Skip to content

[Perf] Tunings for SM100 FP8 CUTLASS kernel #18778

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Jun 4, 2025
Merged

[Perf] Tunings for SM100 FP8 CUTLASS kernel #18778

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e4cc99d
Tunings for SM100 FP8 CUTLASS kernel
mgoin May 27, 2025
90a30b6
Merge branch 'main' into sm100-fp8-tuning
mgoin Jun 3, 2025
f96ed4b
Tune small M a bit more
mgoin Jun 3, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 51 additions & 2 deletions csrc/quantization/cutlass_w8a8/c3x/scaled_mm_sm100_fp8_dispatch.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,6 +15,7 @@ using c3x::cutlass_gemm_caller;
template <typename InType, typename OutType,
template <typename, typename, typename> typename Epilogue>
struct sm100_fp8_config_default {
// M in (128, inf)
static_assert(std::is_same<InType, cutlass::float_e4m3_t>());
using KernelSchedule = cutlass::gemm::collective::KernelScheduleAuto;
using EpilogueSchedule = cutlass::epilogue::collective::EpilogueScheduleAuto;
Expand All @@ -25,6 +26,34 @@ struct sm100_fp8_config_default {
KernelSchedule, EpilogueSchedule>;
};

template <typename InType, typename OutType,
template <typename, typename, typename> typename Epilogue>
struct sm100_fp8_config_M128 {
// M in (64, 128]
static_assert(std::is_same<InType, cutlass::float_e4m3_t>());
using KernelSchedule = cutlass::gemm::collective::KernelScheduleAuto;
using EpilogueSchedule = cutlass::epilogue::collective::EpilogueScheduleAuto;
using TileShape = Shape<_128, _128, _64>;
using ClusterShape = Shape<_2, _2, _1>;
using Cutlass3xGemm =
cutlass_3x_gemm_sm100<InType, OutType, Epilogue, TileShape, ClusterShape,
KernelSchedule, EpilogueSchedule>;
};

template <typename InType, typename OutType,
template <typename, typename, typename> typename Epilogue>
struct sm100_fp8_config_M64 {
// M in [1, 64]
static_assert(std::is_same<InType, cutlass::float_e4m3_t>());
using KernelSchedule = cutlass::gemm::collective::KernelScheduleAuto;
using EpilogueSchedule = cutlass::epilogue::collective::EpilogueScheduleAuto;
using TileShape = Shape<_64, _64, _256>;
using ClusterShape = Shape<_1, _8, _1>;
using Cutlass3xGemm =
cutlass_3x_gemm_sm100<InType, OutType, Epilogue, TileShape, ClusterShape,
KernelSchedule, EpilogueSchedule>;
};

template <typename InType, typename OutType,
template <typename, typename, typename> typename Epilogue,
typename... EpilogueArgs>
Expand All @@ -39,8 +68,28 @@ inline void cutlass_gemm_sm100_fp8_dispatch(torch::Tensor& out,
using Cutlass3xGemmDefault =
typename sm100_fp8_config_default<InType, OutType,
Epilogue>::Cutlass3xGemm;
return cutlass_gemm_caller<Cutlass3xGemmDefault>(
out, a, b, std::forward<EpilogueArgs>(args)...);
using Cutlass3xGemmM64 =
typename sm100_fp8_config_M64<InType, OutType, Epilogue>::Cutlass3xGemm;
using Cutlass3xGemmM128 =
typename sm100_fp8_config_M128<InType, OutType, Epilogue>::Cutlass3xGemm;

uint32_t const m = a.size(0);
uint32_t const mp2 =
std::max(static_cast<uint32_t>(64), next_pow_2(m)); // next power of 2

if (mp2 <= 64) {
// m in [1, 64]
return cutlass_gemm_caller<Cutlass3xGemmM64>(
out, a, b, std::forward<EpilogueArgs>(args)...);
} else if (mp2 <= 128) {
// m in (64, 128]
return cutlass_gemm_caller<Cutlass3xGemmM128>(
out, a, b, std::forward<EpilogueArgs>(args)...);
} else {
// m in (128, inf)
return cutlass_gemm_caller<Cutlass3xGemmDefault>(
out, a, b, std::forward<EpilogueArgs>(args)...);
}
}

template <template <typename, typename, typename> typename Epilogue,
Expand Down