From db2f547434cd989bc76f3fc579be692e6756b006 Mon Sep 17 00:00:00 2001 From: Nguyen Dac Nam Date: Thu, 27 Feb 2020 23:14:25 +0700 Subject: shader: FMUL switch to using LUT (#3441) * shader: add FmulPostFactor LUT table * shader: FMUL apply LUT * Update src/video_core/engines/shader_bytecode.h Co-Authored-By: Mat M. * nit: mistype * clang-format & add missing import * shader: remove post factor LUT. * shader: move post factor LUT to function and fix incorrect order. * clang-format * shader: FMUL: add static to post factor LUT * nit: typo Co-authored-by: Mat M. --- src/video_core/shader/decode/arithmetic.cpp | 33 ++++++++++++----------------- 1 file changed, 14 insertions(+), 19 deletions(-) (limited to 'src/video_core/shader/decode') diff --git a/src/video_core/shader/decode/arithmetic.cpp b/src/video_core/shader/decode/arithmetic.cpp index 90240c765..478394682 100644 --- a/src/video_core/shader/decode/arithmetic.cpp +++ b/src/video_core/shader/decode/arithmetic.cpp @@ -53,29 +53,24 @@ u32 ShaderIR::DecodeArithmetic(NodeBlock& bb, u32 pc) { op_b = GetOperandAbsNegFloat(op_b, false, instr.fmul.negate_b); - // TODO(Rodrigo): Should precise be used when there's a postfactor? - Node value = Operation(OperationCode::FMul, PRECISE, op_a, op_b); + static constexpr std::array FmulPostFactor = { + 1.000f, // None + 0.500f, // Divide 2 + 0.250f, // Divide 4 + 0.125f, // Divide 8 + 8.000f, // Mul 8 + 4.000f, // Mul 4 + 2.000f, // Mul 2 + }; if (instr.fmul.postfactor != 0) { - auto postfactor = static_cast(instr.fmul.postfactor); - - // Postfactor encoded as 3-bit 1's complement in instruction, interpreted with below - // logic. - if (postfactor >= 4) { - postfactor = 7 - postfactor; - } else { - postfactor = 0 - postfactor; - } - - if (postfactor > 0) { - value = Operation(OperationCode::FMul, NO_PRECISE, value, - Immediate(static_cast(1 << postfactor))); - } else { - value = Operation(OperationCode::FDiv, NO_PRECISE, value, - Immediate(static_cast(1 << -postfactor))); - } + op_a = Operation(OperationCode::FMul, NO_PRECISE, op_a, + Immediate(FmulPostFactor[instr.fmul.postfactor])); } + // TODO(Rodrigo): Should precise be used when there's a postfactor? + Node value = Operation(OperationCode::FMul, PRECISE, op_a, op_b); + value = GetSaturatedFloat(value, instr.alu.saturate_d); SetInternalFlagsFromFloat(bb, value, instr.generates_cc); -- cgit v1.2.3