Re: [PATCH] SVE popcount support - Mailing list pgsql-hackers

From Nathan Bossart
Subject Re: [PATCH] SVE popcount support
Date
Msg-id Z8Y0pKrlQPPbhVKC@nathan
Whole thread Raw
In response to Re: [PATCH] SVE popcount support  ("Chiranmoy.Bhattacharya@fujitsu.com" <Chiranmoy.Bhattacharya@fujitsu.com>)
List pgsql-hackers
On Wed, Feb 19, 2025 at 09:31:50AM +0000, Chiranmoy.Bhattacharya@fujitsu.com wrote:
>> Hm.  Any idea why that is?  I wonder if the compiler isn't using as many
>> SVE registers as it could for this.
> 
> Not sure, we tried forcing loop unrolling using the below line in the MakeFile
> but the results are the same.
> 
> pg_popcount_sve.o: CFLAGS += ${CFLAGS_UNROLL_LOOPS} -march=native

Interesting.  I do see different assembly with the 2 and 4 register
versions, but I didn't get to testing it on a machine with SVE support
today.

Besides some additional benchmarking, I might make some small adjustments
to the patch.  But overall, it seems to be in decent shape.

-- 
nathan



pgsql-hackers by date:

Previous
From: "Devulapalli, Raghuveer"
Date:
Subject: RE: Improve CRC32C performance on SSE4.2
Next
From: Tom Lane
Date:
Subject: Re: proposal - plpgsql - support standard syntax for named arguments for cursors