Revert D71711852

Summary:
This diff reverts D71711852
causes qps regression as per:

https://fb.workplace.com/groups/1982103978790468/posts/2575339479466912/?comment_id=2577572732576920

bypass-github-export-checks "these should not have triggered for DK revert"

Reviewed By: izaitsevfb

Differential Revision: D72140190

fbshipit-source-id: 94dfe23dd6b6cee3d95a40fcb28558d4e23c7893

Author: Dark Knight

userbenchmark/dynamo/dynamobench/common.py

@@ -3545,16 +3545,6 @@ def run(runner, args, original_dir=None):
         if args.devices == ["xpu"]:
             torch.use_deterministic_algorithms(True, warn_only=True)
         os.environ["CUBLAS_WORKSPACE_CONFIG"] = ":4096:8"
-        # TODO(eqy): revisit when cuBLASLt workspace size is bumped
-        # if args.only is not None and args.only in {
-        #     "DebertaForQuestionAnswering",
-        #     "RobertaForQuestionAnswering",
-        #     "nvidia_deeprecommender",
-        #     "volo_d1_224",
-        # }:
-        #     # These seem unhappy with numerics of larger cuBLASLt workspace
-        #     # sizes following #145130 (due to enabling split-k?)
-        #     torch.backends.cuda.matmul.allow_fp16_reduced_precision_reduction = False
         torch.backends.cudnn.deterministic = True
         torch.backends.cudnn.allow_tf32 = False
         torch.backends.cudnn.benchmark = False