[linux-mm-cc] First experience of compressed cache

Nitin Gupta nitingupta910 at gmail.com
Wed Apr 2 02:46:45 EDT 2008


On Wed, Apr 2, 2008 at 11:09 AM, Nai Xia <nai.xia at gmail.com> wrote:
> Hi, Nitin,
>
>
>  On Tue, Apr 1, 2008 at 9:46 PM, Nitin Gupta <nitingupta910 at gmail.com> wrote:
>  > On Tue, Apr 1, 2008 at 7:08 PM, John McCabe-Dansted <gmatht at gmail.com> wrote:
>  >  > On Tue, Apr 1, 2008 at 9:19 PM, Nitin Gupta <nitingupta910 at gmail.com> wrote:
>  >  >  >  that page is freed. I am holding changes to fix these issues till it
>  >  >  >  at least present code
>  >  >  >  get mainline.
>  >  >
>  >  >  I understand that Linus likes small initial patches, so it seems
>  >  >  reasonable to submit to mainline before adding any more features. The
>  >  >  code has already been tested in a number of environments, under both
>  >  >  i386 and x86_64...
>  >  >
>  >  >  So what needs to be done before it is ready for mainline?
>  >  >
>  >  >
>  >  >
>  >
>  >  I already submitted compcache as kernel patch to LKML.
>  >  But I need to get more numbers on allocator performance. See:
>
>  Hi, I took a look at the allocator wiki,
>  My previous experience with Vmware gave me the impression
>  that profiling of functions' call time in it is sometimes unstable.
>  Do you want a new profiling result based on real hardware?
>

That will be helpful. Now, profiling is required for the kernel port as
I made some (minor) changes that can impact performance.

>
>  >
>  >  http://lkml.org/lkml/2008/3/24/181
>
>  The comments in this thread seemed to suggest
>  the direct comparison with kmalloc,
>  while the current wiki benchmark is based on user space.
>  If you already have the test suit for kernel,
>  I wanna have a try in my environment.
>
>

I don't yet have testing infrastructure for kernel allocator. I tried
systemtap some time back to instrument kmalloc code but that didn't
work as expected. I will now planning to use swap replay with some
additional helper mods to get these numbers. Swap Replay will also
allow us to reproduce same test results again easily.

- Nitin


More information about the linux-mm-cc mailing list