From 272f0748356c53e6314d83ce38e781ce530291ad Mon Sep 17 00:00:00 2001 From: danielhanchen Date: Wed, 22 Jul 2026 14:18:35 +0000 Subject: [PATCH 1/2] Prebuilt: carry macOS 14 paravirtual mmap fix; drop stale GGML_METAL_USE_BF16 define Pin unslothai/llama.cpp#37 in pr-set.json: on macOS 14 VMs the Apple Paravirtual GPU returns stale data for mmap-backed weight buffers, so the patch makes the Metal backend stop advertising buffer_from_host_ptr there and llama.cpp falls back to copied weights (same path as --no-mmap). GGML_METAL_USE_BF16 no longer exists in current llama.cpp (bf16 is runtime-probed since the ggml-metal rewrite), so the define was a no-op; remove it from the macOS matrix entry. --- .github/workflows/unsloth-prebuilt.yml | 2 +- scripts/unsloth/pr-set.json | 3 ++- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/.github/workflows/unsloth-prebuilt.yml b/.github/workflows/unsloth-prebuilt.yml index 8f41f233b923..ebdc6605ef65 100644 --- a/.github/workflows/unsloth-prebuilt.yml +++ b/.github/workflows/unsloth-prebuilt.yml @@ -328,7 +328,7 @@ jobs: # pins 13.3 (matches upstream's Intel leg, covers 2017 Intel Macs # capped at Ventura). MACOS_INCLUDE='[ - {"build":"arm64","runner":"macos-26", "expect_arch":"arm64", "deploy_target":"14.0","defines":"-DGGML_METAL_USE_BF16=ON -DGGML_METAL_EMBED_LIBRARY=ON"}, + {"build":"arm64","runner":"macos-26", "expect_arch":"arm64", "deploy_target":"14.0","defines":"-DGGML_METAL_EMBED_LIBRARY=ON"}, {"build":"x64", "runner":"macos-15-intel","expect_arch":"x86_64","deploy_target":"13.3","defines":"-DGGML_METAL=OFF"} ]' MACOS_INCLUDE="$(echo "$MACOS_INCLUDE" | jq -c .)" diff --git a/scripts/unsloth/pr-set.json b/scripts/unsloth/pr-set.json index 2f67d797e58e..752e8587b777 100644 --- a/scripts/unsloth/pr-set.json +++ b/scripts/unsloth/pr-set.json @@ -11,6 +11,7 @@ "prs": [ "https://github.com/ggml-org/llama.cpp/pull/24423/commits/c3fb97241295c196e09b783e705e84b96cd1bd74", "https://github.com/ggml-org/llama.cpp/pull/24523/commits/66f43aa655a07999c7746fe9ff5ede94835e921e", - "https://github.com/ggml-org/llama.cpp/pull/25731/commits/ce16fff2ad1d1ddfb39365fedfb4fcfd6db178f1" + "https://github.com/ggml-org/llama.cpp/pull/25731/commits/ce16fff2ad1d1ddfb39365fedfb4fcfd6db178f1", + "https://github.com/unslothai/llama.cpp/pull/37/commits/c417c7f2fd98e29048c4bdfd9482e759c11acfd2" ] } \ No newline at end of file From 92d996806153cdc3fa0bbddea79f2fcae7291205 Mon Sep 17 00:00:00 2001 From: danielhanchen Date: Thu, 23 Jul 2026 09:06:46 +0000 Subject: [PATCH 2/2] Keep the pin-set unchanged; this PR only drops the stale define GGML_METAL_USE_BF16 no longer exists as a build option in the current source tree (bf16 is runtime-probed), so passing it was a no-op. The macOS 14 paravirtual mmap pin moves to a separate decision. --- scripts/unsloth/pr-set.json | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/scripts/unsloth/pr-set.json b/scripts/unsloth/pr-set.json index 752e8587b777..2f67d797e58e 100644 --- a/scripts/unsloth/pr-set.json +++ b/scripts/unsloth/pr-set.json @@ -11,7 +11,6 @@ "prs": [ "https://github.com/ggml-org/llama.cpp/pull/24423/commits/c3fb97241295c196e09b783e705e84b96cd1bd74", "https://github.com/ggml-org/llama.cpp/pull/24523/commits/66f43aa655a07999c7746fe9ff5ede94835e921e", - "https://github.com/ggml-org/llama.cpp/pull/25731/commits/ce16fff2ad1d1ddfb39365fedfb4fcfd6db178f1", - "https://github.com/unslothai/llama.cpp/pull/37/commits/c417c7f2fd98e29048c4bdfd9482e759c11acfd2" + "https://github.com/ggml-org/llama.cpp/pull/25731/commits/ce16fff2ad1d1ddfb39365fedfb4fcfd6db178f1" ] } \ No newline at end of file