Re: [PATCH] write-tree performance problems

From: Linus Torvalds <torvalds@osdl.org>
Date: Apr 20, 2005, 17:06 UTC
Message-ID: <Pine.LNX.4.58.0504200957030.6467@ppc970.osdl.org>
In-Reply-To: <200504201237.38374.mason@suse.com>

On Wed, 20 Apr 2005, Chris Mason wrote:

> 
> At any rate, the time for a single write-tree is pretty consistent.  Before it 
> was around .5 seconds, and with this change it goes down to .128s.

Oh, wow.

I bet your SHA1 implementation is done with hand-optimized and scheduled x86 MMX code or something, while my poor G5 is probably using some slow generic routine. As a result, it only improved by 33% for me since the compression was just part of the picture, but with your cheap SHA1 the compression costs really dominated, and so it's almost four times faster for you.

Anyway, that's good. It definitely means that I consider tree writing to be "fast enough". You can commit patches in a third of a second on your machine.

I'll consider the problem solved for now. Yeah, I realize that it still takes you half a minute to commit the 100 quilt patches, but I just can't bring myself to think it's a huge problem in the kind of usage patterns I think are realistic.

If somebody really wants to replace quilt with git, he'd need to spend some effort on it. If you just want to work together reasonably well, I think 3 patches per second is pretty much there.

			Linus