| From: | Thomas Munro <thomas(dot)munro(at)enterprisedb(dot)com> |
|---|---|
| To: | Heikki Linnakangas <hlinnaka(at)iki(dot)fi> |
| Cc: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Andres Freund <andres(at)anarazel(dot)de>, Yuqi Gu <Yuqi(dot)Gu(at)arm(dot)com>, "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Optimize Arm64 crc32c implementation in Postgresql |
| Date: | 2018-04-04 11:47:11 |
| Message-ID: | CAEepm=02Run-Pk3xyt+RV3p1N+7cKZxN95_MamaJw8Cnw+DwjQ@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Wed, Apr 4, 2018 at 11:21 PM, Heikki Linnakangas <hlinnaka(at)iki(dot)fi> wrote:
> Yep. I got the code before the main loop, to handle the first 1-7 unaligned
> bytes, wrong. Apparently those are the only tests that call the CRC function
> with very short and unaligned input.
BTW I did some googling just now and found out that some libraries use
a technique they call "CPU probing": just try it and see if you get
SIGILL. Is that a bad idea for some reason? Here is a quick hack --
anyone got an ARM system without crc that they could test it on?
--
Thomas Munro
http://www.enterprisedb.com
| Attachment | Content-Type | Size |
|---|---|---|
| cpu-probe-hack.patch | application/octet-stream | 1.4 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Nikhil Sontakke | 2018-04-04 11:58:54 | Re: [HACKERS] logical decoding of two-phase transactions |
| Previous Message | Dmitry Dolgov | 2018-04-04 11:27:16 | Re: json(b)_to_tsvector with numeric values |