[Sparse24] [Deprecation] Remove Sparse24 CT integration and kernels (#36799)

Signed-off-by: Kyle Sayers <kylesayrs@gmail.com>
This commit is contained in:
Kyle Sayers
2026-03-23 16:03:29 -04:00
committed by GitHub
parent fafe76b4af
commit 38364a7e32
17 changed files with 9 additions and 2674 deletions
-10
View File
@@ -285,16 +285,6 @@ void cutlass_scaled_mm_azp(torch::Tensor& out, torch::Tensor const& a,
std::optional<torch::Tensor> const& azp,
std::optional<torch::Tensor> const& bias);
bool cutlass_sparse_scaled_mm_supported(int64_t cuda_device_capability);
void cutlass_scaled_sparse_mm(torch::Tensor& out, torch::Tensor const& a,
torch::Tensor const& b, torch::Tensor const& e,
torch::Tensor const& a_scales,
torch::Tensor const& b_scales,
std::optional<torch::Tensor> const& bias);
std::vector<torch::Tensor> cutlass_sparse_compress(torch::Tensor const& a);
std::tuple<torch::Tensor, torch::Tensor> scaled_fp4_quant_func(
torch::Tensor const& input, torch::Tensor const& input_scale,
bool is_sf_swizzled_layout);