Skip to content

[ET-VK] Replace Uniform buffers with push constants for native layer norm op #9831

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 2 commits into from
Apr 3, 2025
Merged

[ET-VK] Replace Uniform buffers with push constants for native layer norm op #9831

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
591ee4f
[ET-VK] Replace Uniform buffers with push constants for native layer …
trivedivivek Apr 2, 2025
3767208
Update on "[ET-VK] Replace Uniform buffers with push constants for na…
trivedivivek Apr 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
8 changes: 5 additions & 3 deletions backends/vulkan/runtime/graph/ops/glsl/native_layer_norm.glsl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,9 +27,11 @@ ${layout_declare_tensor(B, "r", "t_in", DTYPE, STORAGE)}
${layout_declare_tensor(B, "r", "t_weight", DTYPE, STORAGE)}
${layout_declare_tensor(B, "r", "t_bias", DTYPE, STORAGE)}

${layout_declare_ubo(B, "ivec3", "out_limits")}
${layout_declare_ubo(B, "ivec4", "sizes")}
${layout_declare_ubo(B, "float", "epsilon")}
layout(push_constant) uniform PRECISION restrict Block {
ivec3 out_limits;
ivec4 sizes;
float epsilon;
};

layout(local_size_x_id = 0, local_size_y_id = 1, local_size_z_id = 2) in;

Expand Down
13 changes: 7 additions & 6 deletions backends/vulkan/runtime/graph/ops/impl/NativeLayerNorm.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -101,19 +101,20 @@ void add_native_layer_norm_node(
vkapi::MemoryAccessType::WRITE},
{{in, arg_weight, arg_bias}, vkapi::MemoryAccessType::READ}},
// Shader params buffers
{
t_out->logical_limits_ubo(),
t_out->sizes_ubo(),
graph.create_params_buffer(epsilon),
},
{},
// Specialization Constants
{
t_input->hashed_layout(),
t_out->hashed_layout(),
},
// Resizing Logic
resize_native_layer_norm_node,
{normalized_shape}));
{normalized_shape},
{
graph.logical_limits_pc_of(out_val->at(0)),
graph.sizes_pc_of(out_val->at(0)),
PushConstantDataInfo(&epsilon, sizeof(epsilon)),
}));
}

void native_layer_norm(ComputeGraph& graph, const std::vector<ValueRef>& args) {
Expand Down
Loading