Date: Sat, 6 Aug 2022 11:13:26 -0700
From: Nathan Bossart <nathandbossart@gmail.com>
To: Andres Freund <andres@anarazel.de>
Cc: John Naylor <john.naylor@enterprisedb.com>,
	John Naylor <jcnaylor@gmail.com>,
	pgsql-hackers <pgsql-hackers@postgresql.org>
Subject: Re: optimize lookups in snapshot [sub]xip arrays
Message-ID: <20220806181326.GA1216305@nathanxps13>
References: <20220802221301.GA742739@nathanxps13>
 <20220802225539.nm7bgkwvkpzgib5x@awork3.anarazel.de>
 <20220802234357.GA802740@nathanxps13>
 <20220803180658.3ofq6v34wxchg4wc@awork3.anarazel.de>
 <20220803202540.GA1088968@nathanxps13>
 <CAFBsxsH51bBtyr3rf13x7YMNkFfuzZPSQxxbfUepBj8m4=ck=Q@mail.gmail.com>
 <20220804221551.GA1090071@nathanxps13>
 <CAFBsxsGmNefB-LAo8hNXLDAwWN0vHXc6H2A+NfFUQrj=rn50DQ@mail.gmail.com>
 <20220805202510.GA1208450@nathanxps13>
 <20220805220434.6h2aa7eww5rxird6@awork3.anarazel.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
In-Reply-To: <20220805220434.6h2aa7eww5rxird6@awork3.anarazel.de>
Archived-At: 
 <https://www.postgresql.org/message-id/20220806181326.GA1216305%40nathanxps13>
Precedence: bulk

On Fri, Aug 05, 2022 at 03:04:34PM -0700, Andres Freund wrote:
> But mainly I'd expect to find a difference if the SIMD code were optimized a
> further on the basis of not needing to return the offset. E.g. by
> replacing _mm_packs_epi32 with _mm_or_si128, that's cheaper.

I haven't been able to find a significant difference between the two.  If
anything, the _mm_packs_epi* approach actually seems to be slightly faster
in some cases.  For something marginally more concrete, I compared the two
in perf-top and saw the following for the relevant instructions:

_mm_packs_epi*:
	0.19 │       packssdw   %xmm1,%xmm0
	0.62 │       packssdw   %xmm1,%xmm0
	7.14 │       packsswb   %xmm1,%xmm0

_mm_or_si128:
	1.52 │       por        %xmm1,%xmm0
	2.05 │       por        %xmm1,%xmm0
	5.66 │       por        %xmm1,%xmm0

I also tried a combined approach where I replaced _mm_packs_epi16 with
_mm_or_si128:
	1.16 │       packssdw   %xmm1,%xmm0
	1.47 │       packssdw   %xmm1,%xmm0
	8.17 │       por        %xmm1,%xmm0

Of course, this simplistic analysis leaves out the impact of the
surrounding instructions, but it seems to support the idea that the
_mm_packs_epi* approach might have a slight edge.

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