Skip to content

remove templates from soft_max_f32_submitter to allow SYCL graph updates #13724

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 1 commit into
base: master
Choose a base branch
from
+29 −84
Open

remove templates from soft_max_f32_submitter to allow SYCL graph updates #13724

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
113 changes: 29 additions & 84 deletions ggml/src/ggml-sycl/softmax.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,16 +1,16 @@
#include "softmax.hpp"

template <bool vals_smem, int ncols_template, int block_size_template, typename T>
static void soft_max_f32(const float * x, const T * mask, float * dst, const int ncols_par,
template <typename T>
static void soft_max_f32(const float * x, const T * mask, float * dst, int ncols,
const int nrows_y, const float scale, const float max_bias, const float m0,
const float m1, uint32_t n_head_log2, const sycl::nd_item<3> &item_ct1, float *buf) {
const int ncols = ncols_template == 0 ? ncols_par : ncols_template;
const float m1, uint32_t n_head_log2, const sycl::nd_item<3> &item_ct1, float *buf,
const bool vals_smem, const bool check_columns_count) {

const int tid = item_ct1.get_local_id(2);
const int rowx = item_ct1.get_group(2);
const int rowy = rowx % nrows_y; // broadcast the mask (y) in the row dimension

const int block_size = block_size_template == 0 ? item_ct1.get_local_range(2) : block_size_template;
const int block_size = check_columns_count ? item_ct1.get_local_range(2) : std::min(ncols, 1024);

const int warp_id = item_ct1.get_local_id(2) / WARP_SIZE;
const int lane_id = item_ct1.get_local_id(2) % WARP_SIZE;
Expand All @@ -35,7 +35,7 @@ static void soft_max_f32(const float * x, const T * mask, float * dst, const int
for (int col0 = 0; col0 < ncols; col0 += block_size) {
const int col = col0 + tid;

if (ncols_template == 0 && col >= ncols) {
if (check_columns_count && col >= ncols) {
break;
}

Expand Down Expand Up @@ -74,7 +74,7 @@ static void soft_max_f32(const float * x, const T * mask, float * dst, const int
#pragma unroll
for (int col0 = 0; col0 < ncols; col0 += block_size) {
const int col = col0 + tid;
if (ncols_template == 0 && col >= ncols) {
if (check_columns_count && col >= ncols) {
break;
}

Expand Down Expand Up @@ -113,7 +113,7 @@ static void soft_max_f32(const float * x, const T * mask, float * dst, const int
for (int col0 = 0; col0 < ncols; col0 += block_size) {
const int col = col0 + tid;

if (ncols_template == 0 && col >= ncols) {
if (check_columns_count && col >= ncols) {
return;
}

Expand All @@ -122,25 +122,6 @@ static void soft_max_f32(const float * x, const T * mask, float * dst, const int
}
}

template <bool vals_smem, int ncols_template, int block_size_template, typename T>
static void soft_max_f32_submitter(const float * x, const T * mask, float * dst, const int ncols_par,
const int nrows_y, const float scale, const float max_bias, const float m0,
const float m1, uint32_t n_head_log2, sycl::range<3> block_nums, sycl::range<3> block_dims,
const size_t n_local_scratch, queue_ptr stream) {
stream->submit([&](sycl::handler &cgh) {
sycl::local_accessor<float, 1> local_buf_acc(n_local_scratch, cgh);

cgh.parallel_for(
sycl::nd_range<3>(block_nums * block_dims, block_dims),
[=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
soft_max_f32<vals_smem, ncols_template, block_size_template>(x, mask, dst, ncols_par,
nrows_y, scale, max_bias, m0,
m1, n_head_log2, item_ct1,
get_pointer(local_buf_acc));
});
});
}

template<typename T>
static void soft_max_f32_sycl(const float * x, const T * mask,
float * dst, const int ncols_x, const int nrows_x,
Expand All @@ -163,64 +144,28 @@ static void soft_max_f32_sycl(const float * x, const T * mask,
const float m1 = powf(2.0f, -(max_bias / 2.0f) / n_head_log2);

const size_t local_mem_size = stream->get_device().get_info<sycl::info::device::local_mem_size>();
if (n_local_scratch*sizeof(float) < local_mem_size) {
if (ncols_x > max_block_size) {
soft_max_f32_submitter<true, 0, 0>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
return;
}
switch (ncols_x) {
case 32:
soft_max_f32_submitter<true, 32, 32>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
case 64:
soft_max_f32_submitter<true, 64, 64>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
case 128:
soft_max_f32_submitter<true, 128, 128>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
case 256:
soft_max_f32_submitter<true, 256, 256>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
case 512:
soft_max_f32_submitter<true, 512, 512>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
case 1024:
soft_max_f32_submitter<true, 1024, 1024>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
case 2048:
soft_max_f32_submitter<true, 2048, 1024>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
case 4096:
soft_max_f32_submitter<true, 4096, 1024>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
default:
soft_max_f32_submitter<true, 0, 0>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, n_local_scratch, stream);
break;
}

auto soft_max_f32_submit = [=](size_t scratch_size, bool vals_smem, bool check_columns_count) {
stream->submit([=](sycl::handler &cgh) {
sycl::local_accessor<float, 1> local_buf_acc(scratch_size, cgh);
cgh.parallel_for(
sycl::nd_range<3>(block_nums * block_dims, block_dims),
[=](sycl::nd_item<3> item_ct1) [[sycl::reqd_sub_group_size(WARP_SIZE)]] {
soft_max_f32(x, mask, dst, ncols_x, nrows_y, scale, max_bias, m0, m1, n_head_log2,
item_ct1, get_pointer(local_buf_acc), vals_smem, check_columns_count);
});
});
};

if (n_local_scratch*sizeof(float) >= local_mem_size) {
soft_max_f32_submit(WARP_SIZE, false, true);
} else if (ncols_x > max_block_size) {
soft_max_f32_submit(n_local_scratch, true, true);
} else if (ncols_x == 32 || ncols_x == 64 || ncols_x == 128 || ncols_x == 256
|| ncols_x == 512 || ncols_x == 1024 || ncols_x == 2048 || ncols_x == 4096) {
soft_max_f32_submit(n_local_scratch, true, false);
} else {
soft_max_f32_submitter<false, 0, 0>(x, mask, dst, ncols_x, nrows_y, scale,
max_bias, m0, m1, n_head_log2, block_nums,
block_dims, WARP_SIZE, stream);
soft_max_f32_submit(n_local_scratch, true, true);
}
}

Expand Down
Loading