Skip to content

CUDA: add fp kernel for larger batch size MoE #16512

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

Open
wants to merge 8 commits into
base: master
Choose a base branch
from
+418 −68
Open

CUDA: add fp kernel for larger batch size MoE #16512

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
8ee4066
CUDA: kernel for larger batch sizes for MoE
am17an Sep 11, 2025
996bed0
WIP
am17an Oct 10, 2025
4a2388e
WIP
am17an Oct 10, 2025
3120ddf
WIP
am17an Oct 10, 2025
237ade5
WIP
am17an Oct 10, 2025
43e0436
WIP
am17an Oct 10, 2025
56e7362
WIP
am17an Oct 11, 2025
3183a8e
fixup
am17an Oct 11, 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
49 changes: 43 additions & 6 deletions ggml/src/ggml-cuda/mmf.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,10 @@
#include "ggml.h"
#include "mmf.cuh"

void ggml_cuda_launch_mmq_ids_helper(
const int32_t * ids, int32_t * ids_src1, int32_t * ids_dst, int32_t * expert_bounds,
int n_experts, int n_tokens, int n_expert_used, int nchannels_y, int si1, int sis1, cudaStream_t stream);

void ggml_cuda_mul_mat_f(ggml_backend_cuda_context & ctx, const ggml_tensor * src0, const ggml_tensor * src1, const ggml_tensor * ids, ggml_tensor * dst) {
GGML_ASSERT( src1->type == GGML_TYPE_F32);
GGML_ASSERT(!ids || ids->type == GGML_TYPE_I32);
Expand Down Expand Up @@ -37,6 +41,12 @@ void ggml_cuda_mul_mat_f(ggml_backend_cuda_context & ctx, const ggml_tensor * sr
const int64_t ids_s0 = ids ? ids->nb[0] / ggml_type_size(ids->type) : 0;
const int64_t ids_s1 = ids ? ids->nb[1] / ggml_type_size(ids->type) : 0;

mmf_ids_data ids_info{};
mmf_ids_data * ids_info_ptr = nullptr;
ggml_cuda_pool_alloc<int32_t> ids_src_compact_dev;
ggml_cuda_pool_alloc<int32_t> ids_dst_compact_dev;
ggml_cuda_pool_alloc<int32_t> expert_bounds_dev;

// For MUL_MAT_ID the memory layout is different than for MUL_MAT:
const int64_t ncols_dst = ids ? ne2 : ne1;
const int64_t nchannels_dst = ids ? ne1 : ne2;
Expand All @@ -54,30 +64,58 @@ void ggml_cuda_mul_mat_f(ggml_backend_cuda_context & ctx, const ggml_tensor * sr
nchannels_y = ids->ne[0];
}

if (ids && ncols_dst > 16) {
const int64_t n_expert_used = ids->ne[0];
const int64_t n_experts = ne02;
const int64_t n_tokens = ne12;
const int64_t ne_get_rows = n_tokens * n_expert_used;

ids_src_compact_dev.alloc(ctx.pool(), ne_get_rows);
ids_dst_compact_dev.alloc(ctx.pool(), ne_get_rows);
expert_bounds_dev.alloc(ctx.pool(), n_experts + 1);

const int si1 = static_cast<int>(ids_s1);
const int sis1 = static_cast<int>(src1->nb[2] / src1->nb[1]);

GGML_ASSERT(sis1 > 0);

ggml_cuda_launch_mmq_ids_helper(ids_d, ids_src_compact_dev.get(), ids_dst_compact_dev.get(), expert_bounds_dev.get(),
static_cast<int>(n_experts), static_cast<int>(n_tokens), static_cast<int>(n_expert_used), static_cast<int>(ne11), si1, sis1, ctx.stream());
CUDA_CHECK(cudaGetLastError());

ids_info.ids_src_compact = ids_src_compact_dev.get();
ids_info.ids_dst_compact = ids_dst_compact_dev.get();
ids_info.expert_bounds_dev = expert_bounds_dev.get();
ids_info.n_experts = static_cast<int>(n_experts);
ids_info.sis1 = sis1;
ids_info.cols_per_tile = 16;
ids_info_ptr = &ids_info;
}

switch (src0->type) {
case GGML_TYPE_F32: {
const float * src0_d = (const float *) src0->data;
constexpr int vals_per_T = 1;
mul_mat_f_switch_cols_per_block(
src0_d, src1_d, ids_d, dst_d, ne00/vals_per_T, ne01, ncols_dst, s01/vals_per_T, stride_col_y/vals_per_T, stride_col_dst,
ids_s0, ids_s1, ne02, nchannels_y, nchannels_dst, s02/vals_per_T, stride_channel_y, stride_channel_dst,
ne03, ne3, s03/vals_per_T, s13, s3, ctx.stream());
ne03, ne3, s03/vals_per_T, s13, s3, ctx.stream(), ids_info_ptr);
} break;
case GGML_TYPE_F16: {
const half2 * src0_d = (const half2 *) src0->data;
constexpr int vals_per_T = 2;
mul_mat_f_switch_cols_per_block(
src0_d, src1_d, ids_d, dst_d, ne00/vals_per_T, ne01, ncols_dst, s01/vals_per_T, stride_col_y/vals_per_T, stride_col_dst,
ids_s0, ids_s1, ne02, nchannels_y, nchannels_dst, s02/vals_per_T, stride_channel_y, stride_channel_dst,
ne03, ne3, s03/vals_per_T, s13, s3, ctx.stream());
ne03, ne3, s03/vals_per_T, s13, s3, ctx.stream(), ids_info_ptr);
} break;
case GGML_TYPE_BF16: {
const nv_bfloat162 * src0_d = (const nv_bfloat162 *) src0->data;
constexpr int vals_per_T = 2;
mul_mat_f_switch_cols_per_block(
src0_d, src1_d, ids_d, dst_d, ne00/vals_per_T, ne01, ncols_dst, s01/vals_per_T, stride_col_y/vals_per_T, stride_col_dst,
ids_s0, ids_s1, ne02, nchannels_y, nchannels_dst, s02/vals_per_T, stride_channel_y, stride_channel_dst,
ne03, ne3, s03/vals_per_T, s13, s3, ctx.stream());
ne03, ne3, s03/vals_per_T, s13, s3, ctx.stream(), ids_info_ptr);
} break;
default:
GGML_ABORT("unsupported type: %s", ggml_type_name(src0->type));
Expand All @@ -98,10 +136,9 @@ bool ggml_cuda_should_use_mmf(enum ggml_type type, int cc, int warp_size, const
}

if (mul_mat_id) {
if (type == GGML_TYPE_F32 && src1_ncols > 32) {
if (src0_ne[1] <= 1024 && src1_ncols > 512) {
return false;
}
if ((type == GGML_TYPE_F16 || type == GGML_TYPE_BF16) && src1_ncols > 64) {
} else if(src0_ne[1] > 1024 && src1_ncols > 128) {
return false;
}
} else {
Expand Down
Loading
Loading