summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
-rw-r--r--src/video_core/engines/shader_bytecode.h16
-rw-r--r--src/video_core/shader/decode/texture.cpp44
-rw-r--r--src/video_core/shader/shader_ir.h2
3 files changed, 62 insertions, 0 deletions
diff --git a/src/video_core/engines/shader_bytecode.h b/src/video_core/engines/shader_bytecode.h
index ffb3ec3e0..5b32e1249 100644
--- a/src/video_core/engines/shader_bytecode.h
+++ b/src/video_core/engines/shader_bytecode.h
@@ -1232,6 +1232,20 @@ union Instruction {
} texs;
union {
+ BitField<28, 1, u64> is_array;
+ BitField<29, 2, TextureType> texture_type;
+ BitField<35, 1, u64> aoffi;
+ BitField<49, 1, u64> nodep_flag;
+ BitField<50, 1, u64> ms; // Multisample?
+ BitField<54, 1, u64> cl;
+ BitField<55, 1, u64> process_mode;
+
+ TextureProcessMode GetTextureProcessMode() const {
+ return process_mode == 0 ? TextureProcessMode::LZ : TextureProcessMode::LL;
+ }
+ } tld;
+
+ union {
BitField<49, 1, u64> nodep_flag;
BitField<53, 4, u64> texture_info;
@@ -1408,6 +1422,7 @@ public:
TXQ, // Texture Query
TXQ_B, // Texture Query Bindless
TEXS, // Texture Fetch with scalar/non-vec4 source/destinations
+ TLD, // Texture Load
TLDS, // Texture Load with scalar/non-vec4 source/destinations
TLD4, // Texture Load 4
TLD4S, // Texture Load 4 with scalar / non - vec4 source / destinations
@@ -1682,6 +1697,7 @@ private:
INST("1101111101001---", Id::TXQ, Type::Texture, "TXQ"),
INST("1101111101010---", Id::TXQ_B, Type::Texture, "TXQ_B"),
INST("1101-00---------", Id::TEXS, Type::Texture, "TEXS"),
+ INST("11011100--11----", Id::TLD, Type::Texture, "TLD"),
INST("1101101---------", Id::TLDS, Type::Texture, "TLDS"),
INST("110010----111---", Id::TLD4, Type::Texture, "TLD4"),
INST("1101111100------", Id::TLD4S, Type::Texture, "TLD4S"),
diff --git a/src/video_core/shader/decode/texture.cpp b/src/video_core/shader/decode/texture.cpp
index 4a356dbd4..b22831c64 100644
--- a/src/video_core/shader/decode/texture.cpp
+++ b/src/video_core/shader/decode/texture.cpp
@@ -245,6 +245,18 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
}
break;
}
+ case OpCode::Id::TLD: {
+ UNIMPLEMENTED_IF_MSG(instr.tld.aoffi, "AOFFI is not implemented");
+ UNIMPLEMENTED_IF_MSG(instr.tld.ms, "MS is not implemented");
+ UNIMPLEMENTED_IF_MSG(instr.tld.cl, "CL is not implemented");
+
+ if (instr.tld.nodep_flag) {
+ LOG_WARNING(HW_GPU, "TLD.NODEP implementation is incomplete");
+ }
+
+ WriteTexInstructionFloat(bb, instr, GetTldCode(instr));
+ break;
+ }
case OpCode::Id::TLDS: {
const Tegra::Shader::TextureType texture_type{instr.tlds.GetTextureType()};
const bool is_array{instr.tlds.IsArrayTexture()};
@@ -575,6 +587,38 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
return values;
}
+Node4 ShaderIR::GetTldCode(Tegra::Shader::Instruction instr) {
+ const auto texture_type{instr.tld.texture_type};
+ const bool is_array{instr.tld.is_array};
+ const bool lod_enabled{instr.tld.GetTextureProcessMode() == TextureProcessMode::LL};
+ const std::size_t coord_count{GetCoordCount(texture_type)};
+
+ u64 gpr8_cursor{instr.gpr8.Value()};
+ const Node array_register{is_array ? GetRegister(gpr8_cursor++) : nullptr};
+
+ std::vector<Node> coords;
+ for (std::size_t i = 0; i < coord_count; ++i) {
+ coords.push_back(GetRegister(gpr8_cursor++));
+ }
+
+ u64 gpr20_cursor{instr.gpr20.Value()};
+ // const Node bindless_register{is_bindless ? GetRegister(gpr20_cursor++) : nullptr};
+ const Node lod{lod_enabled ? GetRegister(gpr20_cursor++) : Immediate(0u)};
+ // const Node aoffi_register{is_aoffi ? GetRegister(gpr20_cursor++) : nullptr};
+ // const Node multisample{is_multisample ? GetRegister(gpr20_cursor++) : nullptr};
+
+ const auto& sampler = GetSampler(instr.sampler, texture_type, is_array, false);
+
+ Node4 values;
+ for (u32 element = 0; element < values.size(); ++element) {
+ auto coords_copy = coords;
+ MetaTexture meta{sampler, array_register, {}, {}, {}, lod, {}, element};
+ values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
+ }
+
+ return values;
+}
+
Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is_array) {
const std::size_t type_coord_count = GetCoordCount(texture_type);
const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL;
diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h
index edcf2288e..1b84c0672 100644
--- a/src/video_core/shader/shader_ir.h
+++ b/src/video_core/shader/shader_ir.h
@@ -277,6 +277,8 @@ private:
Node4 GetTld4Code(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
bool depth_compare, bool is_array, bool is_aoffi);
+ Node4 GetTldCode(Tegra::Shader::Instruction instr);
+
Node4 GetTldsCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
bool is_array);