A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 06:00 UTC
Support no privileged mode on CPU for docker and kubernetes deployments (#19241)
Signed-off-by: Tsai, Louie <louie.tsai@intel.com>
Louie Tsai committed
Jun 11, 2025 at 04:11 UTC
5c8d34a42cff68dde652128726f7450032b8f474
1 file changed
+2
-3
csrc/cpu/utils.cpp
+2
-3
@@ -54,8 +54,7 @@ std::string init_cpu_threads_env(const std::string& cpu_ids) {
54
*(src_mask->maskp) = *(src_mask->maskp) ^ *(mask->maskp);
55
int page_num = numa_migrate_pages(pid, src_mask, mask);
56
if (page_num == -1) {
57
- TORCH_CHECK(false,
58
- "numa_migrate_pages failed. errno: " + std::to_string(errno));
57
+ TORCH_WARN("numa_migrate_pages failed. errno: " + std::to_string(errno));
58
}
59
60
// restrict memory allocation node.
@@ -105,4 +104,4 @@ std::string init_cpu_threads_env(const std::string& cpu_ids) {
104
105
return ss.str();
106
}
108
-#endif
\ No newline at end of file
107
+#endif