Skip to content

vulkan: improve topk perf for large k, fix overflow in unit tests #17582

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 1 commit into from
Nov 29, 2025
Merged

vulkan: improve topk perf for large k, fix overflow in unit tests #17582

Show file tree
Hide file tree
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
4 changes: 3 additions & 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 @@ -10239,7 +10239,9 @@ static void ggml_vk_topk(ggml_backend_vk_context * ctx, vk_context& subctx, cons

// Prefer going as small as num_topk_pipelines - 3 for perf reasons.
// But if K is larger, then we need a larger workgroup
uint32_t max_pipeline = num_topk_pipelines - 3;
uint32_t max_pipeline = num_topk_pipelines - 1;
uint32_t preferred_pipeline = std::max(num_topk_pipelines - 3, (uint32_t)log2f(float(k)) + 2);
max_pipeline = std::min(preferred_pipeline, max_pipeline);
uint32_t min_pipeline = (uint32_t)log2f(float(k)) + 1;
// require full subgroup
min_pipeline = std::max(min_pipeline, ctx->device->subgroup_size_log2);
Expand Down
8 changes: 5 additions & 3 deletions tests/test-backend-ops.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1446,14 +1446,14 @@ struct test_case {
const uint64_t target_flops_cpu = 8ULL * GFLOP;
const uint64_t target_flops_gpu = 100ULL * GFLOP;
uint64_t target_flops = is_cpu ? target_flops_cpu : target_flops_gpu;
n_runs = std::min<int>(ggml_graph_size(gf) - ggml_graph_n_nodes(gf), target_flops / op_flops(out)) + 1;
n_runs = (int)std::min<int64_t>(ggml_graph_size(gf) - ggml_graph_n_nodes(gf), target_flops / op_flops(out)) + 1;
} else {
// based on memory size
const size_t GB = 1ULL << 30;
const size_t target_size_cpu = 8 * GB;
const size_t target_size_gpu = 32 * GB;
size_t target_size = is_cpu ? target_size_cpu : target_size_gpu;
n_runs = std::min<int>(ggml_graph_size(gf) - ggml_graph_n_nodes(gf), target_size / op_size(out)) + 1;
n_runs = (int)std::min<int64_t>(ggml_graph_size(gf) - ggml_graph_n_nodes(gf), target_size / op_size(out)) + 1;
}

// duplicate the op
Expand Down Expand Up @@ -8043,7 +8043,9 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_perf() {
}

test_cases.emplace_back(new test_argsort(GGML_TYPE_F32, {65000, 16, 1, 1}));
for (auto k : {1, 10, 40}) {

test_cases.emplace_back(new test_top_k(GGML_TYPE_F32, {2, 1, 1, 1}, 1));
for (auto k : {1, 10, 40, 400}) {
for (auto nrows : {1, 16}) {
for (auto cols : {k, 1000, 65000, 200000}) {
test_cases.emplace_back(new test_top_k(GGML_TYPE_F32, {cols, nrows, 1, 1}, k));
Expand Down
Loading