Horton

Re: Avoiding repeated snapshot computation

Andres Freund <andres@anarazel.de>

View thread
From: Andres Freund <andres@anarazel.de>
To: Robert Haas <robertmhaas@gmail.com>
Cc: pgsql-hackers@postgresql.org, Tom Lane <tgl@sss.pgh.pa.us>, Pavan Deolasee <pavan.deolasee@gmail.com>
Date: 2011-11-26T22:49:19Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Slightly reorganize struct SnapshotData.

On Saturday, November 26, 2011 11:39:23 PM Robert Haas wrote:
> On Sat, Nov 26, 2011 at 5:28 PM, Andres Freund <andres@anarazel.de> wrote:
> > On Saturday, November 26, 2011 09:52:17 PM Tom Lane wrote:
> >> I'd just as soon keep the fields in a logical order.
> > 
> > Btw, I don't think the new order is necessarily worse than the old one.
> 
> You forget to attach the benchmark results.
> 
> My impression is that cache lines on modern hardware are 64 or 128
> *bytes*, in which case this wouldn't matter a bit.
All current x86 cpus use 64bytes. The 2nd 128bit reference was a typo. Sorry 
for that.
And why is 72=>56 *bytes* (I even got that one right) not relevant for 64byte 
cachelines?

And yea. I didn't add benchmark results. I don't think I *have* to do that 
when making suggestions to somebody trying to improve something specific. I 
also currently don't have hardware where I can sensibly run at a high enough 
concurrency to see that GetSnapshotData takes ~40% of runtime.
Additional cacheline references around synchronized access can hurt to my 
knowledge...

Andres