Skip to content

vulkan: fix SSM_CONV PP scaling with large ubatch sizes #20379

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
0cc4m merged 2 commits into from
Mar 12, 2026
+17 −11
Merged

vulkan: fix SSM_CONV PP scaling with large ubatch sizes #20379

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5bfba8a
vulkan: optimize SSM_CONV workgroup dispatch for large ubatch
ProgenyAlpha Mar 11, 2026
25a1561
vulkan: remove unused shared memory declaration in SSM_CONV
ProgenyAlpha Mar 11, 2026
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
2 changes: 1 addition & 1 deletion ggml/src/ggml-vulkan/ggml-vulkan.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4567,7 +4567,7 @@ static void ggml_vk_load_shaders(vk_device& device) {
ggml_vk_create_pipeline(device, device->pipeline_ssm_scan_f32_d256, "ssm_scan_256_f32", ssm_scan_f32_len, ssm_scan_f32_data, "main", 8, sizeof(vk_op_ssm_scan_push_constants), {1, 1, 1}, {256, device->subgroup_size, 16}, 1, true, true);
}

ggml_vk_create_pipeline(device, device->pipeline_ssm_conv_f32, "ssm_conv_f32", ssm_conv_f32_len, ssm_conv_f32_data, "main", 3, sizeof(vk_op_ssm_conv_push_constants), {32, 1, 1}, {32}, 1);
ggml_vk_create_pipeline(device, device->pipeline_ssm_conv_f32, "ssm_conv_f32", ssm_conv_f32_len, ssm_conv_f32_data, "main", 3, sizeof(vk_op_ssm_conv_push_constants), {32, 16, 1}, {32, 16}, 1);

ggml_vk_create_pipeline(device, device->pipeline_opt_step_adamw_f32, "opt_step_adamw_f32", opt_step_adamw_f32_len, opt_step_adamw_f32_data, "main", 5, sizeof(vk_op_push_constants), {512, 1, 1}, {}, 1);

Expand Down
28 changes: 18 additions & 10 deletions ggml/src/ggml-vulkan/vulkan-shaders/ssm_conv.comp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,8 +5,9 @@
#include "types.glsl"

layout(constant_id = 0) const uint BLOCK_SIZE = 32;
layout(constant_id = 1) const uint TOKENS_PER_WG = 16;

layout(local_size_x_id = 0, local_size_y = 1, local_size_z = 1) in;
layout(local_size_x_id = 0, local_size_y_id = 1, local_size_z = 1) in;

layout(binding = 0) readonly buffer Src0 { float src0[]; };
layout(binding = 1) readonly buffer Src1 { float src1[]; };
Expand All @@ -19,26 +20,33 @@ layout(push_constant) uniform PushConstants {
uint nc; uint ncs; uint nr; uint n_t; uint n_s;
};

shared float s_conv[4];
Comment thread
0cc4m marked this conversation as resolved.
Outdated

void main() {
const uint global_thread_id = gl_GlobalInvocationID.x;
const uint i2 = gl_WorkGroupID.y;
const uint i1 = gl_GlobalInvocationID.x;
const uint i2 = gl_WorkGroupID.y * TOKENS_PER_WG + gl_LocalInvocationID.y;
const uint i3 = gl_WorkGroupID.z;

if (global_thread_id >= nr || i2 >= n_t || i3 >= n_s) {
if (i1 >= nr || i2 >= n_t || i3 >= n_s) {
return;
}

const uint i1 = global_thread_id;
const uint src0_base = i3 * (nb02 / 4) + i2 + i1 * (nb01 / 4);
const uint src1_base = i1 * (nb11 / 4);
const uint dst_idx = i3 * (dst_nb2 / 4) + i2 * (dst_nb1 / 4) + i1;

float sum = 0.0;
[[unroll]] for (uint i0 = 0; i0 < nc; i0++) {
const uint src0_idx = src0_base + i0;
const uint src1_idx = src1_base + i0;
sum += src0[src0_idx] * src1[src1_idx];

if (nc == 4) {
sum = dot(
vec4(src0[src0_base], src0[src0_base + 1], src0[src0_base + 2], src0[src0_base + 3]),
vec4(src1[src1_base], src1[src1_base + 1], src1[src1_base + 2], src1[src1_base + 3])
);
} else {
[[unroll]] for (uint i0 = 0; i0 < nc; i0++) {
sum += src0[src0_base + i0] * src1[src1_base + i0];
}
}

const uint dst_idx = i3 * (dst_nb2 / 4) + i2 * (dst_nb1 / 4) + i1;
dst[dst_idx] = sum;
}