[haiku-development] Re: Optimizing Painter::_DrawBitmapBilinearCopy32

  • From: Christian Packmann <Christian.Packmann@xxxxxx>
  • To: haiku-development@xxxxxxxxxxxxx
  • Date: Fri, 19 Jun 2009 19:20:23 +0200

André Braga - 2009-06-15 00:16 :

My CPU liked the SSE2 code :)

Wow. I didn't know the K10 had such a good SSE engine. I don't know if you noticed, but it's precisely as fast as my Core2/45nm per clock cycle, and faster than the first-generation Core2s. That means AMD has implemented a full 128-bit engine throughout, while Intel skimped on the early Core2s where only some parts of the SSE engine were 128bit.

While the Core2s do have an advantage of 20% in integer, the K10s performance on "multimedia" code which doesn't use newer SSE versions should be equal. Considering the low prices of the K10, this makes it a real nice performer.

I don't know if this makes a difference for this code, but this is a
AM2 board, not AM2+ (which is the native type for this CPU).

This doesn't matter at all, as the only differences (AFAIK) are in HyperTransport speed and better power savings due to split power planes. Only extremely I/O-heavy programs may see a difference, even though I can't imagine any load on a single-CPU system to saturate even the "slow" HT link.

Christian

Other related posts: