]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
ggml: fix CUDA grid launch condition for large block_nums.y in binbcast (#16742)
authorleejet <redacted>
Fri, 24 Oct 2025 19:39:37 +0000 (03:39 +0800)
committerGitHub <redacted>
Fri, 24 Oct 2025 19:39:37 +0000 (21:39 +0200)
commit55945d2ef51b93821d4b6f4a9b994393344a90db
tree132bd06426b8cab34eab6b8f78b55b27a874f0ef
parent0bcb40b48c6fc6f17ba9672625e526ab2574344b
ggml: fix CUDA grid launch condition for large block_nums.y in binbcast (#16742)

* Fix CUDA grid launch condition for large block_nums.y

* add backend ops test

* reduce test  repetitions
ggml/src/ggml-cuda/binbcast.cu
tests/test-backend-ops.cpp