llama.cpp
ggml-openvino: update to 2026.4.1, optimize performance, expand ops, improve device listing.
#29852
Merged
Merged
ggml-openvino: update to 2026.4.1, optimize performance, expand ops, improve device listing. #29852
ggerganov
merged
23 commits into
ggml-org:master
from
ravi9:pr-2026-10-02
ggml-openvino : Qwen3.5 MoE perf (#312)
197655c6
ggml-openvino: Update doc of compiled model cache
1153f5fc
openvino: implement PRD-compliant device enumeration and memory repor…
c92f17b5
openvino: fix multi-device listing issues from review
71b0973f
openvino : take USM entry points from the selected device platform
60628320
openvino: fuse MoE experts for models with a fused gate_up weight
d6f26f87
openvino: fix rank-3 axis handling so MoE works under stateful execution
2c0dec27
OpenVINO Backend: Upgrade graph cache to use node_idx, src_idx, node …
9e69b58e
ggml-openvino : enable more comprehensive conv fusion
1b6d42a5
enable conv ops
0d6c36ac
Reject kernel size 0 and support IM2COL_3D
52b6655d
openvino : abort when the GPU remote context cannot be created
86b044ca
openvino : fix build warnings
b1d0bab5
OpenVINO Backend: Support common MTMD ops
1c4d07d9
ggml-openvino: give a reshaping view its own ov::Tensor
fbf24bf9
ggml-openvino : compute HARDSIGMOID and EXPM1 in f32
43b399f0
ggml-openvino : update device selection and --list-devices
41939ae4
openvino : remove unreachable OpenCL queue checks
10cf91f6
openvino : update OpenVINO to 2026.4.1 and GPU drivers to 26.35.39758.10
2fc45421
docs : update OpenVINO validated models and GPU driver version
9515e4a8
ggml-openvino : skip empty views when giving a reshaping view its own…
2def4ebd
ggml-openvino : rebind the cached decoder when llama passes a differe…
abf0e740
docs : update OpenVINO validated models
308ccd80
CISC
approved these changes
on 2026-10-02
Login to write a write a comment.
Login via GitHub