Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

cuda : add warp-per-row wkv7 kernel for single-token decode (#26111)

Browse Source
This commit is contained in:
0
2026-08-11 20:46:23 +03:00
committed by GitHub
parent f785fc9ea4
commit 5988633170
2 changed files with 56 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/test-backend-ops.cpp
+1
View File
@@ -8827,6 +8827,7 @@ static std::vector<std::unique_ptr<test_case>> make_test_cases_eval() {
test_cases.emplace_back(new test_rwkv_wkv6(GGML_TYPE_F32, 32, 64, 128, 4));
test_cases.emplace_back(new test_rwkv_wkv7(GGML_TYPE_F32, 32, 64, 1, 1));
test_cases.emplace_back(new test_rwkv_wkv7(GGML_TYPE_F32, 32, 64, 1, 4));
test_cases.emplace_back(new test_rwkv_wkv7(GGML_TYPE_F32, 32, 64, 32, 1));
test_cases.emplace_back(new test_rwkv_wkv7(GGML_TYPE_F32, 32, 64, 32, 4));
test_cases.emplace_back(new test_rwkv_wkv7(GGML_TYPE_F32, 32, 64, 128, 4));