jeffr_tech ([info]jeffr_tech) wrote,
I have updated my scaling results with Linux 2.6.21.5 and glibc 2.6. You can see them at http://people.freebsd.org/~jeff/sysbench.png.

This also has results comparing the old FreeBSD scheduler with no affinity, SCHED_4BSD. SCHED_ULE is the version of ULE that doesn't use per-cpu locks. And SCHED_SMP is actually ULE with per-cpu locks and various other tuning. I actually have had better results than those on the graph but I don't feel like regenerating it again right now.

Linux has definitely improved but they still have some significant problems. I will try with tcmalloc later.

  • Post a new comment

    Error

    Your reply will be screened

    Your IP address will be recorded 

  • 26 comments

Anonymous

June 19 2007, 03:28:06 UTC 4 years ago

Possible problem

Assuming the glibc you are using is doing the optimal malloc thing now, you
still need the following patch to the kernel which AFAIK is not in 2.6.21.x.

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff_plain;h=0a27a14a62921b438bb6f33772690d345a089be6

If you don't know whether your glibc is doing the right thing -- well if the
above patch changes anything, then it is going through that path, and if not
then it is not :)

[info]jeffr_tech

June 19 2007, 04:34:02 UTC 4 years ago

Re: Possible problem

The numbers changed slightly but it's hard to say because there is significant variance.

I don't believe this is the root cause. It looks like the scheduler is simply not keeping all of the processors busy.

Anonymous

June 19 2007, 04:42:09 UTC 4 years ago

Re: Possible problem

>The numbers changed slightly but it's hard to say because there is
>significant variance.

Then I'd guess that your glibc is still using the old path for allocation /
freeing.


>I don't believe this is the root cause. It looks like the scheduler
>is simply not keeping all of the processors busy.

That's what it looks like because there is significant contention on the
mmap_sem semaphore which is shared among all threads. The patches fix that.

[info]jeffr_tech

June 19 2007, 04:49:28 UTC 4 years ago

Re: Possible problem

I downloaded glibc 2.6 from ftp.gnu.org. How do I verify whether or not this is using the 'new' allocation routines.

Anonymous

June 19 2007, 04:53:42 UTC 4 years ago

Re: Possible problem

I'm not completely sure from the version number -- I'm not involved with glibc
development too much.

You could strace MySQL daemon and see whether it is making use of
madvise(MADV_DONTNEED) or if it is doing a lot of mmap() calls. The former
is what you want.

[info]jeffr_tech

June 19 2007, 04:59:34 UTC 4 years ago

Re: Possible problem

The malloc implementation is using MADV_DONTNEED to shrink the heap.

I have uploaded a new graph with some debugging support removed that was built in to the kernel.

Anonymous

June 19 2007, 05:15:49 UTC 4 years ago

Re: Possible problem

Weird. It looks very different to the results I get, that has Linux
following a similar curve to your FreeBSD results.

http://www.thisishull.net/showpost.php?s=5d2bfa8b5a0707286a86d7c57a2c6308&p=1010222&postcount=2

sysbench transactions per sec (higher is better)

kernel is 2.6.21
threads   unpatched tps   patched tps
1,        453             476
2,        831             871
4,       1468            1529
8,       2139            2235
16,      2118            2177
32,      1051            2120
64,       917            1949

With the patched kernel, you see its only losing about 5% from peak at 8
threads to 32 threads (which looks roughly in line with what you see).

OTOH, I was testing with a patched 2.4 glibc IIRC, so maybe something else
is happening in 2.6. I'll have to retest it.

[info]jeffr_tech

June 19 2007, 05:20:35 UTC 4 years ago

Re: Possible problem

Yes, that does look consistent with my findings. Can I see the patch you have for 2.4? I could see how the diff looks against the 2.6 sources I have.

Anonymous

June 19 2007, 05:26:23 UTC 4 years ago

Re: Possible problem

I was using Jakub's uploaded glibc from this post (sorry, it was
2.5.x, not 2.4)

http://www.ussg.iu.edu/hypermail/linux/kernel/0704.2/2064.html

The rpms appear to no longer be available, but I assume his patch
is the only deviation from the glibc cvs at the time.

[info]jeffr_tech

June 19 2007, 05:35:03 UTC 4 years ago

Re: Possible problem

Thanks, it looks like most of this patch went in, but the chunk that tries MADV_FREE is not in. The rest of the accounting was checked in. I'm testing that now.

Is this Nick Piggin, btw?

[info]jeffr_tech

June 19 2007, 05:40:49 UTC 4 years ago

Re: Possible problem

This did not seem to make a difference. There may be some other contention that is masking the effects of this patch. How do I see what resources we're contending on in linux? Is there some lock profiling or tracing?

Anonymous

4 years ago

[info]lvchanel

October 25 2010, 02:50:19 UTC 1 year ago

Re: Possible problem

I Love gucci and chanel.
Some women do not mind buying replica handbags, while some women just love designer handbags if you can distinguish between good and bed from the replica handbag,you also can use low price get high quality enjoy,today use chanel handbag ,tomorrow carry gucci handbag,the day after tomorrow hermes handbad in your hand, this niceness all give the credit to low price,same argument you also can buy replica watchesbuy DVDs louis vuitton...

[info]lvchanel

October 25 2010, 02:47:47 UTC 1 year ago

Re: Possible problem

I Love gucci and chanel.
Some women do not mind buying replica handbags, while some women just love designer handbags if you can distinguish between good and bed from the replica handbag,you also can use low price get high quality enjoy,today use chanel handbag ,tomorrow carry gucci handbag,the day after tomorrow hermes handbad in your hand, this niceness all give the credit to low price,same argument you also can buy replica watchesbuy DVDs louis vuitton...

Anonymous

June 21 2007, 22:05:42 UTC 4 years ago

new Linux scheduler

Jeff, what do you say about new Linux scheduler, please ? -- http://kerneltrap.org/node/8059

[info]lvchanel

October 25 2010, 02:49:54 UTC 1 year ago

I Love gucci and chanel.
Some women do not mind buying replica handbags, while some women just love designer handbags if you can distinguish between good and bed from the replica handbag,you also can use low price get high quality enjoy,today use chanel handbag ,tomorrow carry gucci handbag,the day after tomorrow hermes handbad in your hand, this niceness all give the credit to low price,same argument you also can buy replica watchesbuy DVDs louis vuitton...

[info]lvchanel

October 25 2010, 02:47:35 UTC 1 year ago

I Love gucci and chanel.
Some women do not mind buying replica handbags, while some women just love designer handbags if you can distinguish between good and bed from the replica handbag,you also can use low price get high quality enjoy,today use chanel handbag ,tomorrow carry gucci handbag,the day after tomorrow hermes handbad in your hand, this niceness all give the credit to low price,same argument you also can buy replica watchesbuy DVDs louis vuitton...

[info]pjswb2005

December 15 2010, 06:55:55 UTC 1 year ago

This is a really good read for me, Must admit that you are one of the best bloggers I ever saw.Thanks for posting this informative article.
hermes handbags Coach handbags hermes birkin handbags hermes purses discount hermes handbags hermes kelly handbags hermes lindy handbags hermes wallets hermes birkin bags burberry handbags hermes handbag coach handbags

[info]eniyisi

January 6 2011, 21:13:13 UTC 1 year ago

porno ~
porno ~
sikiş ~
sikiş ~
videofoni ~
vpills ~
azdırıcı damla

[info]kadirboz

January 8 2011, 19:35:04 UTC 1 year ago

ritimmm

Tek tek havayolu şirketlerinin sayfalarına girmenize gerek yok uçak bileti satın almak istiyorsanız sitemizi ziyaret ediniz.
Create an Account
Forgot your login or password?
Facebook Twitter More login options
English • Español • Deutsch • Русский…