Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

I have compiled the data I have so far, and placed it, as well as ministat(1) plots and graphs (boxplots) generated with the R language. All of this is available at . In addition to the real time & CPU times, the above-referenced data inclue all of

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Tue, Oct 26, 2010 at 10:34:08AM -0400, Mike Tancsa wrote: > ... > ULE is the default in 7 as well. Perhaps remove some of the kernel > options not in 7, that are in 8 by default? What is the disk > subsystem ? just ata ? > > They seem innocuous enough, but worth a try > > optionsHWP

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

At 03:36 PM 10/27/2010, Mike Tancsa wrote: At 01:05 PM 10/27/2010, Mike Tancsa wrote: At 12:34 PM 10/27/2010, David Wolfskill wrote: * release/7.1.0, with the following merged in: r186860 from stable/7 r190970 from stable/7 r203072 from head r209964 from stable/7 and using the MAC ke

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

At 01:05 PM 10/27/2010, Mike Tancsa wrote: At 12:34 PM 10/27/2010, David Wolfskill wrote: * release/7.1.0, with the following merged in: r186860 from stable/7 r190970 from stable/7 r203072 from head r209964 from stable/7 and using the MAC kernel config * stable/8 @r214029 using the G

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On 10/27/10 13:19, David Wolfskill wrote: >> note 2x drop in performance between outer and inner tracks. > > OK, but I'm not sure how that's likely to work for a multi-spindle RAID > 0 group Unless the RAID controller is trying to be overly smart (i.e. plays with fire) by somehow alternating

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Wed, Oct 27, 2010 at 01:06:18PM +0200, Ivan Voras wrote: > On 10/27/10 12:55, David Wolfskill wrote: > > > That *is* a problem, as I cannot justify a migration to a branch > > of FreeBSD that imposes about a 23% penalty in elapsed time on this > > workload. I want folks at work to have more re

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On 10/27/10 12:55, David Wolfskill wrote: > That *is* a problem, as I cannot justify a migration to a branch > of FreeBSD that imposes about a 23% penalty in elapsed time on this > workload. I want folks at work to have more reason to want to use > (newer branches of) FreeBSD, not less. That is

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Wed, Oct 27, 2010 at 11:54:07AM +0200, Ivan Voras wrote: > ... > > the 8.x reference machine, and each terminated with a status code of 0: > > > > startstopreal usersys os > > 128867 1288111298 131.14 12.77 17.88 7.1-R+ > > > 1288109542 1288109653 111.26 12.

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On 10/26/10 19:45, David Wolfskill wrote: > On Tue, Oct 26, 2010 at 02:03:34PM +0200, Ivan Voras wrote: >> ... >> Since you now have the two kernels readily available, can you rule out >> NFS by just repeating the step which involves it in both kernels and >> compare the results? > > On Tue, Oct 2

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Tue, Oct 26, 2010 at 02:03:34PM +0200, Ivan Voras wrote: > ... > Since you now have the two kernels readily available, can you rule out > NFS by just repeating the step which involves it in both kernels and > compare the results? On Tue, Oct 26, 2010 at 02:47:11PM +0200, Ivan Voras wrote: > ...

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Tue, Oct 26, 2010 at 10:34:08AM -0400, Mike Tancsa wrote: > At 07:29 AM 10/26/2010, David Wolfskill wrote: > > >OK -- but we were using the default scheduler in each case. The basic > >point I'm making here is the apparent performance regression for > >similarly-configured systems under 7.1 vs

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

At 07:29 AM 10/26/2010, David Wolfskill wrote: OK -- but we were using the default scheduler in each case. The basic point I'm making here is the apparent performance regression for similarly-configured systems under 7.1 vs. 8.1. ULE is the default in 7 as well. Perhaps remove some of the k

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Mon, Oct 25, 2010 at 10:09:53PM -0700, Julian Elischer wrote: > ... > >It appears to me that the last test runs show results that are just > >about identical to the "native" 8.1-S kernel+userland, so if I > >understand the logic correctly, that appears to implicate something in > >the 8.1-S kern

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On 10/25/10 9:55 AM, David Wolfskill wrote: On Wed, Oct 20, 2010 at 04:50:10PM -0700, Julian Elischer wrote: ... try the 7.x machine but running the 8.x kernel.. i.e. change nothing, but boot the new kernel. ... OK; here are results of previous tests, along with the above. As noted earlier, I

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Wed, Oct 20, 2010 at 04:50:10PM -0700, Julian Elischer wrote: > ... > try the 7.x machine but running the 8.x kernel.. i.e. change nothing, > but boot the new kernel. > ... OK; here are results of previous tests, along with the above. As noted earlier, I needed to set the UNAME_r environmant

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Fri, Oct 22, 2010 at 05:07:52PM -0700, Julian Elischer wrote: > On 10/22/10 4:48 PM, David Wolfskill wrote: > >On Fri, Oct 22, 2010 at 04:17:30PM -0700, David Wolfskill wrote: > >>On Wed, Oct 20, 2010 at 04:50:10PM -0700, Julian Elischer wrote: > >>>... > >>>try the 7.x machine but running the 8

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On 10/22/10 4:48 PM, David Wolfskill wrote: On Fri, Oct 22, 2010 at 04:17:30PM -0700, David Wolfskill wrote: On Wed, Oct 20, 2010 at 04:50:10PM -0700, Julian Elischer wrote: ... try the 7.x machine but running the 8.x kernel.. i.e. change nothing, but boot the new kernel. I just started that t

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Fri, Oct 22, 2010 at 04:17:30PM -0700, David Wolfskill wrote: > On Wed, Oct 20, 2010 at 04:50:10PM -0700, Julian Elischer wrote: > > ... > > try the 7.x machine but running the 8.x kernel.. i.e. change nothing, > > but boot the new kernel. > > I just started that test, to run over the weekend.

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Wed, Oct 20, 2010 at 04:50:10PM -0700, Julian Elischer wrote: > ... > try the 7.x machine but running the 8.x kernel.. i.e. change nothing, > but boot the new kernel. I just started that test, to run over the weekend. I did run another test (3 iterations) with the 7.1 OS (kernel+userland) on

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On 10/21/10 23:53, Dan Nelson wrote: > In the last episode (Oct 20), David Wolfskill said: >> Almost 2 years ago, we migrated from a lightly-patched 6.2-R to 7.1-R with >> 5 commits that were made to 7.1-S backported to it. On the same hardware >> (not the HP mentioned above), I measured a 35% red

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On 22 October 2010 06:42, David Wolfskill wrote: >> Julian's suggestion of booting the 8.1 kernel on the 7.1 OS will definitely >> narrow down the list of suspects. > > I'll see about doing something along those lines, but I doubt it will be > all that helpful, actually. It narrows down the susp

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On Thu, Oct 21, 2010 at 04:53:30PM -0500, Dan Nelson wrote: > ... > An observation: on 8.1, both user and sys times are less, but real time is > higher. So 8.1 finished the build using less CPU, but spent more time > waiting for something else. Interesting; thanks for pointing that out. > Disk?

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

In the last episode (Oct 20), David Wolfskill said: > Almost 2 years ago, we migrated from a lightly-patched 6.2-R to 7.1-R with > 5 commits that were made to 7.1-S backported to it. On the same hardware > (not the HP mentioned above), I measured a 35% reduction in elapsed time > for one particula

Re: Possible evidence of performance regression for 8.1-S (vs. 7.1)

On 10/20/10 10:48 AM, David Wolfskill wrote: [...] The 8.x reference machine was created by cloning the 7.x reference machine (the OS "drive" is a RAID 1; I broke the mirror and physically booted the (soon-to-be) 8.x machine from a single drive from the 7.x mirror, changed the hostname& IP ad

Possible evidence of performance regression for 8.1-S (vs. 7.1)

At work, my focus is on facilitating & improving the productivity of the developers. Much of the time they spend is in waiting for a build to complete -- thus, something that reduces the time thus spent is likely to be beneficial (all other things being approximately equal), while something that i