The optimal value for the word size depends mostly on the amount of L1 cache on your machine. We should be able to infer this at compile time. If autotune ever returns to Mojo, this would be a good use-case. https://github.com/fnands/mimage/blob/3805a6ab6e00f47f2a1533e888d172ed20b8e19e/mimage/utils/crc.mojo#L101
The optimal value for the word size depends mostly on the amount of L1 cache on your machine.
We should be able to infer this at compile time. If autotune ever returns to Mojo, this would be a good use-case.
mimage/mimage/utils/crc.mojo
Line 101 in 3805a6a