linux-mips
[Top] [All Lists]

Re: [PATCH 0/3] MIPS vdso and signal delivery optimization (v2)

To: David Daney <ddaney@caviumnetworks.com>
Subject: Re: [PATCH 0/3] MIPS vdso and signal delivery optimization (v2)
From: Ralf Baechle <ralf@linux-mips.org>
Date: Fri, 19 Feb 2010 14:57:27 +0100
Cc: linux-mips@linux-mips.org
In-reply-to: <1266538385-29088-1-git-send-email-ddaney@caviumnetworks.com>
Original-recipient: rfc822;linux-mips@linux-mips.org
References: <1266538385-29088-1-git-send-email-ddaney@caviumnetworks.com>
Sender: linux-mips-bounce@linux-mips.org
User-agent: Mutt/1.5.20 (2009-08-17)
On Thu, Feb 18, 2010 at 04:13:02PM -0800, David Daney wrote:

> Before:
> n64 - Signal handler overhead: 14.517 microseconds
> n32 - Signal handler overhead: 14.497 microseconds
> o32 - Signal handler overhead: 16.637 microseconds
> 
> After:
> 
> n64 - Signal handler overhead: 7.935 microseconds
> n32 - Signal handler overhead: 7.334 microseconds
> o32 - Signal handler overhead: 8.628 microsecond

On a 180MHz 2 CPU single-node IP27:

Before:
Signal handler installation: 3.524 microseconds
Signal handler overhead: 37.009 microseconds
Protection fault: 4.264 microseconds

After:
Signal handler installation: 3.536 microseconds
Signal handler overhead: 14.331 microseconds
Protection fault: 3.600 microseconds

Everything meassured with very ancient O32 lmbench 2-alpha11 binaries.
IP27 has processors in separate packages so the cache-to-cache overhead
and thus the speedup is much higher than you have observed.

In 2.4 we used to have drastically better signal latency numbers btw.  I
wonder where all that performance went down the drain.

  Ralf

<Prev in Thread] Current Thread [Next in Thread>