llama.cpp
CUDA: fix build error from ambiguous __half conversions in conv2d
#15690

Merged

CUDA: fix build error from ambiguous __half conversions in conv2d #15690

qnixsynapse merged 5 commits into master from cuda/fix_conv2d_type

CUDA: fix build error from ambiguous __half conversions in conv2d

22744493

qnixsynapse requested a review from

JohannesGaessler 8 days ago

github-actions added Nvidia GPU

github-actions added ggml

CUDA: Replace custom to_float helper with unified ggml_cuda_cast and …

1d4e7ee9

JohannesGaessler dismissed these changes on 2025-08-31

JohannesGaessler dismissed their stale review 7 days ago

Forgot to check CI passing.

CUDA: Add missing convert.cuh header

80a679b9

CUDA: remove unnecessary extension in ggml_cuda_cast

a704126b

qnixsynapse requested a review from

JohannesGaessler 7 days ago

JohannesGaessler approved these changes on 2025-08-31

CUDA: Address review comment, remove second type template argument

5251a404

qnixsynapse merged b66df9d9 into master 7 days ago

qnixsynapse deleted the cuda/fix_conv2d_type branch 7 days ago

Reviewers

JohannesGaessler

Assignees

No one assigned

Labels

Nvidia GPU ggml

Milestone

No milestone