]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
vulkan: fix GPU deduplication logic. (#19222)
authorOleksandr Kuvshynov <redacted>
Thu, 5 Feb 2026 08:06:59 +0000 (03:06 -0500)
committerGitHub <redacted>
Thu, 5 Feb 2026 08:06:59 +0000 (09:06 +0100)
commita498c75ad1d7019bc2cc1e5d83e11bb8a062861f
tree8fe908908e7576507aceb6c765aaf173ba053b99
parent3409ab842d988c3e0dc0d3110dd793a5e187d37e
vulkan: fix GPU deduplication logic. (#19222)

* vulkan: fix GPU deduplication logic.

As reported in https://github.com/ggml-org/llama.cpp/issues/19221, the
(same uuid, same driver) logic is problematic for windows+intel igpu.

Let's just avoid filtering for MoltenVK which is apple-specific, and
keep the logic the  same as before 88d23ad5 - just dedup based on UUID.

Verified that MacOS + 4xVega still reports 4 GPUs with this version.

* vulkan: only skip dedup when both drivers are moltenVk
ggml/src/ggml-vulkan/ggml-vulkan.cpp