• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Re: Build slowdown with 4 processors vs. 2 or 1?
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Build slowdown with 4 processors vs. 2 or 1?


  • Subject: Re: Build slowdown with 4 processors vs. 2 or 1?
  • From: Jim Wintermyre <email@hidden>
  • Date: Fri, 8 Dec 2006 09:34:51 -0800

At 12:29 AM -0600 12/8/06, Syd Polk wrote:
On Dec 7, 2006, at 11:47 PM, Jim Wintermyre wrote:

Here's a weird one. In one of my projects, when I build in XCode on the Mac Pro, it grinds to a halt after 4 files. There is a flurry of disk activity, and hardly any CPU activity. Top seems to indicate a lot of swapping going on. This goes on for around a couple minutes. What's weird is that if I set the number of processors to 2 or 1, the problem goes away. The entire project builds in less than 20 seconds, and the CPU load is pegged pretty much the entire time, which is what I'd expect. Any ideas what could be going on here? Any settings I could tweak?

Thanks,
Jim

How much RAM do you have? You need double the physical RAM for 4 compiles as your need for two.

1 GB. The weird thing is that this is only happening with *one* of my plugin projects. All the other ones are fine w/4 processors enabled. They all use the same precompiled header, and are all set up the same; the only difference is the actual plugin source files.


At 10:39 PM -0800 12/7/06, Chris Espinosa wrote:
We see this sometimes with processor scaling in Mac OS X. gcc maps in the precompiled header into each process as a private write file, then writes to it, forcing the kernel to do a copy-on-write operation each time. If you have 4 copies of gcc running simultaneously all mapping in the same precompiled header file and trying to write to private copies of it, the kernel can get saturated with the copy-on-write requests.

So after all that work to get your precompiled headers coherent, you may be amazed to find that that's the cause of the slowdown when scaling up to more processors.

Would this still make sense in light of the fact that this is not happening in any of my other plugin projects, which all use the same precompiled header?


Thanks,
Jim
_______________________________________________
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


References: 
 >Build slowdown with 4 processors vs. 2 or 1? (From: Jim Wintermyre <email@hidden>)
 >Re: Build slowdown with 4 processors vs. 2 or 1? (From: Syd Polk <email@hidden>)

  • Prev by Date: Re: Illegal byte sequence
  • Next by Date: Re: Illegal byte sequence
  • Previous by thread: Re: Build slowdown with 4 processors vs. 2 or 1?
  • Next by thread: Re: Build slowdown with 4 processors vs. 2 or 1?
  • Index(es):
    • Date
    • Thread