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, 09:45 UTC
[Misc] reformat the collect-env output (#18285)
Signed-off-by: reidliu41 <reid201711@gmail.com> Co-authored-by: reidliu41 <reid201711@gmail.com>
Reid committed
May 17, 2025 at 10:46 UTC
60017dc84195a8c854ce833b95679523dcac8bb8
1 file changed
+54
-31
vllm/collect_env.py
+54
-31
@@ -637,33 +637,50 @@ def get_env_info():
637
638
639
env_info_fmt = """
640
-PyTorch version: {torch_version}
641
-Is debug build: {is_debug_build}
642
-CUDA used to build PyTorch: {cuda_compiled_version}
643
-ROCM used to build PyTorch: {hip_compiled_version}
644
-
645
-OS: {os}
646
-GCC version: {gcc_version}
647
-Clang version: {clang_version}
648
-CMake version: {cmake_version}
649
-Libc version: {libc_version}
650
-
651
-Python version: {python_version}
652
-Python platform: {python_platform}
653
-Is CUDA available: {is_cuda_available}
654
-CUDA runtime version: {cuda_runtime_version}
655
-CUDA_MODULE_LOADING set to: {cuda_module_loading}
656
-GPU models and configuration: {nvidia_gpu_models}
657
-Nvidia driver version: {nvidia_driver_version}
658
-cuDNN version: {cudnn_version}
659
-HIP runtime version: {hip_runtime_version}
660
-MIOpen runtime version: {miopen_runtime_version}
661
-Is XNNPACK available: {is_xnnpack_available}
662
-
663
-CPU:
640
+==============================
641
+ System Info
642
+==============================
643
+OS : {os}
644
+GCC version : {gcc_version}
645
+Clang version : {clang_version}
646
+CMake version : {cmake_version}
647
+Libc version : {libc_version}
648
+
649
+==============================
650
+ PyTorch Info
651
+==============================
652
+PyTorch version : {torch_version}
653
+Is debug build : {is_debug_build}
654
+CUDA used to build PyTorch : {cuda_compiled_version}
655
+ROCM used to build PyTorch : {hip_compiled_version}
656
+
657
+==============================
658
+ Python Environment
659
+==============================
660
+Python version : {python_version}
661
+Python platform : {python_platform}
662
+
663
+==============================
664
+ CUDA / GPU Info
665
+==============================
666
+Is CUDA available : {is_cuda_available}
667
+CUDA runtime version : {cuda_runtime_version}
668
+CUDA_MODULE_LOADING set to : {cuda_module_loading}
669
+GPU models and configuration : {nvidia_gpu_models}
670
+Nvidia driver version : {nvidia_driver_version}
671
+cuDNN version : {cudnn_version}
672
+HIP runtime version : {hip_runtime_version}
673
+MIOpen runtime version : {miopen_runtime_version}
674
+Is XNNPACK available : {is_xnnpack_available}
675
+
676
+==============================
677
+ CPU Info
678
+==============================
679
{cpu_info}
680
666
-Versions of relevant libraries:
681
+==============================
682
+Versions of relevant libraries
683
+==============================
684
{pip_packages}
685
{conda_packages}
686
""".strip()
@@ -671,17 +688,23 @@ Versions of relevant libraries:
688
# both the above code and the following code use `strip()` to
689
# remove leading/trailing whitespaces, so we need to add a newline
690
# in between to separate the two sections
674
-env_info_fmt += "\n"
691
+env_info_fmt += "\n\n"
692
693
env_info_fmt += """
677
-ROCM Version: {rocm_version}
678
-Neuron SDK Version: {neuron_sdk_version}
679
-vLLM Version: {vllm_version}
694
+==============================
695
+ vLLM Info
696
+==============================
697
+ROCM Version : {rocm_version}
698
+Neuron SDK Version : {neuron_sdk_version}
699
+vLLM Version : {vllm_version}
700
vLLM Build Flags:
681
-{vllm_build_flags}
701
+ {vllm_build_flags}
702
GPU Topology:
683
-{gpu_topo}
703
+ {gpu_topo}
704
705
+==============================
706
+ Environment Variables
707
+==============================
708
{env_vars}
709
""".strip()
710