Klaus Post
Klaus Post
(project will be published under minio)
Keeping it around as a TODO item. I close the ones I don't plan to do.
LGTM. That would solve #235 for me.
A PR would make the change easy to add. Meanwhile I wouldn't expect `XORQ(tmp, tmp)` to be a huge overhead and use that as base. For most CPUs zeroing a...
Because you are swapping inputs? Remember that compared to Intel syntax *all* parameters are swapped. `VPCMPUQ(imm8, zmm3, zmm2, k1)`
Yes, there are pure SSE2 (non+VEX)/AVX2 (VEX) versions of these. There are "duplicates": ``` 66 0F 3A 44 /r ib PCLMULQDQ xmm1, xmm2/m128, imm8 VEX.128.66.0F3A.WIG 44 /r ib VPCLMULQDQ xmm1,...
@harshavardhana Will it prefer 5 over 10, 6 over 12, 7 over 14 - or will this just affect sizes divisible by 16?
Yeah.. Like I get it for 16 -> 8 and 14 -> 7.... But 10 -> 5 and maybe 12 ->6 seems like a rather big drop...
This seems like a metric that should be impossible to return in a scalable fashion. Could you maybe explain exactly what it is?
@shtripat Is it a 1 minute window, or how are expecting to keep it scalable? Even 1 minute can accumulate a huge number of requests. It seems like just logging...