Re: CRC32C Parallel Computation Optimization on ARM

On Wed, Oct 25, 2023 at 07:17:55AM +0900, Michael Paquier wrote:
> If you are looking at computing the CRC of records with arbitrary
> sizes, why not just generating a series with
> pg_logical_emit_message() before doing a comparison with pg_waldump or
> a custom replay loop to go through the records?  At least it would
> make the results more predictible.

I tried this.  pg_waldump on 2 million ~8kB records took around 8.1 seconds
without the patch and around 7.4 seconds with it (an 8% improvement).
pg_waldump on 1 million ~16kB records took around 3.2 seconds without the
patch and around 2.4 seconds with it (a 25% improvement).

Given the performance characteristics and relative simplicity of the patch,
I think this could be worth doing.  I suspect we'll want to do something
similar for x86, too.

-- 
Nathan Bossart
Amazon Web Services: https://aws.amazon.com