The Intel SSE 4.2 instruction set added the new CRC32C computation instructions. They are super cool and they are SO MUCH faster than doing the CRC calculation in software.

But, as I was coding up crc32c for something, I noticed that the CRC32 instruction has a mode that returns a 64 bit value for 64 bit source operands, and a 64 bit value for an 8 bit source operand (in addition to 32 bit values for all the others).

https://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-32-architectures-software-developer-instruction-set-reference-manual-325383.pdf

See page 2A 3-225 and 226.

This means there is a strange

unsigned __int64 _mm_crc32_u64() in the Intel intrinsics (see https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#ig_expand=1494,1494&techs=SSE4_2).

It isn't a big deal, return a 64 bit value with the top two bytes set to zero, the result is correct. But, I'm sure there was some reason why Intel did this, has anyone found a good explanation for this somewhere - I'm asking only out of curiosity, not asserting that it is wrong or something.

I would have thought unsigned __int32 _mm_crc32_u64() would be more reasonable at least for the intrinsic, no?