[TorchBench] Performance Signal Detected

[zzpmiracle]

TorchBench CI has detected a performance signal.

Affected Tests:

• eval-cuda-fp32:
  • functorch_dp_cifar10[disc (latency)] 1.779 -> 1.679, +5.6211%
  • functorch_maml_omniglot[blade (latency)] 0.71 -> 0.595, +16.1972%
  • hf_Bert_mini[blade (latency)] 0.546 -> 0.663, -21.4286%
  • maml_omniglot[dynamo-blade (latency)] 0.495 -> 0.462, +6.6667%
• eval-cuda-fp16:
  • dlrm[dynamo-disc (latency)] 1.754 -> 1.647, +6.1003%
  • functorch_maml_omniglot[blade (latency)] 0.521 -> 0.636, -22.0729%
  • maml_omniglot[dynamo-blade (latency)] 0.486 -> 0.363, +25.3086%

detail data can be seen in oss://bladedisc-ci/TorchBench/gpu/full/20230630-15 created by TorchBench CI automatically