Re: [PATCH v3] lib/raid/xor: x86: Add AVX-512 optimized xor_gen()

[David Laight <david.laight.linux@gmail.com>]

On Wed, 17 Jun 2026 07:56:53 +0200
Christoph Hellwig <hch@lst.de> wrote:

> Can use the xor: prefix used for all other commits to lib/raid/xor?
> 
> > Benchmark on AMD Ryzen 9 9950X (Zen 5):
> > 
> >     src_cnt    avx          avx512       Improvement
> >     =======    ==========   ==========   ===========
> >     1          56353 MB/s   75388 MB/s   33%
> >     2          54274 MB/s   68409 MB/s   26%
> >     3          44649 MB/s   64042 MB/s   43%
> >     4          41315 MB/s   55002 MB/s   33%  
> 
> On my Zen 5 mobile (AMD Ryzen AI 7 PRO 350) both the existing
> AVX2 and this AVX512 code give numbers in the 200+ GB/s range.  Not
> sure if is just the different benchmarking or something else going on.

I'd expect benchmarking of the xor loop to show that it is doing
2 memory reads every clock.
At 5GHz (I'm sure my zen5 will reach that single threaded) that
is 320 GB/s for src_cnt == 1 and 160 GB/s for src_cnt == 3.

200 GB/s with 32 byte cache reads would need a 200/32 = 6GHz cpu
just for the reads.

But I expect Eric is benchmarking more code and may be limited
by data cache refills.

> 
> FYI, one or 2 sources are basically useless as they RAID5 configs
> that have no benefits over simple mirroring and thus the numbers
> aren't too interesting.

With three disks you xor two buffers (src_count == 1) to get the parity
to write to the third - so that is a valid RAID5 config.

> 
> > +DO_XOR_BLOCKS(avx512_inner, xor_avx512_2, xor_avx512_3, xor_avx512_4,
> > +	      xor_avx512_5);  
> 
> Is there really much of a benefit of doing the historic DO_XOR_BLOCKS
> vs doing the loop manually?  Especially as the common cases for a
> modern RAID will usually loop over more disks than this was built
> for.  I.e., in practice one or two source buffers only happen at the
> end of a loop over more disks.

I stopped looking at what was being tested at that point :-)

	David