[quant][gpu][core][bug fix] Added memset to CacheKey for quantized cudnn add op (#76445)
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/76445
See https://github.com/pytorch/pytorch/pull/76436. It is the same idea
here.
Test Plan: python test/test_quantization.py -k test_qadd_relu_cudnn
Reviewed By: jerryzh168
Differential Revision: D35970281
Pulled By: dzdang
fbshipit-source-id: 9e983b15fa630409fa0a0906f24d94f18f1f3e7b
(cherry picked from commit 8b9be06ea7a2423e64f6ff760b1acab7423ad528)