Update readme for openmp on mac

[krrishnarraj]

No description provided.

[karpathy]

I'm just really unable to get any substantial gains here, even following the new readme. Am I alone?

[krrishnarraj]

Then gain was more than 2x on my macbook with openmp enabled.

Chip: M1 pro with 10 cpu cores
Memory: 32GB
OS: 13.5
/usr/bin/gcc --version: Apple clang version 14.0.3
/opt/homebrew/opt/llvm/bin/clang --version: Homebrew clang version 16.0.6

Launched the binary as: ./run llama2_7b.bin

Run	tok/s
make runfast	1.79
make runfast CC=/opt/homebrew/opt/llvm/bin/clang	1.79
make runomp CC=/opt/homebrew/opt/llvm/bin/clang	4.01
above binary with OMP_NUM_THREADS=4	3.96

Not setting OMP_NUM_THREADS gave the best performance.

For further debugging, let us know your setup

[karpathy]

ok let's go with this for now ty