From 4e35177e23069ad7a4cb0fdfa2ad5b34300c44f7 Mon Sep 17 00:00:00 2001
From: ReinUsesLisp <reinuseslisp@airmail.cc>
Date: Fri, 9 Aug 2019 23:50:21 -0300
Subject: shader_ir: Implement VOTE

Implement VOTE using Nvidia's intrinsics. Documentation about these can
be found here
https://developer.nvidia.com/reading-between-threads-shader-intrinsics

Instead of using portable ARB instructions I opted to use Nvidia
intrinsics because these are the closest we have to how Tegra X1
hardware renders.

To stub VOTE on non-Nvidia drivers (including nouveau) this commit
simulates a GPU with a warp size of one, returning what is meaningful
for the instruction being emulated:

* anyThreadNV(value) -> value
* allThreadsNV(value) -> value
* allThreadsEqualNV(value) -> true

ballotARB, also known as "uint64_t(activeThreadsNV())", emits

VOTE.ANY Rd, PT, PT;

on nouveau's compiler. This doesn't match exactly to Nvidia's code

VOTE.ALL Rd, PT, PT;

Which is emulated with activeThreadsNV() by this commit. In theory this
shouldn't really matter since .ANY, .ALL and .EQ affect the predicates
(set to PT on those cases) and not the registers.
---
 src/video_core/shader/decode.cpp      |  1 +
 src/video_core/shader/decode/warp.cpp | 55 +++++++++++++++++++++++++++++++++++
 src/video_core/shader/node.h          |  5 ++++
 src/video_core/shader/shader_ir.h     |  1 +
 4 files changed, 62 insertions(+)
 create mode 100644 src/video_core/shader/decode/warp.cpp

(limited to 'src/video_core/shader')

diff --git a/src/video_core/shader/decode.cpp b/src/video_core/shader/decode.cpp
index b547d8323..47a9fd961 100644
--- a/src/video_core/shader/decode.cpp
+++ b/src/video_core/shader/decode.cpp
@@ -176,6 +176,7 @@ u32 ShaderIR::DecodeInstr(NodeBlock& bb, u32 pc) {
         {OpCode::Type::Ffma, &ShaderIR::DecodeFfma},
         {OpCode::Type::Hfma2, &ShaderIR::DecodeHfma2},
         {OpCode::Type::Conversion, &ShaderIR::DecodeConversion},
+        {OpCode::Type::Warp, &ShaderIR::DecodeWarp},
         {OpCode::Type::Memory, &ShaderIR::DecodeMemory},
         {OpCode::Type::Texture, &ShaderIR::DecodeTexture},
         {OpCode::Type::Image, &ShaderIR::DecodeImage},
diff --git a/src/video_core/shader/decode/warp.cpp b/src/video_core/shader/decode/warp.cpp
new file mode 100644
index 000000000..04ca74f46
--- /dev/null
+++ b/src/video_core/shader/decode/warp.cpp
@@ -0,0 +1,55 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "video_core/engines/shader_bytecode.h"
+#include "video_core/shader/node_helper.h"
+#include "video_core/shader/shader_ir.h"
+
+namespace VideoCommon::Shader {
+
+using Tegra::Shader::Instruction;
+using Tegra::Shader::OpCode;
+using Tegra::Shader::Pred;
+using Tegra::Shader::VoteOperation;
+
+namespace {
+OperationCode GetOperationCode(VoteOperation vote_op) {
+    switch (vote_op) {
+    case VoteOperation::All:
+        return OperationCode::VoteAll;
+    case VoteOperation::Any:
+        return OperationCode::VoteAny;
+    case VoteOperation::Eq:
+        return OperationCode::VoteEqual;
+    default:
+        UNREACHABLE_MSG("Invalid vote operation={}", static_cast<u64>(vote_op));
+        return OperationCode::VoteAll;
+    }
+}
+} // Anonymous namespace
+
+u32 ShaderIR::DecodeWarp(NodeBlock& bb, u32 pc) {
+    const Instruction instr = {program_code[pc]};
+    const auto opcode = OpCode::Decode(instr);
+
+    switch (opcode->get().GetId()) {
+    case OpCode::Id::VOTE: {
+        const Node value = GetPredicate(instr.vote.value, instr.vote.negate_value != 0);
+        const Node active = Operation(OperationCode::BallotThread, value);
+        const Node vote = Operation(GetOperationCode(instr.vote.operation), value);
+        SetRegister(bb, instr.gpr0, active);
+        SetPredicate(bb, instr.vote.dest_pred, vote);
+        break;
+    }
+    default:
+        UNIMPLEMENTED_MSG("Unhandled warp instruction: {}", opcode->get().GetName());
+        break;
+    }
+
+    return pc;
+}
+
+} // namespace VideoCommon::Shader
diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h
index 5f0852364..5db9313c4 100644
--- a/src/video_core/shader/node.h
+++ b/src/video_core/shader/node.h
@@ -168,6 +168,11 @@ enum class OperationCode {
     WorkGroupIdY,       /// () -> uint
     WorkGroupIdZ,       /// () -> uint
 
+    BallotThread, /// (bool) -> uint
+    VoteAll,      /// (bool) -> bool
+    VoteAny,      /// (bool) -> bool
+    VoteEqual,    /// (bool) -> bool
+
     Amount,
 };
 
diff --git a/src/video_core/shader/shader_ir.h b/src/video_core/shader/shader_ir.h
index 59a083d90..99d06ff4a 100644
--- a/src/video_core/shader/shader_ir.h
+++ b/src/video_core/shader/shader_ir.h
@@ -167,6 +167,7 @@ private:
     u32 DecodeFfma(NodeBlock& bb, u32 pc);
     u32 DecodeHfma2(NodeBlock& bb, u32 pc);
     u32 DecodeConversion(NodeBlock& bb, u32 pc);
+    u32 DecodeWarp(NodeBlock& bb, u32 pc);
     u32 DecodeMemory(NodeBlock& bb, u32 pc);
     u32 DecodeTexture(NodeBlock& bb, u32 pc);
     u32 DecodeImage(NodeBlock& bb, u32 pc);
-- 
cgit v1.2.3