llama.cpp
CUDA & CPU: support F32 kernel type for `CONV_TRANSPOSE_2D`
#17094
Merged
Merged
CUDA & CPU: support F32 kernel type for `CONV_TRANSPOSE_2D` #17094
am17an
merged
8 commits into
ggml-org:master
from
AgainstEntropy:conv2d-transpose
AgainstEntropy
changed the title
am17an
approved these changes
on 2025-11-15
am17an
approved these changes
on 2026-03-25
Refactor CUDA 2D transpose implementation to support multiple kernel …
72f60ccc
Refactor test cases for 2D convolution transpose to support dynamic k…
de1f4d7c
Refactor ggml_compute_forward_conv_transpose_2d to support both F16 a…
76a761e7
Refactor conv2d transpose kernel to use a template for kernel type, e…
96e7ae2b
Update ggml/src/ggml-cuda/conv2d-transpose.cu
d6c14331
Update ggml/src/ggml-cpu/ggml-cpu.c
c530bde2
Refactor conv2d transpose implementation by removing the conv2d_trans…
f63287db
Enhance cpu conv2d transpose implementation by introducing a template…
fd81795b
CISC
approved these changes
on 2026-03-25
Login to write a write a comment.
Login via GitHub