Closed ESI-SYD closed 1 year ago
Compare with the ww02.4, there is a accuracy failure on Torchbench model Super_SloMo on TorchInductor CPU Performance Dashboard on ww03.2:
WW03.2 SW info:
WW02.4 SW info:
cpu eval Super_SloMo [2023-01-16 06:56:05,488] torch._dynamo.utils: [ERROR] RMSE (res-fp64): 0.01816, (ref-fp64): 0.00000 and shape=torch.Size([]) FAIL
python -m torch.backends.xeon.run_cpu --core_list 0 --ncores_per_instance 1 benchmarks/dynamo/torchbench.py --accuracy --float32 -dcpu -n50 --inductor --no-skip --dashboard --only Super_SloMo --cold_start_latency --batch_size 1 --threads 1
The issue has been fixed according to WW03.4 dashboard: https://github.com/pytorch/pytorch/issues/93531
🐛 Describe the bug
Compare with the ww02.4, there is a accuracy failure on Torchbench model Super_SloMo on TorchInductor CPU Performance Dashboard on ww03.2:
WW03.2 SW info:
WW02.4 SW info:
Error logs
Minified repro