[Performance] Faster RNNs #8469

Job	Run time
build-wheel-mac (3.8, 3.8)	1m 20s
build-wheel-linux (3.8, cp38-cp38, --index-url https://download.pytorch.org/whl/cpu, "['cpu', '11...	1m 38s
build-wheel-mac (3.9, 3.9)	2m 50s
build-wheel-linux (3.9, cp39-cp39, --index-url https://download.pytorch.org/whl/cpu, "['cpu', '11...	1m 31s
build-wheel-mac (3.10, 3.10.3)	1m 49s
build-wheel-linux (3.10, cp310-cp310, --index-url https://download.pytorch.org/whl/cpu, "['cpu', ...	1m 32s
build-wheel-windows (3.8, 3.8)	2m 43s
build-wheel-mac (3.11, 3.11)	1m 36s
build-wheel-linux (3.11, cp311-cp311, --index-url https://download.pytorch.org/whl/cpu, "['cpu', ...	1m 34s
build-wheel-windows (3.9, 3.9)	3m 21s
build-wheel-windows (3.10, 3.10.3)	3m 55s
build-wheel-windows (3.11, 3.11)	3m 22s
test-wheel (linux, ubuntu-20.04, 3.8)	1m 12s
test-wheel (linux, ubuntu-20.04, 3.9)	1m 15s
test-wheel (linux, ubuntu-20.04, 3.10)	1m 16s
test-wheel (linux, ubuntu-20.04, 3.11)	1m 14s
test-wheel (mac, macos-latest, 3.8)	1m 13s
test-wheel (mac, macos-latest, 3.9)	1m 9s
test-wheel (mac, macos-latest, 3.10)	1m 9s
test-wheel (mac, macos-latest, 3.11)	1m 4s
test-wheel-windows	1s
	36m 44s

Provide feedback