Home » Mailing list archives » linux-kernel » 2008 » August » 15

Re: pthread_create() slow for many threads; also time to revisit 64b context switch optimization?

!MAILaRCHIVE_VOTE_RePLACE
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
[view in full thread]
From: Andi Kleen <andi@...>
To: Ingo Molnar <mingo@...>
Cc: Andi Kleen <andi@...>, Ulrich Drepper <drepper@...>, Arjan van de Ven <arjan@...>, <akpm@...>, <hugh@...>, <linux-mm@...>, <linux-kernel@...>, <briangrant@...>, <cgd@...>, <mbligh@...>, Linus Torvalds <torvalds@...>, Thomas Gleixner <tglx@...>, H. Peter Anvin <hpa@...>
Subject: Re: pthread_create() slow for many threads; also time to revisit 64b context switch optimization?
Date: Friday, August 15, 2008 - 9:33 am

On Fri, Aug 15, 2008 at 02:43:50PM +0200, Ingo Molnar wrote:

I used them as 64bit machines and they weren't painful at all.


MAP_32BIT was not actually added for this originally. It 
was originally added for the X server's old dynamic loader, which
needed 2GB memory.

It's main failing, which I freely admit, was to not call it MAP_31BIT.


Not sure what the semantics of that would be. For me it would
seem ugly to hardcode specific semantics in the kernel for this
("mechanism not policy")

But for most possible semantics I can think of the data structure would still 
need to be fixed I think.


IMHO the correct solution is to fix the data structure to not have such
a bad complexity in this corner case. We typically do this for all
other data structures as we discover such cases. No reason the VMAs
should be any different. 

-Andi

--
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
Messages in current thread:
Re: pthread_create() slow for many threads; also time to rev..., Pardo, (Tue Aug 12, 8:45 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Wed Aug 13, 6:44 am)
Re: pthread_create() slow for many threads; also time to rev..., Arjan van de Ven, (Wed Aug 13, 9:35 am)
Re: pthread_create() slow for many threads; also time to rev..., Ulrich Drepper, (Wed Aug 13, 10:21 am)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Wed Aug 13, 10:25 am)
Re: pthread_create() slow for many threads; also time to rev..., Ulrich Drepper, (Wed Aug 13, 10:36 am)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Wed Aug 13, 11:10 am)
Re: pthread_create() slow for many threads; also time to rev..., Andi Kleen, (Wed Aug 13, 4:42 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Fri Aug 15, 8:43 am)
Re: pthread_create() slow for many threads; also time to rev..., Andi Kleen, (Fri Aug 15, 9:33 am)
Re: pthread_create() slow for many threads; also time to rev..., Andrew Morton, (Wed Aug 13, 4:56 pm)
Re: pthread_create() slow for many threads; also time to rev..., Andi Kleen, (Wed Aug 13, 5:46 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ulrich Drepper, (Wed Aug 13, 11:21 am)
Re: pthread_create() slow for many threads; also time to rev..., H. Peter Anvin, (Wed Aug 13, 12:05 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Wed Aug 13, 11:40 am)
Re: pthread_create() slow for many threads; also time to rev..., Ulrich Drepper, (Wed Aug 13, 11:55 am)
Re: pthread_create() slow for many threads; also time to rev..., Linus Torvalds, (Wed Aug 13, 1:09 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ulrich Drepper, (Wed Aug 13, 2:04 pm)
Re: pthread_create() slow for many threads; also time to rev..., Arjan van de Ven, (Wed Aug 13, 2:16 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ulrich Drepper, (Wed Aug 13, 2:22 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Wed Aug 13, 12:02 pm)
Re: pthread_create() slow for many threads; also time to rev..., Jamie Lokier, (Fri Aug 15, 11:54 am)
Re: pthread_create() slow for many threads; also time to rev..., Ulrich Drepper, (Fri Aug 15, 1:13 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Fri Aug 15, 1:19 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ulrich Drepper, (Fri Aug 15, 1:23 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Fri Aug 15, 3:00 pm)
Re: pthread_create() slow for many threads; also time to rev..., Ingo Molnar, (Fri Aug 15, 12:03 pm)