Re: Armel userspace performance regression

2007-08-15 Thread Daniel Jacobowitz
On Wed, Aug 15, 2007 at 11:25:54AM +0300, Eugene Sanivsky wrote: > Previous throughput measured on custom rootfs using > GLIBC2.3.5pre/GCC3.4.4 (CS2005q3-2 Provided by CodeSourcery). I don't really want to toot our own horn here, but you're comparing a generic Debian compiler - focus on correctnes

Re: Armel userspace performance regression

2007-08-15 Thread Riku Voipio
On Wed, Aug 15, 2007 at 11:25:54AM +0300, Eugene Sanivsky wrote: >I am experiencing serious performance regression with new armel >libc/compiler. >I see up to 45% degradation in Samba throughput for example. Samba is a very complex piece of software, and you have changed every single

Re: Armel userspace performance regression

2007-08-15 Thread Lennert Buytenhek
On Wed, Aug 15, 2007 at 11:25:54AM +0300, Eugene Sanivsky wrote: > Hi, Hi, > I am experiencing serious performance regression with new armel > libc/compiler. > > I see up to 45% degradation in Samba throughput for example. > > Previous throughput measured on custom rootfs using > GLIBC2.3.5pr

Armel userspace performance regression

2007-08-15 Thread Eugene Sanivsky
Hi, I am experiencing serious performance regression with new armel libc/compiler. I see up to 45% degradation in Samba throughput for example. Previous throughput measured on custom rootfs using GLIBC2.3.5pre/GCC3.4.4 (CS2005q3-2 Provided by CodeSourcery). Same Linux kernel used, both