Carry-less multiplication

[kreuzerkrieg]

Is carry-less multiplication supported by EVE? Something like PCLMULQDQ instruction?

[jfalcou]

We don't at the moment. What are your use case ?

[kreuzerkrieg]

SIMD accelerated CRC32 calculation following this document. I have intrinsic version of this code and would like to try and implement it using EVE

[jfalcou]

OK. At the moment, we don't support those. We can surely does, this means finding a way to implement it on non-X86 architectures.
Meanwhile, eve::wide can be directly passed to intrinsics so you can rewrite the rest fo the code without issue then just pass w/e to these intrinsics.

[DenisYaroshevskiy]

Posted an example below

[kreuzerkrieg]

We can surely does, this means finding a way to implement it on non-X86 architectures.

https://en.wikipedia.org/wiki/CLMUL_instruction_set
Intel, AMD, ARMv8 and even SPARC

[DenisYaroshevskiy]

FYI: you can always mix eve with intrinsics if you need to https://godbolt.org/z/Mvbhj77ch

[kreuzerkrieg]

ha! looks interesting, will try to do that