CUTLASS 2.6 (#298)

CUTLASS 2.6
This commit is contained in:
Manish Gupta
2021-07-23 00:40:53 -04:00
committed by GitHub
parent 6c29fe20ba
commit e5d51840e8
308 changed files with 32408 additions and 4722 deletions
@@ -121,7 +121,7 @@ public:
// warp-level, arch-level (instruction), math operator
using WarpMmaOperator = typename GemmKernel::Mma::Policy::Operator;
using ArchMmaOperator = typename WarpMmaOperator::ArchMmaOperator;
using MathOperator = typename ArchMmaOperator::Operator;
using MathOperator = typename WarpMmaOperator::MathOperator;
// Operator class and arch tag extract bottom-up
// set it for top-level gemm device-level template
@@ -161,13 +161,11 @@ public:
using TensorRefC = TensorRef<ElementC const, LayoutC>;
using TensorRefD = TensorRef<ElementC, LayoutC>;
using ElementAccumulator = typename GemmKernel::Mma::Policy::Operator::ElementC;
static int const kStages = GemmKernel::Mma::kStages;
using EpilogueOutputOp = typename GemmKernel::EpilogueOutputOp;
using ElementAccumulator = typename EpilogueOutputOp::ElementAccumulator;
using ThreadblockSwizzle = typename GemmKernel::ThreadblockSwizzle;
using Operator = typename GemmKernel::Operator;
using UnderlyingOperator = GemmUniversalBase<GemmKernel>;
using Arguments = typename UnderlyingOperator::Arguments;