zulifeng e49ea32094 feat: 新增多机 nccl test 测试脚本

2026-05-25 14:19:02 +08:00

..

__init__.py

init: project scaffolding with README, config, and requirements

2026-04-25 17:23:27 +08:00

benchmark.py

feat: 新增多机 nccl test 测试脚本

2026-05-25 14:19:02 +08:00

gpu_info.py

fix: resolve stress OOM, D2D efficiency calculation, NCCL execution failures

2026-05-07 18:09:22 +08:00

gpu_specs.py

feat: 新增多机 nccl test 测试脚本

2026-05-25 14:19:02 +08:00

health_check.py

refactor: remove hardcoding, fix AMP bug, unify English output

2026-05-07 21:32:35 +08:00

multinode_nccl_test.py

feat: 新增多机 nccl test 测试脚本

2026-05-25 14:19:02 +08:00

nccl_test.py

feat: 按 H100 生产验收标准更新测试指标与判定逻辑

2026-05-13 14:52:41 +08:00

rdma_test.py

feat: 新增 H20 支持、优化算力测试精度并修复多项稳定性问题

2026-05-12 21:41:46 +08:00

report.py

feat: 按 H100 生产验收标准更新测试指标与判定逻辑

2026-05-13 14:52:41 +08:00

stress_test.py

feat: 新增 H20 支持、优化算力测试精度并修复多项稳定性问题

2026-05-12 21:41:46 +08:00

training_sim.py

refactor: remove hardcoding, fix AMP bug, unify English output

2026-05-07 21:32:35 +08:00