[10/n] Migrate cuda_view and silu_and_mul_per_block_quant kernels to torch stale ABI. (#44334)

This commit is contained in:
Chris Leonard
2026-06-04 20:14:43 -07:00
committed by GitHub
parent 063ce98fb7
commit 56aff0dd15
25 changed files with 186 additions and 163 deletions
@@ -20,7 +20,7 @@
#include "cutlass/util/packed_stride.hpp"
#include "core/math.hpp"
#include "cutlass_extensions/common.hpp"
#include "libtorch_stable/cutlass_extensions/common.hpp"
// clang-format on
namespace vllm::c3x {
@@ -15,7 +15,7 @@
#include "cutlass/gemm/collective/collective_builder.hpp"
#include "core/math.hpp"
#include "cutlass_extensions/common.hpp"
#include "libtorch_stable/cutlass_extensions/common.hpp"
// clang-format on
/*
@@ -1,7 +1,7 @@
#include <torch/csrc/stable/tensor.h>
#include <torch/headeronly/core/ScalarType.h>
#include "cuda_utils.h"
#include "cutlass_extensions/common.hpp"
#include "libtorch_stable/cutlass_extensions/common.hpp"
template <typename Fp8Func, typename Int8Func, typename BlockwiseFunc>
void dispatch_scaled_mm(torch::stable::Tensor& c,
@@ -8,7 +8,7 @@
#include <torch/csrc/stable/ops.h>
#include "cutlass_extensions/epilogue/scaled_mm_epilogues_c3x.hpp"
#include "cutlass_extensions/common.hpp"
#include "libtorch_stable/cutlass_extensions/common.hpp"
#include "get_group_starts.cuh"
using namespace cute;
@@ -23,7 +23,7 @@
#include "cutlass/gemm/kernel/default_gemm_universal_with_visitor.h"
#include "core/math.hpp"
#include "cutlass_extensions/common.hpp"
#include "libtorch_stable/cutlass_extensions/common.hpp"
// clang-format on
using namespace cute;
@@ -4,7 +4,7 @@
#include "libtorch_stable/torch_utils.h"
#include "cutlass_extensions/common.hpp"
#include "libtorch_stable/cutlass_extensions/common.hpp"
void cutlass_scaled_mm_sm75(torch::stable::Tensor& c,
torch::stable::Tensor const& a,