cuda, hip: GDN chunked kernel (#29353)
* cuda, hip: GDN chunked kernel Ported-from: https://github.com/tetherto/qvac-fabric-llm.cpp/pull/257 Source-commit: 53f0027d8070715fe9ad2131f3201bb0001d4eed Co-authored-by: Aman Gupta <aman.gupta@tether.io> * fix older builds * review fixes * cuda: enable GDN MMA on tested Ampere and Ada GPUs Enable SM86 and SM89 after RTX 3090 and RTX 4090 correctness and prefill benchmarks. Keep other untested architectures on the existing fallback. Assisted-by: Codex * cuda : use refactored swizzle API in GDN MMA * ggml : GDN only guarantees the final state when n_tokens > K * cuda : add short GDN MMA phase comments --------- Co-authored-by: gianni-cor <gianfranco.cordella@tether.io> Co-authored-by: Aman Gupta <aman.gupta@tether.io>
This commit is contained in:
7 files changed
+564
-14
No files matched your search
@@ -11950,6 +11950,18 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 4, 64, 33, 1, 1, false, true));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 4, 64, 100, 1, 1, false, true));
|
||||
|
||||
// MMA-eligible GDN shapes: whole heads, value slices, tails, and the HIP prefill threshold.
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 65, 1, 1));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 65, 3, 4));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 64, 1, 4));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 257, 1, 3));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 512, 1, 2));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 513, 1, 3));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 513, 2, 2, true));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 2048, 1, 2));
|
||||
// Keep the WebGPU dispatch below 65535 workgroups (32 per token for this shape).
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 16, 128, 2047, 2, 3));
|
||||
|
||||
// K > 1: output keeps the last min(n_tokens, K) per-token snapshots, ordered most-recent-first
|
||||
// (slot 0 = final state, slot s = state s tokens back).
|
||||
// exact-match cases (K == n_seq_tokens):
|
||||
@@ -11959,9 +11971,6 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 8, 128, 4, 1, 1, false, false, /*K=*/4));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 4, 64, 4, 2, 1, false, true, /*K=*/4));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 8, 32, 4, 2, 2, false, true, /*K=*/4));
|
||||
// overflow: n_tokens > K — only the last K snapshots kept.
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 4, 32, 8, 1, 1, false, false, /*K=*/3));
|
||||
test_cases.emplace_back(new test_gated_delta_net(GGML_TYPE_F32, 4, 64, 16, 2, 1, false, false, /*K=*/4));
|
||||
|
||||
// gdn + cache cpy fusion (K > 1)
|
||||
test_cases.emplace_back(new test_gated_delta_net_cache_fusion(GGML_TYPE_F32, 4, 32, 2, 1, 2));
|
||||
|
||||
Reference in new issue
Block a user