Skip to content

[ET-VK] Tuning native layer norm local workgroup size to improve thread occupancy during reduce. #9984

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 6 commits into
base: gh/trivedivivek/75/base
Choose a base branch
from
Open

[ET-VK] Tuning native layer norm local workgroup size to improve thread occupancy during reduce. #9984

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
b5ed426
[ET-VK] Tuning native layer norm local workgroup size to improve thre…
trivedivivek Apr 8, 2025
78559e1
Update on "[ET-VK] Tuning native layer norm local workgroup size to i…
trivedivivek Apr 11, 2025
4fc4454
Update on "[ET-VK] Tuning native layer norm local workgroup size to i…
trivedivivek Apr 14, 2025
0e1a282
Update on "[ET-VK] Tuning native layer norm local workgroup size to i…
trivedivivek Apr 14, 2025
e0860b4
Update on "[ET-VK] Tuning native layer norm local workgroup size to i…
trivedivivek Apr 16, 2025
9a46e90
Update on "[ET-VK] Tuning native layer norm local workgroup size to i…
trivedivivek Apr 16, 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
13 changes: 12 additions & 1 deletion 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 @@ -84,7 +84,18 @@ void add_native_layer_norm_node(
std::vector<int64_t> in_sizes = t_input->sizes();

utils::uvec3 global_size = t_out->logical_limits();
utils::uvec3 local_size = graph.create_local_wg_size(global_size);
utils::uvec3 local_size;

// Since the shader sets shared memory scale factor > 1, if dispatch is
// greater than maximum WG size. Setting WG size in X axis to max WG size,
// would allow best thread utilization.
if (global_size[0] > 64) {
local_size = {64, 1, 1};
} else {
// If thread size in X axis is smaller or equal to maximum WG size, we can
// let the function decide the best WG size.
local_size = graph.create_local_wg_size(global_size);
}

std::string kernel_name("native_layer_norm");
kernel_name.reserve(kShaderNameReserve);
Expand Down
Loading