submission 188802

from dataclasses import dataclass
import os

import torch

import cutlass
import cutlass.cute as cute
import cutlass.utils as utils
import cutlass.pipeline as pipeline
from cutlass.pipeline import pipeline_init_arrive, pipeline_init_wait
from cutlass.cute.nvgpu import cpasync, tcgen05
import cutlass.utils.blackwell_helpers as sm100_utils
import cutlass.utils.blockscaled_layout as blockscaled_utils
from cutlass.cute.runtime import make_ptr

from task import input_t, output_t
from torch.utils.cpp_extension import load_inline


@dataclass(frozen=True)
class KernelConfig:
    name: str
    mma_tiler_mn: tuple
    cluster_shape_mn: tuple
    swizzle_size: int
    raster_along_m: bool
    occupancy: int


CONFIGS = {
    "n7168_k16384": KernelConfig(
        name="n7168_k16384",
        mma_tiler_mn=(256, 128),
        cluster_shape_mn=(2, 1),
        swizzle_size=4,
        raster_along_m=True,
        occupancy=1,
    ),
    "n4096_k7168": KernelConfig(
        name="n4096_k7168",
        mma_tiler_mn=(256, 128),
        cluster_shape_mn=(2, 1),
        swizzle_size=1,
        raster_along_m=False,
        occupancy=1,
    ),
    "n7168_k2048": KernelConfig(
        name="n7168_k2048",
        mma_tiler_mn=(128, 192),
        cluster_shape_mn=(1, 1),
        swizzle_size=2,
        raster_along_m=False,
        occupancy=1,
    ),
    "default": KernelConfig(
        name="default",
        mma_tiler_mn=(128, 128),
        cluster_shape_mn=(1, 1),
        swizzle_size=1,
        raster_along_m=True,
        occupancy=1,
    ),
}


ab_dtype = cutlass.Float4E2M1FN
sf_dtype = cutlass.Float8E4M3FN
c_dtype = cutlass.Float16
sf_vec_size = 16

_cutlass_ext = None
_cutlass_src = r"""
#include <torch/extension.h>

#include "cutlass/cutlass.h"

#include "cute/tensor.hpp"
#include "cutlass/detail/sm100_blockscaled_layout.hpp"
#include "cutlass/epilogue/collective/collective_builder.hpp"
#include "cutlass/epilogue/fusion/operations.hpp"
#include "cutlass/gemm/collective/collective_builder.hpp"
#include "cutlass/gemm/device/gemm_universal_adapter.h"
#include "cutlass/gemm/kernel/gemm_universal.hpp"
#include "cutlass/util/packed_stride.hpp"

using ElementA = cutlass::nv_float4_t<cutlass::float_e2m1_t>;
using ElementB = cutlass::nv_float4_t<cutlass::float_e2m1_t>;
using ElementC = cutlass::half_t;
using ElementD = cutlass::half_t;
using ElementAccumulator = float;
using ElementCompute = float;

using LayoutATag = cutlass::layout::RowMajor;
using LayoutBTag = cutlass::layout::ColumnMajor;
using LayoutCTag = cutlass::layout::RowMajor;
using LayoutDTag = cutlass::layout::RowMajor;

constexpr int AlignmentA = 32;
constexpr int AlignmentB = 32;
constexpr int AlignmentC = 128 / cutlass::sizeof_bits<ElementC>::value;
constexpr int AlignmentD = 128 / cutlass::sizeof_bits<ElementD>::value;

using ArchTag = cutlass::arch::Sm100;
using OperatorClass = cutlass::arch::OpClassBlockScaledTensorOp;

using ClusterShape = cute::Shape<cute::_1, cute::_1, cute::_1>;

using FusionOp = cutlass::epilogue::fusion::LinearCombination<
    ElementD,
    ElementCompute,
    ElementC,
    ElementCompute>;

template <class MmaTileShape>
struct GemmDef {
  using CollectiveEpilogue = typename cutlass::epilogue::collective::CollectiveBuilder<
      ArchTag,
      OperatorClass,
      MmaTileShape,
      ClusterShape,
      cutlass::epilogue::collective::EpilogueTileAuto,
      ElementAccumulator,
      ElementAccumulator,
      ElementC,
      LayoutCTag,
      AlignmentC,
      ElementD,
      LayoutDTag,
      AlignmentD,
      cutlass::epilogue::collective::EpilogueScheduleAuto,
      FusionOp>::CollectiveOp;

  using CollectiveMainloop = typename cutlass::gemm::collective::CollectiveBuilder<
      ArchTag,
      OperatorClass,
      ElementA,
      LayoutATag,
      AlignmentA,
      ElementB,
      LayoutBTag,
      AlignmentB,
      ElementAccumulator,
      MmaTileShape,
      ClusterShape,
      cutlass::gemm::collective::StageCountAutoCarveout<
          static_cast<int>(sizeof(typename CollectiveEpilogue::SharedStorage))>,
      cutlass::gemm::collective::KernelScheduleAuto>::CollectiveOp;

  using Kernel = cutlass::gemm::kernel::GemmUniversal<
      cute::Shape<int, int, int, int>,
      CollectiveMainloop,
      CollectiveEpilogue,
      void>;

  using Gemm = cutlass::gemm::device::GemmUniversalAdapter<Kernel>;
};

using Gemm128x128 = GemmDef<cute::Shape<cute::_128, cute::_128, cute::_256>>::Gemm;
using Gemm128x256 = GemmDef<cute::Shape<cute::_128, cute::_256, cute::_256>>::Gemm;
using Gemm256x128 = GemmDef<cute::Shape<cute::_256, cute::_128, cute::_256>>::Gemm;

template <class Gemm>
torch::Tensor run_gemm(
    torch::Tensor a,
    torch::Tensor b,
    torch::Tensor sfa,
    torch::Tensor sfb,
    torch::Tensor c) {
  using StrideA = typename Gemm::GemmKernel::StrideA;
  using StrideB = typename Gemm::GemmKernel::StrideB;
  using StrideC = typename Gemm::GemmKernel::StrideC;
  using StrideD = typename Gemm::GemmKernel::StrideD;
  using LayoutSFA = typename Gemm::GemmKernel::CollectiveMainloop::LayoutSFA;
  using LayoutSFB = typename Gemm::GemmKernel::CollectiveMainloop::LayoutSFB;

  const int64_t m = a.size(0);
  const int64_t k2 = a.size(1);
  const int64_t n = b.size(0);
  const int64_t k = k2 * 2;

  auto stride_a = StrideA{a.stride(0), a.stride(1), a.stride(2)};
  auto stride_b = StrideB{b.stride(0), b.stride(1), b.stride(2)};
  auto stride_c = StrideC{c.stride(0), c.stride(1), c.stride(2)};
  auto stride_d = StrideD{c.stride(0), c.stride(1), c.stride(2)};

  auto layout_sfa = typename Gemm::GemmKernel::CollectiveMainloop::Sm1xxBlkScaledConfig::
      tile_atom_to_shape_SFA(cute::make_shape(m, n, k, 1));
  auto layout_sfb = typename Gemm::GemmKernel::CollectiveMainloop::Sm1xxBlkScaledConfig::
      tile_atom_to_shape_SFB(cute::make_shape(m, n, k, 1));

  typename Gemm::Arguments args{
      cutlass::gemm::GemmUniversalMode::kGemm,
      {int(m), int(n), int(k), 1},
      {reinterpret_cast<ElementA*>(a.data_ptr()),
       stride_a,
       reinterpret_cast<ElementB*>(b.data_ptr()),
       stride_b,
       reinterpret_cast<typename LayoutSFA::Element*>(sfa.data_ptr()),
       layout_sfa,
       reinterpret_cast<typename LayoutSFB::Element*>(sfb.data_ptr()),
       layout_sfb},
      {{ElementCompute(1.0f), ElementCompute(0.0f)},
       reinterpret_cast<ElementC*>(c.data_ptr()),
       stride_c,
       reinterpret_cast<ElementD*>(c.data_ptr()),
       stride_d}};

  args.scheduler.max_swizzle_size = 4;

  size_t workspace_bytes = Gemm::get_workspace_size(args);
  void* workspace_ptr = nullptr;
  if (workspace_bytes > 0) {
    cudaMalloc(&workspace_ptr, workspace_bytes);
  }

  Gemm gemm;
  auto status = gemm.can_implement(args);
  if (status != cutlass::Status::kSuccess) {
    if (workspace_ptr) {
      cudaFree(workspace_ptr);
    }
    return c;
  }

  status = gemm.initialize(args, workspace_ptr);
  if (status != cutlass::Status::kSuccess) {
    if (workspace_ptr) {
      cudaFree(workspace_ptr);
    }
    return c;
  }

  status = gemm.run();
  if (workspace_ptr) {
    cudaFree(workspace_ptr);
  }
  return c;
}

torch::Tensor nvfp4_cutlass_gemm(
    torch::Tensor a,
    torch::Tensor b,
    torch::Tensor sfa,
    torch::Tensor sfb,
    torch::Tensor c) {
  const int64_t n = b.size(0);
  const int64_t k2 = a.size(1);
  const int64_t k = k2 * 2;

  if (n == 4096 && k >= 7168) {
    return run_gemm<Gemm128x128>(a, b, sfa, sfb, c);
  }
  if (n == 7168 && k >= 16384) {
    return run_gemm<Gemm128x256>(a, b, sfa, sfb, c);
  }
  if (n == 7168 && k <= 2048) {
    return run_gemm<Gemm256x128>(a, b, sfa, sfb, c);
  }
  return run_gemm<Gemm128x128>(a, b, sfa, sfb, c);
}

PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
  m.def("nvfp4_cutlass_gemm", &nvfp4_cutlass_gemm);
}
"""


class PersistentNvfp4Gemm:
    def __init__(self, cfg: KernelConfig):
        self.sf_vec_size = sf_vec_size
        self.mma_tiler_mn = cfg.mma_tiler_mn
        self.cluster_shape_mn = cfg.cluster_shape_mn
        self.swizzle_size = cfg.swizzle_size
        self.raster_along_m = cfg.raster_along_m
        self.occupancy = cfg.occupancy

        self.acc_dtype = cutlass.Float32
        self.use_2cta_instrs = self.mma_tiler_mn[0] == 256
        self.cta_group = (
            tcgen05.CtaGroup.TWO if self.use_2cta_instrs else tcgen05.CtaGroup.ONE
        )

        self.epilog_warp_id = (0, 1, 2, 3)
        self.mma_warp_id = 4
        self.tma_warp_id = 5
        self.threads_per_cta = 32 * len(
            (self.mma_warp_id, self.tma_warp_id, *self.epilog_warp_id)
        )

        self.epilog_sync_barrier = pipeline.NamedBarrier(
            barrier_id=1, num_threads=32 * len(self.epilog_warp_id)
        )
        self.tmem_alloc_barrier = pipeline.NamedBarrier(
            barrier_id=2, num_threads=32 * len((self.mma_warp_id, *self.epilog_warp_id))
        )

        self.smem_capacity = utils.get_smem_capacity_in_bytes("sm_100")
        self.num_tmem_alloc_cols = 512

    def _setup_attributes(self, a_tensor, b_tensor, c_tensor):
        self.a_dtype = a_tensor.element_type
        self.b_dtype = b_tensor.element_type
        self.sf_dtype = sf_dtype
        self.c_dtype = c_tensor.element_type
        self.a_major_mode = utils.LayoutEnum.from_tensor(a_tensor).mma_major_mode()
        self.b_major_mode = utils.LayoutEnum.from_tensor(b_tensor).mma_major_mode()
        self.c_layout = utils.LayoutEnum.from_tensor(c_tensor)

        self.mma_inst_shape_mn = (self.mma_tiler_mn[0], self.mma_tiler_mn[1])
        self.mma_inst_shape_mn_sfb = (
            self.mma_inst_shape_mn[0] // (2 if self.use_2cta_instrs else 1),
            cute.round_up(self.mma_inst_shape_mn[1], 128),
        )

        tiled_mma = sm100_utils.make_blockscaled_trivial_tiled_mma(
            self.a_dtype,
            self.a_major_mode,
            self.b_major_mode,
            self.sf_dtype,
            self.sf_vec_size,
            self.cta_group,
            self.mma_inst_shape_mn,
        )

        tiled_mma_sfb = sm100_utils.make_blockscaled_trivial_tiled_mma(
            self.a_dtype,
            self.a_major_mode,
            self.b_major_mode,
            self.sf_dtype,
            self.sf_vec_size,
            tcgen05.CtaGroup.ONE,
            self.mma_inst_shape_mn_sfb,
        )

        mma_inst_shape_k = cute.size(tiled_mma.shape_mnk, mode=[2])
        mma_inst_tile_k = 4
        self.mma_tiler = (
            self.mma_inst_shape_mn[0],
            self.mma_inst_shape_mn[1],
            mma_inst_shape_k * mma_inst_tile_k,
        )
        self.mma_tiler_sfb = (
            self.mma_inst_shape_mn_sfb[0],
            self.mma_inst_shape_mn_sfb[1],
            mma_inst_shape_k * mma_inst_tile_k,
        )

        self.cta_tile_shape_mnk = (
            self.mma_tiler[0] // cute.size(tiled_mma.thr_id.shape),
            self.mma_tiler[1],
            self.mma_tiler[2],
        )
        self.cta_tile_shape_mnk_sfb = (
            self.mma_tiler_sfb[0] // cute.size(tiled_mma.thr_id.shape),
            self.mma_tiler_sfb[1],
            self.mma_tiler_sfb[2],
        )

        self.cluster_layout_vmnk = cute.tiled_divide(
            cute.make_layout((*self.cluster_shape_mn, 1)),
            (tiled_mma.thr_id.shape,),
        )
        self.cluster_layout_sfb_vmnk = cute.tiled_divide(
            cute.make_layout((*self.cluster_shape_mn, 1)),
            (tiled_mma_sfb.thr_id.shape,),
        )

        self.num_mcast_ctas_a = cute.size(self.cluster_layout_vmnk.shape[2])
        self.num_mcast_ctas_b = cute.size(self.cluster_layout_vmnk.shape[1])
        self.num_mcast_ctas_sfb = cute.size(self.cluster_layout_sfb_vmnk.shape[1])
        self.is_a_mcast = self.num_mcast_ctas_a > 1
        self.is_b_mcast = self.num_mcast_ctas_b > 1
        self.is_sfb_mcast = self.num_mcast_ctas_sfb > 1

        self.epi_tile = sm100_utils.compute_epilogue_tile_shape(
            self.cta_tile_shape_mnk,
            self.use_2cta_instrs,
            self.c_layout,
            self.c_dtype,
        )
        self.epi_tile_n = cute.size(self.epi_tile[1])

        self.num_acc_stage, self.num_ab_stage, self.num_c_stage = self._compute_stages(
            tiled_mma,
            self.mma_tiler,
            self.a_dtype,
            self.b_dtype,
            self.epi_tile,
            self.c_dtype,
            self.c_layout,
            self.sf_dtype,
            self.sf_vec_size,
            self.smem_capacity,
            self.occupancy,
        )

        self.a_smem_layout_staged = sm100_utils.make_smem_layout_a(
            tiled_mma, self.mma_tiler, self.a_dtype, self.num_ab_stage
        )
        self.b_smem_layout_staged = sm100_utils.make_smem_layout_b(
            tiled_mma, self.mma_tiler, self.b_dtype, self.num_ab_stage
        )
        self.sfa_smem_layout_staged = blockscaled_utils.make_smem_layout_sfa(
            tiled_mma, self.mma_tiler, self.sf_vec_size, self.num_ab_stage
        )
        self.sfb_smem_layout_staged = blockscaled_utils.make_smem_layout_sfb(
            tiled_mma, self.mma_tiler, self.sf_vec_size, self.num_ab_stage
        )
        self.c_smem_layout_staged = sm100_utils.make_smem_layout_epi(
            self.c_dtype, self.c_layout, self.epi_tile, self.num_c_stage
        )

        self.overlapping_accum = self.num_acc_stage == 1

        sf_atom_mn = 32
        self.num_sfa_tmem_cols = (
            self.cta_tile_shape_mnk[0] // sf_atom_mn
        ) * mma_inst_tile_k
        self.num_sfb_tmem_cols = (
            self.cta_tile_shape_mnk_sfb[1] // sf_atom_mn
        ) * mma_inst_tile_k
        self.num_sf_tmem_cols = self.num_sfa_tmem_cols + self.num_sfb_tmem_cols
        if not self.overlapping_accum:
            self.num_accumulator_tmem_cols = (
                self.cta_tile_shape_mnk[1] * self.num_acc_stage
            )
        else:
            self.num_accumulator_tmem_cols = (
                self.cta_tile_shape_mnk[1] * 2 - self.num_sf_tmem_cols
            )

        self.iter_acc_early_release_in_epilogue = self.num_sf_tmem_cols // self.epi_tile_n

    @cute.jit
    def __call__(
        self,
        a_tensor: cute.Tensor,
        b_tensor: cute.Tensor,
        sfa_tensor: cute.Tensor,
        sfb_tensor: cute.Tensor,
        c_tensor: cute.Tensor,
        max_active_clusters: cutlass.Constexpr,
    ):
        self._setup_attributes(a_tensor, b_tensor, c_tensor)

        sfa_layout = blockscaled_utils.tile_atom_to_shape_SF(
            a_tensor.shape, self.sf_vec_size
        )
        sfa_tensor = cute.make_tensor(sfa_tensor.iterator, sfa_layout)

        sfb_layout = blockscaled_utils.tile_atom_to_shape_SF(
            b_tensor.shape, self.sf_vec_size
        )
        sfb_tensor = cute.make_tensor(sfb_tensor.iterator, sfb_layout)

        tiled_mma = sm100_utils.make_blockscaled_trivial_tiled_mma(
            self.a_dtype,
            self.a_major_mode,
            self.b_major_mode,
            self.sf_dtype,
            self.sf_vec_size,
            self.cta_group,
            self.mma_inst_shape_mn,
        )
        tiled_mma_sfb = sm100_utils.make_blockscaled_trivial_tiled_mma(
            self.a_dtype,
            self.a_major_mode,
            self.b_major_mode,
            self.sf_dtype,
            self.sf_vec_size,
            tcgen05.CtaGroup.ONE,
            self.mma_inst_shape_mn_sfb,
        )

        atom_thr_size = cute.size(tiled_mma.thr_id.shape)

        a_op = sm100_utils.cluster_shape_to_tma_atom_A(
            self.cluster_shape_mn, tiled_mma.thr_id
        )
        a_smem_layout = cute.slice_(self.a_smem_layout_staged, (None, None, None, 0))
        tma_atom_a, tma_tensor_a = cute.nvgpu.make_tiled_tma_atom_A(
            a_op,
            a_tensor,
            a_smem_layout,
            self.mma_tiler,
            tiled_mma,
            self.cluster_layout_vmnk.shape,
        )

        b_op = sm100_utils.cluster_shape_to_tma_atom_B(
            self.cluster_shape_mn, tiled_mma.thr_id
        )
        b_smem_layout = cute.slice_(self.b_smem_layout_staged, (None, None, None, 0))
        tma_atom_b, tma_tensor_b = cute.nvgpu.make_tiled_tma_atom_B(
            b_op,
            b_tensor,
            b_smem_layout,
            self.mma_tiler,
            tiled_mma,
            self.cluster_layout_vmnk.shape,
        )

        sfa_op = sm100_utils.cluster_shape_to_tma_atom_A(
            self.cluster_shape_mn, tiled_mma.thr_id
        )
        sfa_smem_layout = cute.slice_(self.sfa_smem_layout_staged, (None, None, None, 0))
        tma_atom_sfa, tma_tensor_sfa = cute.nvgpu.make_tiled_tma_atom_A(
            sfa_op,
            sfa_tensor,
            sfa_smem_layout,
            self.mma_tiler,
            tiled_mma,
            self.cluster_layout_vmnk.shape,
            internal_type=cutlass.Int16,
        )

        sfb_op = sm100_utils.cluster_shape_to_tma_atom_SFB(
            self.cluster_shape_mn, tiled_mma.thr_id
        )
        sfb_smem_layout = cute.slice_(self.sfb_smem_layout_staged, (None, None, None, 0))
        tma_atom_sfb, tma_tensor_sfb = cute.nvgpu.make_tiled_tma_atom_B(
            sfb_op,
            sfb_tensor,
            sfb_smem_layout,
            self.mma_tiler_sfb,
            tiled_mma_sfb,
            self.cluster_layout_sfb_vmnk.shape,
            internal_type=cutlass.Int16,
        )
        if cutlass.const_expr(self.cta_tile_shape_mnk[1] == 192):
            x = tma_tensor_sfb.stride[0][1]
            y = cute.ceil_div(tma_tensor_sfb.shape[0][1], 4)
            new_shape = (
                (tma_tensor_sfb.shape[0][0], ((2, 2), y)),
                tma_tensor_sfb.shape[1],
                tma_tensor_sfb.shape[2],
            )
            x_times_3 = 3 * x
            new_stride = (
                (tma_tensor_sfb.stride[0][0], ((x, x), x_times_3)),
                tma_tensor_sfb.stride[1],
                tma_tensor_sfb.stride[2],
            )
            tma_tensor_sfb_new_layout = cute.make_layout(
                new_shape, stride=new_stride
            )
            tma_tensor_sfb = cute.make_tensor(
                tma_tensor_sfb.iterator, tma_tensor_sfb_new_layout
            )

        a_copy_size = cute.size_in_bytes(self.a_dtype, a_smem_layout)
        b_copy_size = cute.size_in_bytes(self.b_dtype, b_smem_layout)
        sfa_copy_size = cute.size_in_bytes(self.sf_dtype, sfa_smem_layout)
        sfb_copy_size = cute.size_in_bytes(self.sf_dtype, sfb_smem_layout)
        self.num_tma_load_bytes = (
            a_copy_size + b_copy_size + sfa_copy_size + sfb_copy_size
        ) * atom_thr_size

        epi_smem_layout = cute.slice_(self.c_smem_layout_staged, (None, None, 0))
        tma_atom_c, tma_tensor_c = cpasync.make_tiled_tma_atom(
            cpasync.CopyBulkTensorTileS2GOp(),
            c_tensor,
            epi_smem_layout,
            self.epi_tile,
        )

        self.tile_sched_params, grid = self._compute_grid(
            c_tensor,
            self.cta_tile_shape_mnk,
            self.cluster_shape_mn,
            max_active_clusters,
            self.swizzle_size,
            self.raster_along_m,
        )

        self.buffer_align_bytes = 1024
        self._define_shared_storage()

        self.kernel(
            tiled_mma,
            tiled_mma_sfb,
            tma_atom_a,
            tma_tensor_a,
            tma_atom_b,
            tma_tensor_b,
            tma_atom_sfa,
            tma_tensor_sfa,
            tma_atom_sfb,
            tma_tensor_sfb,
            tma_atom_c,
            tma_tensor_c,
            self.cluster_layout_vmnk,
            self.cluster_layout_sfb_vmnk,
            self.a_smem_layout_staged,
            self.b_smem_layout_staged,
            self.sfa_smem_layout_staged,
            self.sfb_smem_layout_staged,
            self.c_smem_layout_staged,
            self.epi_tile,
            self.tile_sched_params,
        ).launch(
            grid=grid,
            block=[self.threads_per_cta, 1, 1],
            cluster=(*self.cluster_shape_mn, 1),
            min_blocks_per_mp=1,
        )
        return

    def _define_shared_storage(self):
        @cute.struct
        class SharedStorage:
            ab_full_mbar_ptr: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage]
            ab_empty_mbar_ptr: cute.struct.MemRange[cutlass.Int64, self.num_ab_stage]
            acc_full_mbar_ptr: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage]
            acc_empty_mbar_ptr: cute.struct.MemRange[cutlass.Int64, self.num_acc_stage]
            tmem_dealloc_mbar_ptr: cutlass.Int64
            tmem_holding_buf: cutlass.Int32
            sC: cute.struct.Align[
                cute.struct.MemRange[
                    self.c_dtype,
                    cute.cosize(self.c_smem_layout_staged.outer),
                ],
                self.buffer_align_bytes,
            ]
            sA: cute.struct.Align[
                cute.struct.MemRange[
                    self.a_dtype, cute.cosize(self.a_smem_layout_staged.outer)
                ],
                self.buffer_align_bytes,
            ]
            sB: cute.struct.Align[
                cute.struct.MemRange[
                    self.b_dtype, cute.cosize(self.b_smem_layout_staged.outer)
                ],
                self.buffer_align_bytes,
            ]
            sSFA: cute.struct.Align[
                cute.struct.MemRange[
                    self.sf_dtype,
                    cute.cosize(self.sfa_smem_layout_staged),
                ],
                self.buffer_align_bytes,
            ]
            sSFB: cute.struct.Align[
                cute.struct.MemRange[
                    self.sf_dtype,
                    cute.cosize(self.sfb_smem_layout_staged),
                ],
                self.buffer_align_bytes,
            ]

        self.shared_storage = SharedStorage

    @cute.kernel
    def kernel(
        self,
        tiled_mma: cute.TiledMma,
        tiled_mma_sfb: cute.TiledMma,
        tma_atom_a: cute.CopyAtom,
        mA_mkl: cute.Tensor,
        tma_atom_b: cute.CopyAtom,
        mB_nkl: cute.Tensor,
        tma_atom_sfa: cute.CopyAtom,
        mSFA_mkl: cute.Tensor,
        tma_atom_sfb: cute.CopyAtom,
        mSFB_nkl: cute.Tensor,
        tma_atom_c: cute.CopyAtom,
        mC_mnl: cute.Tensor,
        cluster_layout_vmnk: cute.Layout,
        cluster_layout_sfb_vmnk: cute.Layout,
        a_smem_layout_staged: cute.ComposedLayout,
        b_smem_layout_staged: cute.ComposedLayout,
        sfa_smem_layout_staged: cute.Layout,
        sfb_smem_layout_staged: cute.Layout,
        c_smem_layout_staged: cute.ComposedLayout,
        epi_tile: cute.Tile,
        tile_sched_params: utils.PersistentTileSchedulerParams,
    ):
        warp_idx = cute.arch.warp_idx()
        warp_idx = cute.arch.make_warp_uniform(warp_idx)

        if warp_idx == self.tma_warp_id:
            cpasync.prefetch_descriptor(tma_atom_a)
            cpasync.prefetch_descriptor(tma_atom_b)
            cpasync.prefetch_descriptor(tma_atom_sfa)
            cpasync.prefetch_descriptor(tma_atom_sfb)
            cpasync.prefetch_descriptor(tma_atom_c)

        use_2cta_instrs = cute.size(tiled_mma.thr_id.shape) == 2

        bidx, bidy, bidz = cute.arch.block_idx()
        mma_tile_coord_v = bidx % cute.size(tiled_mma.thr_id.shape)
        is_leader_cta = mma_tile_coord_v == 0
        cta_rank_in_cluster = cute.arch.make_warp_uniform(
            cute.arch.block_idx_in_cluster()
        )
        block_in_cluster_coord_vmnk = cluster_layout_vmnk.get_flat_coord(
            cta_rank_in_cluster
        )
        block_in_cluster_coord_sfb_vmnk = cluster_layout_sfb_vmnk.get_flat_coord(
            cta_rank_in_cluster
        )
        tidx, _, _ = cute.arch.thread_idx()

        smem = utils.SmemAllocator()
        storage = smem.allocate(self.shared_storage)

        ab_pipeline_producer_group = pipeline.CooperativeGroup(pipeline.Agent.Thread)
        num_tma_producer = self.num_mcast_ctas_a + self.num_mcast_ctas_b - 1
        ab_pipeline_consumer_group = pipeline.CooperativeGroup(
            pipeline.Agent.Thread, num_tma_producer
        )
        ab_pipeline = pipeline.PipelineTmaUmma.create(
            barrier_storage=storage.ab_full_mbar_ptr.data_ptr(),
            num_stages=self.num_ab_stage,
            producer_group=ab_pipeline_producer_group,
            consumer_group=ab_pipeline_consumer_group,
            tx_count=self.num_tma_load_bytes,
            cta_layout_vmnk=cluster_layout_vmnk,
            defer_sync=True,
        )

        acc_pipeline_producer_group = pipeline.CooperativeGroup(pipeline.Agent.Thread)
        num_acc_consumer_threads = len(self.epilog_warp_id) * (
            2 if use_2cta_instrs else 1
        )
        acc_pipeline_consumer_group = pipeline.CooperativeGroup(
            pipeline.Agent.Thread, num_acc_consumer_threads
        )
        acc_pipeline = pipeline.PipelineUmmaAsync.create(
            barrier_storage=storage.acc_full_mbar_ptr.data_ptr(),
            num_stages=self.num_acc_stage,
            producer_group=acc_pipeline_producer_group,
            consumer_group=acc_pipeline_consumer_group,
            cta_layout_vmnk=cluster_layout_vmnk,
            defer_sync=True,
        )

        tmem = utils.TmemAllocator(
            storage.tmem_holding_buf,
            barrier_for_retrieve=self.tmem_alloc_barrier,
            allocator_warp_id=self.epilog_warp_id[0],
            is_two_cta=use_2cta_instrs,
            two_cta_tmem_dealloc_mbar_ptr=storage.tmem_dealloc_mbar_ptr,
        )

        pipeline_init_arrive(cluster_shape_mn=self.cluster_shape_mn, is_relaxed=True)

        sC = storage.sC.get_tensor(
            c_smem_layout_staged.outer, swizzle=c_smem_layout_staged.inner
        )
        sA = storage.sA.get_tensor(
            a_smem_layout_staged.outer, swizzle=a_smem_layout_staged.inner
        )
        sB = storage.sB.get_tensor(
            b_smem_layout_staged.outer, swizzle=b_smem_layout_staged.inner
        )
        sSFA = storage.sSFA.get_tensor(sfa_smem_layout_staged)
        sSFB = storage.sSFB.get_tensor(sfb_smem_layout_staged)

        a_full_mcast_mask = None
        b_full_mcast_mask = None
        sfa_full_mcast_mask = None
        sfb_full_mcast_mask = None
        if cutlass.const_expr(self.is_a_mcast or self.is_b_mcast or use_2cta_instrs):
            a_full_mcast_mask = cpasync.create_tma_multicast_mask(
                cluster_layout_vmnk, block_in_cluster_coord_vmnk, mcast_mode=2
            )
            b_full_mcast_mask = cpasync.create_tma_multicast_mask(
                cluster_layout_vmnk, block_in_cluster_coord_vmnk, mcast_mode=1
            )
            sfa_full_mcast_mask = cpasync.create_tma_multicast_mask(
                cluster_layout_vmnk, block_in_cluster_coord_vmnk, mcast_mode=2
            )
            sfb_full_mcast_mask = cpasync.create_tma_multicast_mask(
                cluster_layout_sfb_vmnk, block_in_cluster_coord_sfb_vmnk, mcast_mode=1
            )

        gA_mkl = cute.local_tile(
            mA_mkl, cute.slice_(self.mma_tiler, (None, 0, None)), (None, None, None)
        )
        gB_nkl = cute.local_tile(
            mB_nkl, cute.slice_(self.mma_tiler, (0, None, None)), (None, None, None)
        )
        gSFA_mkl = cute.local_tile(
            mSFA_mkl, cute.slice_(self.mma_tiler, (None, 0, None)), (None, None, None)
        )
        gSFB_nkl = cute.local_tile(
            mSFB_nkl,
            cute.slice_(self.mma_tiler_sfb, (0, None, None)),
            (None, None, None),
        )
        gC_mnl = cute.local_tile(
            mC_mnl, cute.slice_(self.mma_tiler, (None, None, 0)), (None, None, None)
        )
        k_tile_cnt = cute.size(gA_mkl, mode=[3])

        thr_mma = tiled_mma.get_slice(mma_tile_coord_v)
        thr_mma_sfb = tiled_mma_sfb.get_slice(mma_tile_coord_v)
        tCgA = thr_mma.partition_A(gA_mkl)
        tCgB = thr_mma.partition_B(gB_nkl)
        tCgSFA = thr_mma.partition_A(gSFA_mkl)
        tCgSFB = thr_mma_sfb.partition_B(gSFB_nkl)
        tCgC = thr_mma.partition_C(gC_mnl)

        a_cta_layout = cute.make_layout(
            cute.slice_(cluster_layout_vmnk, (0, 0, None, 0)).shape
        )
        tAsA, tAgA = cpasync.tma_partition(
            tma_atom_a,
            block_in_cluster_coord_vmnk[2],
            a_cta_layout,
            cute.group_modes(sA, 0, 3),
            cute.group_modes(tCgA, 0, 3),
        )
        b_cta_layout = cute.make_layout(
            cute.slice_(cluster_layout_vmnk, (0, None, 0, 0)).shape
        )
        tBsB, tBgB = cpasync.tma_partition(
            tma_atom_b,
            block_in_cluster_coord_vmnk[1],
            b_cta_layout,
            cute.group_modes(sB, 0, 3),
            cute.group_modes(tCgB, 0, 3),
        )

        sfa_cta_layout = a_cta_layout
        tAsSFA, tAgSFA = cpasync.tma_partition(
            tma_atom_sfa,
            block_in_cluster_coord_vmnk[2],
            sfa_cta_layout,
            cute.group_modes(sSFA, 0, 3),
            cute.group_modes(tCgSFA, 0, 3),
        )
        tAsSFA = cute.filter_zeros(tAsSFA)
        tAgSFA = cute.filter_zeros(tAgSFA)

        sfb_cta_layout = cute.make_layout(
            cute.slice_(cluster_layout_sfb_vmnk, (0, None, 0, 0)).shape
        )
        tBsSFB, tBgSFB = cpasync.tma_partition(
            tma_atom_sfb,
            block_in_cluster_coord_sfb_vmnk[1],
            sfb_cta_layout,
            cute.group_modes(sSFB, 0, 3),
            cute.group_modes(tCgSFB, 0, 3),
        )
        tBsSFB = cute.filter_zeros(tBsSFB)
        tBgSFB = cute.filter_zeros(tBgSFB)

        tCrA = tiled_mma.make_fragment_A(sA)
        tCrB = tiled_mma.make_fragment_B(sB)
        acc_shape = tiled_mma.partition_shape_C(self.mma_tiler[:2])
        if cutlass.const_expr(self.overlapping_accum):
            num_acc_stage_overlapped = 2
            tCtAcc_fake = tiled_mma.make_fragment_C(
                cute.append(acc_shape, num_acc_stage_overlapped)
            )
            tCtAcc_fake = cute.make_tensor(
                tCtAcc_fake.iterator,
                cute.make_layout(
                    tCtAcc_fake.shape,
                    stride=(
                        tCtAcc_fake.stride[0],
                        tCtAcc_fake.stride[1],
                        tCtAcc_fake.stride[2],
                        (256 - self.num_sf_tmem_cols) * tCtAcc_fake.stride[0][1],
                    ),
                ),
            )
        else:
            tCtAcc_fake = tiled_mma.make_fragment_C(
                cute.append(acc_shape, self.num_acc_stage)
            )

        pipeline_init_wait(cluster_shape_mn=self.cluster_shape_mn)

        if warp_idx == self.tma_warp_id:
            tile_sched = utils.StaticPersistentTileScheduler.create(
                tile_sched_params, cute.arch.block_idx(), cute.arch.grid_dim()
            )
            work_tile = tile_sched.initial_work_tile_info()

            ab_producer_state = pipeline.make_pipeline_state(
                pipeline.PipelineUserType.Producer, self.num_ab_stage
            )

            while work_tile.is_valid_tile:
                cur_tile_coord = work_tile.tile_idx
                mma_tile_coord_mnl = (
                    cur_tile_coord[0] // cute.size(tiled_mma.thr_id.shape),
                    cur_tile_coord[1],
                    cur_tile_coord[2],
                )

                tAgA_slice = tAgA[
                    (None, mma_tile_coord_mnl[0], None, mma_tile_coord_mnl[2])
                ]
                tBgB_slice = tBgB[
                    (None, mma_tile_coord_mnl[1], None, mma_tile_coord_mnl[2])
                ]
                tAgSFA_slice = tAgSFA[
                    (None, mma_tile_coord_mnl[0], None, mma_tile_coord_mnl[2])
                ]

                slice_n = mma_tile_coord_mnl[1]
                if cutlass.const_expr(self.cta_tile_shape_mnk[1] == 64):
                    slice_n = mma_tile_coord_mnl[1] // 2
                tBgSFB_slice = tBgSFB[(None, slice_n, None, mma_tile_coord_mnl[2])]

                ab_producer_state.reset_count()
                peek_ab_empty_status = cutlass.Boolean(1)
                if ab_producer_state.count < k_tile_cnt:
                    peek_ab_empty_status = ab_pipeline.producer_try_acquire(
                        ab_producer_state
                    )

                for k_tile in cutlass.range(0, k_tile_cnt, 1, unroll=1):
                    ab_pipeline.producer_acquire(
                        ab_producer_state, peek_ab_empty_status
                    )

                    cute.copy(
                        tma_atom_a,
                        tAgA_slice[(None, ab_producer_state.count)],
                        tAsA[(None, ab_producer_state.index)],
                        tma_bar_ptr=ab_pipeline.producer_get_barrier(
                            ab_producer_state
                        ),
                        mcast_mask=a_full_mcast_mask,
                    )
                    cute.copy(
                        tma_atom_b,
                        tBgB_slice[(None, ab_producer_state.count)],
                        tBsB[(None, ab_producer_state.index)],
                        tma_bar_ptr=ab_pipeline.producer_get_barrier(
                            ab_producer_state
                        ),
                        mcast_mask=b_full_mcast_mask,
                    )
                    cute.copy(
                        tma_atom_sfa,
                        tAgSFA_slice[(None, ab_producer_state.count)],
                        tAsSFA[(None, ab_producer_state.index)],
                        tma_bar_ptr=ab_pipeline.producer_get_barrier(
                            ab_producer_state
                        ),
                        mcast_mask=sfa_full_mcast_mask,
                    )
                    cute.copy(
                        tma_atom_sfb,
                        tBgSFB_slice[(None, ab_producer_state.count)],
                        tBsSFB[(None, ab_producer_state.index)],
                        tma_bar_ptr=ab_pipeline.producer_get_barrier(
                            ab_producer_state
                        ),
                        mcast_mask=sfb_full_mcast_mask,
                    )

                    ab_producer_state.advance()
                    peek_ab_empty_status = cutlass.Boolean(1)
                    if ab_producer_state.count < k_tile_cnt:
                        peek_ab_empty_status = ab_pipeline.producer_try_acquire(
                            ab_producer_state
                        )

                tile_sched.advance_to_next_work()
                work_tile = tile_sched.get_current_work()

            ab_pipeline.producer_tail(ab_producer_state)

        if warp_idx == self.mma_warp_id:
            tmem.wait_for_alloc()

            acc_tmem_ptr = tmem.retrieve_ptr(self.acc_dtype)
            tCtAcc_base = cute.make_tensor(acc_tmem_ptr, tCtAcc_fake.layout)

            sfa_tmem_ptr = cute.recast_ptr(
                acc_tmem_ptr + self.num_accumulator_tmem_cols, dtype=self.sf_dtype
            )
            tCtSFA_layout = blockscaled_utils.make_tmem_layout_sfa(
                tiled_mma,
                self.mma_tiler,
                self.sf_vec_size,
                cute.slice_(sfa_smem_layout_staged, (None, None, None, 0)),
            )
            tCtSFA = cute.make_tensor(sfa_tmem_ptr, tCtSFA_layout)

            sfb_tmem_ptr = cute.recast_ptr(
                acc_tmem_ptr
                + self.num_accumulator_tmem_cols
                + self.num_sfa_tmem_cols,
                dtype=self.sf_dtype,
            )
            tCtSFB_layout = blockscaled_utils.make_tmem_layout_sfb(
                tiled_mma,
                self.mma_tiler,
                self.sf_vec_size,
                cute.slice_(sfb_smem_layout_staged, (None, None, None, 0)),
            )
            tCtSFB = cute.make_tensor(sfb_tmem_ptr, tCtSFB_layout)

            (
                tiled_copy_s2t_sfa,
                tCsSFA_compact_s2t,
                tCtSFA_compact_s2t,
            ) = self.mainloop_s2t_copy_and_partition(sSFA, tCtSFA)
            (
                tiled_copy_s2t_sfb,
                tCsSFB_compact_s2t,
                tCtSFB_compact_s2t,
            ) = self.mainloop_s2t_copy_and_partition(sSFB, tCtSFB)

            tile_sched = utils.StaticPersistentTileScheduler.create(
                tile_sched_params, cute.arch.block_idx(), cute.arch.grid_dim()
            )
            work_tile = tile_sched.initial_work_tile_info()

            ab_consumer_state = pipeline.make_pipeline_state(
                pipeline.PipelineUserType.Consumer, self.num_ab_stage
            )
            acc_producer_state = pipeline.make_pipeline_state(
                pipeline.PipelineUserType.Producer, self.num_acc_stage
            )

            while work_tile.is_valid_tile:
                cur_tile_coord = work_tile.tile_idx
                mma_tile_coord_mnl = (
                    cur_tile_coord[0] // cute.size(tiled_mma.thr_id.shape),
                    cur_tile_coord[1],
                    cur_tile_coord[2],
                )

                if cutlass.const_expr(self.overlapping_accum):
                    acc_stage_index = acc_producer_state.phase ^ 1
                else:
                    acc_stage_index = acc_producer_state.index

                tCtAcc = tCtAcc_base[(None, None, None, acc_stage_index)]

                ab_consumer_state.reset_count()
                acc_producer_state.reset_count()

                peek_ab_full_status = cutlass.Boolean(1)
                if ab_consumer_state.count < k_tile_cnt and is_leader_cta:
                    peek_ab_full_status = ab_pipeline.consumer_try_wait(
                        ab_consumer_state
                    )

                if is_leader_cta:
                    acc_pipeline.producer_acquire(acc_producer_state)

                tCtSFB_mma = tCtSFB
                if cutlass.const_expr(self.cta_tile_shape_mnk[1] == 192):
                    offset = (
                        cutlass.Int32(2)
                        if mma_tile_coord_mnl[1] % 2 == 1
                        else cutlass.Int32(0)
                    )
                    shifted_ptr = cute.recast_ptr(
                        acc_tmem_ptr
                        + self.num_accumulator_tmem_cols
                        + self.num_sfa_tmem_cols
                        + offset,
                        dtype=self.sf_dtype,
                    )
                    tCtSFB_mma = cute.make_tensor(shifted_ptr, tCtSFB_layout)
                elif cutlass.const_expr(self.cta_tile_shape_mnk[1] == 64):
                    offset = cutlass.Int32((mma_tile_coord_mnl[1] % 2) * 2)
                    shifted_ptr = cute.recast_ptr(
                        acc_tmem_ptr
                        + self.num_accumulator_tmem_cols
                        + self.num_sfa_tmem_cols
                        + offset,
                        dtype=self.sf_dtype,
                    )
                    tCtSFB_mma = cute.make_tensor(shifted_ptr, tCtSFB_layout)

                tiled_mma.set(tcgen05.Field.ACCUMULATE, False)

                for k_tile in range(k_tile_cnt):
                    if is_leader_cta:
                        ab_pipeline.consumer_wait(
                            ab_consumer_state, peek_ab_full_status
                        )

                        s2t_stage_coord = (
                            None,
                            None,
                            None,
                            None,
                            ab_consumer_state.index,
                        )
                        tCsSFA_compact_s2t_staged = tCsSFA_compact_s2t[s2t_stage_coord]
                        tCsSFB_compact_s2t_staged = tCsSFB_compact_s2t[s2t_stage_coord]
                        cute.copy(
                            tiled_copy_s2t_sfa,
                            tCsSFA_compact_s2t_staged,
                            tCtSFA_compact_s2t,
                        )
                        cute.copy(
                            tiled_copy_s2t_sfb,
                            tCsSFB_compact_s2t_staged,
                            tCtSFB_compact_s2t,
                        )

                        num_kblocks = cute.size(tCrA, mode=[2])
                        for kblock_idx in cutlass.range(num_kblocks, unroll_full=True):
                            kblock_coord = (
                                None,
                                None,
                                kblock_idx,
                                ab_consumer_state.index,
                            )

                            sf_kblock_coord = (None, None, kblock_idx)
                            tiled_mma.set(
                                tcgen05.Field.SFA,
                                tCtSFA[sf_kblock_coord].iterator,
                            )
                            tiled_mma.set(
                                tcgen05.Field.SFB,
                                tCtSFB_mma[sf_kblock_coord].iterator,
                            )

                            cute.gemm(
                                tiled_mma,
                                tCtAcc,
                                tCrA[kblock_coord],
                                tCrB[kblock_coord],
                                tCtAcc,
                            )

                            tiled_mma.set(tcgen05.Field.ACCUMULATE, True)

                        ab_pipeline.consumer_release(ab_consumer_state)

                    ab_consumer_state.advance()
                    peek_ab_full_status = cutlass.Boolean(1)
                    if ab_consumer_state.count < k_tile_cnt:
                        if is_leader_cta:
                            peek_ab_full_status = ab_pipeline.consumer_try_wait(
                                ab_consumer_state
                            )

                if is_leader_cta:
                    acc_pipeline.producer_commit(acc_producer_state)
                acc_producer_state.advance()

                tile_sched.advance_to_next_work()
                work_tile = tile_sched.get_current_work()

            acc_pipeline.producer_tail(acc_producer_state)

        if warp_idx < self.mma_warp_id:
            tmem.allocate(self.num_tmem_alloc_cols)
            tmem.wait_for_alloc()

            acc_tmem_ptr = tmem.retrieve_ptr(self.acc_dtype)
            tCtAcc_base = cute.make_tensor(acc_tmem_ptr, tCtAcc_fake.layout)

            epi_tidx = tidx
            (
                tiled_copy_t2r,
                tTR_tAcc_base,
                tTR_rAcc,
            ) = self.epilog_tmem_copy_and_partition(
                epi_tidx, tCtAcc_base, tCgC, epi_tile, use_2cta_instrs
            )

            tTR_rC = cute.make_rmem_tensor(tTR_rAcc.shape, self.c_dtype)
            tiled_copy_r2s, tRS_rC, tRS_sC = self.epilog_smem_copy_and_partition(
                tiled_copy_t2r, tTR_rC, epi_tidx, sC
            )
            (
                tma_atom_c,
                bSG_sC,
                bSG_gC_partitioned,
            ) = self.epilog_gmem_copy_and_partition(
                epi_tidx, tma_atom_c, tCgC, epi_tile, sC
            )

            tile_sched = utils.StaticPersistentTileScheduler.create(
                tile_sched_params, cute.arch.block_idx(), cute.arch.grid_dim()
            )
            work_tile = tile_sched.initial_work_tile_info()

            acc_consumer_state = pipeline.make_pipeline_state(
                pipeline.PipelineUserType.Consumer, self.num_acc_stage
            )

            c_producer_group = pipeline.CooperativeGroup(
                pipeline.Agent.Thread,
                32 * len(self.epilog_warp_id),
            )
            c_pipeline = pipeline.PipelineTmaStore.create(
                num_stages=self.num_c_stage,
                producer_group=c_producer_group,
            )

            while work_tile.is_valid_tile:
                cur_tile_coord = work_tile.tile_idx
                mma_tile_coord_mnl = (
                    cur_tile_coord[0] // cute.size(tiled_mma.thr_id.shape),
                    cur_tile_coord[1],
                    cur_tile_coord[2],
                )

                bSG_gC = bSG_gC_partitioned[
                    (
                        None,
                        None,
                        None,
                        *mma_tile_coord_mnl,
                    )
                ]

                if cutlass.const_expr(self.overlapping_accum):
                    acc_stage_index = acc_consumer_state.phase
                    reverse_subtile = (
                        cutlass.Boolean(True)
                        if acc_stage_index == 0
                        else cutlass.Boolean(False)
                    )
                else:
                    acc_stage_index = acc_consumer_state.index

                tTR_tAcc = tTR_tAcc_base[
                    (None, None, None, None, None, acc_stage_index)
                ]

                acc_pipeline.consumer_wait(acc_consumer_state)

                tTR_tAcc = cute.group_modes(tTR_tAcc, 3, cute.rank(tTR_tAcc))
                bSG_gC = cute.group_modes(bSG_gC, 1, cute.rank(bSG_gC))

                subtile_cnt = cute.size(tTR_tAcc.shape, mode=[3])
                num_prev_subtiles = tile_sched.num_tiles_executed * subtile_cnt
                for subtile_idx in cutlass.range(subtile_cnt):
                    real_subtile_idx = subtile_idx
                    if cutlass.const_expr(self.overlapping_accum):
                        if reverse_subtile:
                            real_subtile_idx = (
                                self.cta_tile_shape_mnk[1] // self.epi_tile_n
                                - 1
                                - subtile_idx
                            )
                    tTR_tAcc_mn = tTR_tAcc[(None, None, None, real_subtile_idx)]
                    cute.copy(tiled_copy_t2r, tTR_tAcc_mn, tTR_rAcc)

                    if cutlass.const_expr(self.overlapping_accum):
                        if subtile_idx == self.iter_acc_early_release_in_epilogue:
                            cute.arch.fence_view_async_tmem_load()
                            with cute.arch.elect_one():
                                acc_pipeline.consumer_release(acc_consumer_state)
                            acc_consumer_state.advance()

                    acc_vec = tiled_copy_r2s.retile(tTR_rAcc).load()
                    acc_vec = acc_vec.to(self.c_dtype)
                    tRS_rC.store(acc_vec)

                    c_buffer = (
                        num_prev_subtiles + real_subtile_idx
                    ) % self.num_c_stage
                    cute.copy(
                        tiled_copy_r2s,
                        tRS_rC,
                        tRS_sC[(None, None, None, c_buffer)],
                    )
                    cute.arch.fence_proxy(
                        cute.arch.ProxyKind.async_shared,
                        space=cute.arch.SharedSpace.shared_cta,
                    )
                    self.epilog_sync_barrier.arrive_and_wait()

                    if warp_idx == self.epilog_warp_id[0]:
                        cute.copy(
                            tma_atom_c,
                            bSG_sC[(None, c_buffer)],
                            bSG_gC[(None, real_subtile_idx)],
                        )
                        c_pipeline.producer_commit()
                        c_pipeline.producer_acquire()
                    self.epilog_sync_barrier.arrive_and_wait()

                if cutlass.const_expr(not self.overlapping_accum):
                    with cute.arch.elect_one():
                        acc_pipeline.consumer_release(acc_consumer_state)
                    acc_consumer_state.advance()

                tile_sched.advance_to_next_work()
                work_tile = tile_sched.get_current_work()

            tmem.relinquish_alloc_permit()
            self.epilog_sync_barrier.arrive_and_wait()
            tmem.free(acc_tmem_ptr)
            c_pipeline.producer_tail()

    def mainloop_s2t_copy_and_partition(self, sSF: cute.Tensor, tSF: cute.Tensor):
        tCsSF_compact = cute.filter_zeros(sSF)
        tCtSF_compact = cute.filter_zeros(tSF)

        copy_atom_s2t = cute.make_copy_atom(
            tcgen05.Cp4x32x128bOp(self.cta_group), self.sf_dtype
        )
        tiled_copy_s2t = tcgen05.make_s2t_copy(copy_atom_s2t, tCtSF_compact)
        thr_copy_s2t = tiled_copy_s2t.get_slice(0)

        tCsSF_compact_s2t_ = thr_copy_s2t.partition_S(tCsSF_compact)
        tCsSF_compact_s2t = tcgen05.get_s2t_smem_desc_tensor(
            tiled_copy_s2t, tCsSF_compact_s2t_
        )
        tCtSF_compact_s2t = thr_copy_s2t.partition_D(tCtSF_compact)

        return tiled_copy_s2t, tCsSF_compact_s2t, tCtSF_compact_s2t

    def epilog_tmem_copy_and_partition(
        self,
        tidx: cutlass.Int32,
        tAcc: cute.Tensor,
        gC_mnl: cute.Tensor,
        epi_tile: cute.Tile,
        use_2cta_instrs,
    ):
        copy_atom_t2r = sm100_utils.get_tmem_load_op(
            self.cta_tile_shape_mnk,
            self.c_layout,
            self.c_dtype,
            self.acc_dtype,
            epi_tile,
            use_2cta_instrs,
        )
        tAcc_epi = cute.flat_divide(tAcc[((None, None), 0, 0, None)], epi_tile)
        tiled_copy_t2r = tcgen05.make_tmem_copy(
            copy_atom_t2r, tAcc_epi[(None, None, 0, 0, 0)]
        )

        thr_copy_t2r = tiled_copy_t2r.get_slice(tidx)
        tTR_tAcc = thr_copy_t2r.partition_S(tAcc_epi)

        gC_mnl_epi = cute.flat_divide(
            gC_mnl[((None, None), 0, 0, None, None, None)], epi_tile
        )
        tTR_gC = thr_copy_t2r.partition_D(gC_mnl_epi)
        tTR_rAcc = cute.make_rmem_tensor(
            tTR_gC[(None, None, None, 0, 0, 0, 0, 0)].shape, self.acc_dtype
        )
        return tiled_copy_t2r, tTR_tAcc, tTR_rAcc

    def epilog_smem_copy_and_partition(
        self,
        tiled_copy_t2r: cute.TiledCopy,
        tTR_rC: cute.Tensor,
        tidx: cutlass.Int32,
        sC: cute.Tensor,
    ):
        copy_atom_r2s = sm100_utils.get_smem_store_op(
            self.c_layout, self.c_dtype, self.acc_dtype, tiled_copy_t2r
        )
        tiled_copy_r2s = cute.make_tiled_copy_D(copy_atom_r2s, tiled_copy_t2r)
        thr_copy_r2s = tiled_copy_r2s.get_slice(tidx)
        tRS_sC = thr_copy_r2s.partition_D(sC)
        tRS_rC = tiled_copy_r2s.retile(tTR_rC)
        return tiled_copy_r2s, tRS_rC, tRS_sC

    def epilog_gmem_copy_and_partition(
        self,
        tidx: cutlass.Int32,
        atom: cute.CopyAtom,
        gC_mnl: cute.Tensor,
        epi_tile: cute.Tile,
        sC: cute.Tensor,
    ):
        gC_epi = cute.flat_divide(gC_mnl[((None, None), 0, 0, None, None, None)], epi_tile)

        tma_atom_c = atom
        sC_for_tma_partition = cute.group_modes(sC, 0, 2)
        gC_for_tma_partition = cute.group_modes(gC_epi, 0, 2)
        bSG_sC, bSG_gC = cpasync.tma_partition(
            tma_atom_c,
            0,
            cute.make_layout(1),
            sC_for_tma_partition,
            gC_for_tma_partition,
        )
        return tma_atom_c, bSG_sC, bSG_gC

    @staticmethod
    def _compute_stages(
        tiled_mma: cute.TiledMma,
        mma_tiler_mnk,
        a_dtype,
        b_dtype,
        epi_tile,
        c_dtype,
        c_layout,
        sf_dtype,
        sf_vec_size,
        smem_capacity,
        occupancy,
    ):
        num_acc_stage = 1 if mma_tiler_mnk[1] == 256 else 2
        num_c_stage = 2

        a_smem_layout_stage_one = sm100_utils.make_smem_layout_a(
            tiled_mma, mma_tiler_mnk, a_dtype, 1
        )
        b_smem_layout_staged_one = sm100_utils.make_smem_layout_b(
            tiled_mma, mma_tiler_mnk, b_dtype, 1
        )
        sfa_smem_layout_staged_one = blockscaled_utils.make_smem_layout_sfa(
            tiled_mma, mma_tiler_mnk, sf_vec_size, 1
        )
        sfb_smem_layout_staged_one = blockscaled_utils.make_smem_layout_sfb(
            tiled_mma, mma_tiler_mnk, sf_vec_size, 1
        )

        c_smem_layout_staged_one = sm100_utils.make_smem_layout_epi(
            c_dtype, c_layout, epi_tile, 1
        )

        ab_bytes_per_stage = (
            cute.size_in_bytes(a_dtype, a_smem_layout_stage_one)
            + cute.size_in_bytes(b_dtype, b_smem_layout_staged_one)
            + cute.size_in_bytes(sf_dtype, sfa_smem_layout_staged_one)
            + cute.size_in_bytes(sf_dtype, sfb_smem_layout_staged_one)
        )
        mbar_helpers_bytes = 1024
        c_bytes_per_stage = cute.size_in_bytes(c_dtype, c_smem_layout_staged_one)
        c_bytes = c_bytes_per_stage * num_c_stage

        num_ab_stage = (
            smem_capacity // occupancy - (mbar_helpers_bytes + c_bytes)
        ) // ab_bytes_per_stage

        num_c_stage += (
            smem_capacity
            - occupancy * ab_bytes_per_stage * num_ab_stage
            - occupancy * (mbar_helpers_bytes + c_bytes)
        ) // (occupancy * c_bytes_per_stage)

        return num_acc_stage, num_ab_stage, num_c_stage

    @staticmethod
    def _compute_grid(
        c: cute.Tensor,
        cta_tile_shape_mnk,
        cluster_shape_mn,
        max_active_clusters: cutlass.Constexpr,
        swizzle_size: int,
        raster_along_m: bool,
    ):
        problem_shape_ntile_mnl = (
            cute.ceil_div(c.shape[0], cta_tile_shape_mnk[0]),
            cute.ceil_div(c.shape[1], cta_tile_shape_mnk[1]),
            c.shape[2],
        )
        tile_sched_params = utils.PersistentTileSchedulerParams(
            problem_shape_ntile_mnl,
            (cluster_shape_mn[0], cluster_shape_mn[1], 1),
            swizzle_size=swizzle_size,
            raster_along_m=raster_along_m,
        )
        grid = utils.StaticPersistentTileScheduler.get_grid_shape(
            tile_sched_params, max_active_clusters
        )
        return tile_sched_params, grid


def select_config(m: int, n: int, k: int) -> KernelConfig:
    if n == 7168 and k >= 16384:
        return CONFIGS["n7168_k16384"]
    if n == 4096 and k >= 7168:
        return CONFIGS["n4096_k7168"]
    if n == 7168 and k <= 2048:
        return CONFIGS["n7168_k2048"]
    return CONFIGS["default"]




_compiled_kernel_cache = {}


def _get_cutlass_ext():
    global _cutlass_ext
    if _cutlass_ext is not None:
        return _cutlass_ext
    os.environ.setdefault("TORCH_CUDA_ARCH_LIST", "10.0a")
    cuda_lib = "/usr/local/cuda-13.0/lib64"
    ld_path = os.environ.get("LD_LIBRARY_PATH", "")
    if cuda_lib not in ld_path.split(":"):
        os.environ["LD_LIBRARY_PATH"] = (
            f"{cuda_lib}:{ld_path}" if ld_path else cuda_lib
        )
    build_root = os.path.join(os.path.dirname(__file__), ".cutlass_ext_build")
    _cutlass_ext = load_inline(
        name="nvfp4_cutlass_ext",
        cpp_sources="",
        cuda_sources=_cutlass_src,
        extra_include_paths=[
            "/usr/local/cutlass/include",
            "/usr/local/cutlass/tools/util/include",
        ],
        extra_cuda_cflags=[
            "-O3",
            "-std=c++17",
            "-gencode=arch=compute_100a,code=sm_100a",
        ],
        extra_cflags=["-O3", "-std=c++17"],
        with_cuda=True,
        build_directory=build_root,
        verbose=False,
    )
    return _cutlass_ext


def compile_kernel(cfg: KernelConfig):
    compiled = _compiled_kernel_cache.get(cfg.name)
    if compiled is not None:
        return compiled

    a_ptr = make_ptr(ab_dtype, 0, cute.AddressSpace.gmem, assumed_align=16)
    b_ptr = make_ptr(ab_dtype, 0, cute.AddressSpace.gmem, assumed_align=16)
    c_ptr = make_ptr(c_dtype, 0, cute.AddressSpace.gmem, assumed_align=16)
    sfa_ptr = make_ptr(sf_dtype, 0, cute.AddressSpace.gmem, assumed_align=32)
    sfb_ptr = make_ptr(sf_dtype, 0, cute.AddressSpace.gmem, assumed_align=32)

    gemm = PersistentNvfp4Gemm(cfg)

    hardware_info = utils.HardwareInfo()
    max_active = hardware_info.get_max_active_clusters(
        cfg.cluster_shape_mn[0] * cfg.cluster_shape_mn[1]
    )

    @cute.jit
    def wrapper(
        a_ptr: cute.Pointer,
        b_ptr: cute.Pointer,
        sfa_ptr: cute.Pointer,
        sfb_ptr: cute.Pointer,
        c_ptr: cute.Pointer,
        problem_size: tuple,
    ):
        m, n, k, l = problem_size
        a_tensor = cute.make_tensor(
            a_ptr,
            cute.make_layout(
                (m, cute.assume(k, 32), l),
                stride=(cute.assume(k, 32), 1, cute.assume(m * k, 32)),
            ),
        )
        b_tensor = cute.make_tensor(
            b_ptr,
            cute.make_layout(
                (n, cute.assume(k, 32), l),
                stride=(cute.assume(k, 32), 1, cute.assume(n * k, 32)),
            ),
        )
        c_tensor = cute.make_tensor(
            c_ptr,
            cute.make_layout((cute.assume(m, 32), n, l), stride=(n, 1, m * n)),
        )
        sfa_layout = blockscaled_utils.tile_atom_to_shape_SF(
            a_tensor.shape, sf_vec_size
        )
        sfa_tensor = cute.make_tensor(sfa_ptr, sfa_layout)
        sfb_layout = blockscaled_utils.tile_atom_to_shape_SF(
            b_tensor.shape, sf_vec_size
        )
        sfb_tensor = cute.make_tensor(sfb_ptr, sfb_layout)
        gemm(
            a_tensor,
            b_tensor,
            sfa_tensor,
            sfb_tensor,
            c_tensor,
            max_active,
        )
        return

    compiled = cute.compile(
        wrapper,
        a_ptr,
        b_ptr,
        sfa_ptr,
        sfb_ptr,
        c_ptr,
        (0, 0, 0, 0),
    )
    _compiled_kernel_cache[cfg.name] = compiled
    return compiled


def custom_kernel(data: input_t) -> output_t:
    a = data[0]
    b = data[1]
    sfa_permuted = data[4]
    sfb_permuted = data[5]
    c = data[6]

    m, k, l = a.shape
    n, _, _ = b.shape
    k = k * 2

    use_cutlass = os.getenv("NVFP4_USE_CUTLASS", "0") == "1"
    if use_cutlass:
        try:
            ext = _get_cutlass_ext()
            ext.nvfp4_cutlass_gemm(a, b, sfa_permuted, sfb_permuted, c)
            return c
        except Exception:
            pass

    cfg = select_config(m, n, k)
    try:
        compiled_func = compile_kernel(cfg)
    except Exception:
        cfg = select_config(m, n, k)
        compiled_func = compile_kernel(cfg)

    a_ptr = make_ptr(ab_dtype, a.data_ptr(), cute.AddressSpace.gmem, assumed_align=16)
    b_ptr = make_ptr(ab_dtype, b.data_ptr(), cute.AddressSpace.gmem, assumed_align=16)
    c_ptr = make_ptr(c_dtype, c.data_ptr(), cute.AddressSpace.gmem, assumed_align=16)
    sfa_ptr = make_ptr(
        sf_dtype, sfa_permuted.data_ptr(), cute.AddressSpace.gmem, assumed_align=32
    )
    sfb_ptr = make_ptr(
        sf_dtype, sfb_permuted.data_ptr(), cute.AddressSpace.gmem, assumed_align=32
    )

    compiled_func(a_ptr, b_ptr, sfa_ptr, sfb_ptr, c_ptr, (m, n, k, l))
    return c