},
+                         {"S_BITSET1_B64", 2},
+                         {"S_BITSET0_B64", 2},
+                         {"S_SETPRIO 0x00000003", 1}};
+  return test;
+}
+
 TestCase ScalarShiftLeftAddSccCarryEdges() {
   using O = ShaderOpcode;
 
@@ -14155,6 +14247,55 @@ TestCase VectorAddcWritesPerLaneCarryOut() {
   return test;
 }
 
+TestCase VectorSubCoCiU32CompactAndVop3() {
+  using O = ShaderOpcode;
+
+  std::vector<u32> code;
+  code.push_back(EncodeVop2(0x1a, 4, InlineU32(2), 0));
+  AppendBufferLoadDword(&code, 1, 4);
+  code.push_back(EncodeVop2(0x25, 5, InlineU32(16), 4));
+  AppendBufferLoadDword(&code, 2, 5);
+  code.push_back(EncodeVop2(0x1b, 3, InlineU32(1), 0));
+  code.push_back(EncodeVopc(0xc5, InlineU32(0), 3));
+  code.push_back(EncodeSop1(0x04, 20, 106));
+  code.push_back(EncodeVop2(0x29, 10, Vgpr(1), 2));
+  AppendVMovU32(&code, 6, 1);
+  code.push_back(EncodeVop2(0x01, 11, InlineU32(0), 6));
+  AppendStoreVgprAtLaneDwordOffset(&code, 10, 0, 0);
+  AppendStoreVgprAtLaneDwordOffset(&code, 11, 0, 4);
+
+  code.push_back(EncodeVopc(0xc2, InlineU32(0), 6));
+  AppendVop3B(&code, 0x129u, 12, 22, Vgpr(1), Vgpr(2), 20);
+  AppendVop3(&code, 0x101u, 13, InlineU32(0), Vgpr(6), 22);
+  code.push_back(EncodeVop2(0x01, 14, InlineU32(0), 6));
+  AppendStoreVgprAtLaneDwordOffset(&code, 12, 0, 8);
+  AppendStoreVgprAtLaneDwordOffset(&code, 13, 0, 12);
+  AppendStoreVgprAtLaneDwordOffset(&code, 14, 0, 16);
+  AppendEnd(&code);
+
+  TestCase test;
+  test.name = "VectorSubCoCiU32CompactAndVop3";
+  test.code = std::move(code);
+  test.initial = {5u, 5u, 3u, 3u, 3u, 3u, 5u, 3u, 0u, 0u,
+                  0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u};
+  test.expected = {2u, 1u, 0xfffffffeu, 0xffffffffu, 0u, 0u, 1u,
+                   1u, 2u, 1u, 0xfffffffeu, 0xffffffffu, 0u, 0u,
+                   1u, 1u, 0u, 0u, 0u, 0u};
+  test.opcodes = {O::V_MOV_B32, O::V_LSHLREV_B32, O::V_ADD_NC_U32,
+                  O::V_AND_B32, O::V_CMP_NE_U32, O::V_CMP_EQ_U32,
+                  O::V_SUB_CO_CI_U32, O::V_CNDMASK_B32, O::S_MOV_B64,
+                  O::BUFFER_LOAD_DWORD, O::BUFFER_STORE_DWORD, O::S_ENDPGM};
+  test.decoded_counts = {{"V_SUB_CO_CI_U32", 2}};
+  test.required_spirv = {"OpISub", "OpUGreaterThan", "OpLogicalOr"};
+  test.compute_info.threads_num[0] = 4;
+  test.compute_info.threads_num[1] = 1;
+  test.compute_info.threads_num[2] = 1;
+  test.compute_info.thread_ids_num = 1;
+  test.compute_info.wave_size = 32;
+  test.has_compute_info = true;
+  return test;
+}
+
 TestCase VectorSubrevCoCiU32ExactRawOnGpu() {
   using O = ShaderOpcode;
 
@@ -21146,6 +21287,7 @@ std::vector<TestCase> MakeCases() {
   AddCase(ScalarArithmeticSccCarryBorrowOverflow);
   AddCase(ScalarMinMaxSccComparisonEdges);
   AddCase(ScalarAbsI32UpdatesScc);
+  AddCase(ScalarAbsdiffBitset64AndSetprio);
   AddCase(ScalarShiftLeftAddSccCarryEdges);
   AddCase(ScalarCompareOps);
   AddCase(ScalarShiftAddAndMaskOps);
@@ -21191,6 +21333,7 @@ std::vector<TestCase> MakeCases() {
   AddCase(VectorMbcntUsesThreadMask);
   AddCase(VectorAddcWritesPerLaneCarryOut);
   AddCase(VectorAddcUsesPerLaneCarryIn);
+  AddCase(VectorSubCoCiU32CompactAndVop3);
   AddCase(VectorSubrevCoCiU32ExactRawOnGpu);
   AddCase(VectorVop3BSubrevCoCiUsesEncodedMasks);
   AddCase(VectorVop3BCarryOutWritesSgprMask);
@@ -23439,6 +23582,7 @@ void CheckStorageTextureDepthTileUploadLayout() {
           "1x1 R8_UINT depth tile lost its 64 KiB source footprint");
   std::printf("[host]    %-32s ok\n", "StorageTextureDepthTileUpload");
 }
+
 void CheckImageSamplerSpecialization() {
   ShaderSamplerResource filtered_sampler{};
   filtered_sampler.fields[2] =

tests/shaderCfgTests.cpp

@@ -8811,6 +8811,45 @@ void TestNewShaderRecompilerExpVertexOutputs() {
   CheckSpirvBinaryValidates(result.spirv);
 }
 
+void TestNewShaderRecompilerClipDisabledPosition() {
+  const uint32_t shader[] = {
+      EncodeExp0(0x0c, 0xf), EncodeExp1(0, 1, 2, 3), 0xbf810000u,
+  };
+  const auto compile = [&](const ShaderVertexInputInfo &vertex) {
+    auto options = MakeCompileOptions(ShaderType::Vertex);
+    options.input_info.vertex = &vertex;
+    return ShaderRecompiler::Recompile(shader, options);
+  };
+
+  ShaderVertexInputInfo regular{};
+  auto clipped = regular;
+  clipped.clip_space = {
+      .scale = {640.0f, 360.0f},
+      .offset = {640.0f, 360.0f},
+      .half_extent = {8192.0f, 8192.0f},
+      .enabled = true,
+  };
+  const auto regular_result = compile(regular);
+  const auto clipped_result = compile(clipped);
+  CheckSpirvBinaryValidates(clipped_result.spirv);
+  Check(SpirvInstructionOpcodeCount(regular_result.spirv, 136u) == 0u &&
+            SpirvInstructionOpcodeCount(clipped_result.spirv, 133u) == 2u &&
+            SpirvInstructionOpcodeCount(clipped_result.spirv, 129u) == 2u &&
+            SpirvInstructionOpcodeCount(clipped_result.spirv, 136u) == 2u &&
+            SpirvInstructionOpcodeCount(clipped_result.spirv, 131u) == 2u,
+        "clip-disabled position export did not convert both screen coordinates to NDC");
+  Check(MakeStageStaticKey(regular) != MakeStageStaticKey(clipped),
+        "clip-disabled vertex shader shared the regular shader cache key");
+  auto disabled = clipped;
+  disabled.clip_space.enabled = false;
+  Check(MakeStageStaticKey(regular) == MakeStageStaticKey(disabled),
+        "disabled clip-space payload affected the shader cache key");
+  auto shifted = clipped;
+  shifted.clip_space.offset[0] += 1.0f;
+  Check(MakeStageStaticKey(clipped) != MakeStageStaticKey(shifted),
+        "clip-disabled viewport transform is absent from the shader cache key");
+}
+
 void TestNewShaderRecompilerAuxPositionExports() {
   const auto compile = [](std::span<const uint32_t> shader, uint32_t control) {
     ShaderVertexInputInfo vertex{};
@@ -11908,6 +11947,7 @@ int main() {
   TestNewShaderRecompilerSpirvSizeBaselines();
   TestDemandDrivenSpirvDeclarations();
   TestNewShaderRecompilerSMovB32();
+  TestNewShaderRecompilerClipDisabledPosition();
   TestNewShaderRecompilerAuxPositionExports();
   TestNewShaderRecompilerNativeWideScalarMemoryIr();
   TestNewShaderRecompilerNativeWideBufferIr();