only run embeddingbag op on cpu (#30163)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/30163
as title
Test Plan:
```
buck run mode/opt //caffe2/benchmarks/operator_benchmark:benchmark_all_other_test -- --tag_filter all --iterations 1 --device cuda --operators embeddingbag
Parsing buck files: finished in 0.9 sec
Building: finished in 02:32.5 min (100%) 7358/7358 jobs, 1 updated
Total time: 02:33.5 min
# ----------------------------------------
# PyTorch/Caffe2 Operator Micro-benchmarks
# ----------------------------------------
# Tag : all
buck run mode/opt //caffe2/benchmarks/operator_benchmark:benchmark_all_other_test -- --tag_filter all --iterations 1 --operators embeddingbag
Parsing buck files: finished in 0.9 sec
Building: finished in 5.3 sec (100%) 5604/5604 jobs, 0 updated
Total time: 6.3 sec
# ----------------------------------------
# PyTorch/Caffe2 Operator Micro-benchmarks
# ----------------------------------------
# Tag : all
# Benchmarking PyTorch: embeddingbag
# Mode: Eager
# Name: embeddingbag_embeddingbags80_dim64_modesum_input_size8_offset0_sparseTrue_cpu
# Input: embeddingbags: 80, dim: 64, mode: sum, input_size: 8, offset: 0, sparse: True, device: cpu
Forward Execution Time (us) : 62.608
...
Reviewed By: hl475
Differential Revision: D18617540
fbshipit-source-id: 062dd73c455db8b67749078603745651b55254b2