fix: remove expandable_segments — CoreX CUDACachingAllocator不支持

BI-V100 CoreX PyTorch的CUDACachingAllocator.cpp:545没有实现
expandable segment特性,导致模型加载阶段(VocabParallelEmbedding)
直接INTERNAL ASSERT FAILED崩溃。

替换为max_split_size_mb:512减少内存碎片化。
This commit is contained in:
dylanyunlon
2026-08-13 09:08:47 +00:00
committed by root
parent 1ea2100cb8
commit c1e7065076

View File

@@ -48,14 +48,4 @@ env:
- name: BI100_MOE_COREX_TOPK_SOFTMAX
value: '1'
- name: PYTORCH_CUDA_ALLOC_CONF
value: expandable_segments:True
- name: LD_PRELOAD
value: /workspace/qwen3_6_scripts/cccl_preload_allocator.so
- name: CCCL_ALLOC_BIN_GROWTH
value: '8'
- name: CCCL_ALLOC_MIN_BIN
value: '3'
- name: CCCL_ALLOC_MAX_BIN
value: '13'
- name: CCCL_ALLOC_MAX_CACHED_MB
value: '4096'
value: max_split_size_mb:512