Re: Is -falign-loops=16 ignored sometimes?
Re: Is -falign-loops=16 ignored sometimes?
- Subject: Re: Is -falign-loops=16 ignored sometimes?
- From: Rustam Muginov <email@hidden>
- Date: Mon, 16 Apr 2007 19:04:21 +0400
I remember i had this problem on PowerMac G5/Panther with gcc 3.3 I had to add another flag, -falign-loops-max-skip=15, and this generally solved the problem
-- Sincerely, Rustam Muginov
On 16.04.2007, at 18:58, Rich Siegel wrote: [Grr, sorry about the empty message. Not enough caffeine.]
This may end up being a question for the performance list, but since it appears to involve a compiler behavior, I thought I'd ask here first.
On i386, are there known circumstances in which -falign-loops=16 is ignored? The reason I ask is because a particular hotspot function of mine is taking a big hit for an unaligned loop (according to Shark), even though the file is compiled with -falign-loops=16 (I double-checked the build transcript to verify this).
Or is 16-byte loop alignment inappropriate for i386 and I should use a 32-byte alignment? And what are the implications of this for PPC code?
Thanks for any advice,
R.
|
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden