|
|
@@ -221,6 +221,17 @@ RUN --mount=type=bind,from=vllm-builder,source=/wheels,target=/vllm-wheels \
|
|
|
&& python3 /tmp/patch_cuda_sm12x.py \
|
|
|
&& find /usr/local/lib/python3.12/dist-packages/vllm /opt/lmcache \
|
|
|
-name "__pycache__" -type d -prune -exec rm -rf {} + 2>/dev/null || true
|
|
|
+# b12x loader allocation patch (discord, 2026-09-07): pinned_wc -> managed
|
|
|
+# in b12x/integration/vllm/loader.py — two occurrences (the weight_pool call
|
|
|
+# and the matching audit label). Reported prefill 1350 -> ~2000 tok/s. No
|
|
|
+# recompiles; hard-gated: any pinned_wc left behind fails the build.
|
|
|
+RUN set -eux; \
|
|
|
+ F=/usr/local/lib/python3.12/dist-packages/b12x/integration/vllm/loader.py; \
|
|
|
+ sed -i 's/allocation="pinned_wc"/allocation="managed"/; \
|
|
|
+ s/"allocation": "pinned_wc"/"allocation": "managed"/' "$F"; \
|
|
|
+ test "$(grep -c 'allocation="managed"' "$F")" = "1"; \
|
|
|
+ grep -Fq '"allocation": "managed"' "$F"; \
|
|
|
+ ! grep -q 'pinned_wc' "$F"
|
|
|
# Build gate: the r27 stack must import with the GB10 patches in place and
|
|
|
# carry the qualified package trees (marker spot-checks from the source lock).
|
|
|
RUN set -eux; \
|
