Find the answer to your Linux question:
Results 1 to 2 of 2
Hello, I have PC with 2 Gigs of RAM and I was curious if I can get performance gain disabling CONFIG_HIGHMEM and using CONFIG_VMSPLIT_2G_OPT . I have tested performance by ...
Enjoy an ad free experience by logging in. Not a member yet? Register.
  1. #1
    Just Joined!
    Join Date
    May 2009

    Is it possible, that 32-bit kernel with HIGHMEM is faster?

    I have PC with 2 Gigs of RAM and I was curious if I can get performance gain disabling CONFIG_HIGHMEM and using CONFIG_VMSPLIT_2G_OPT. I have tested performance by running mplayer with some HD trailers (GPU is old Radeon with OSS drivers - only Xv acceleration). From results I see, that kernel with HIGHMEM is slightly faster. I know, I haven't used much memory only by running mplayer alone, but I have expected, that every access to memory will be slower with HIGHMEM. I don't understand, why is HIGHMEM faster, but almost all articles about CONFIG_VMSPLIT_* and CONFIG_PAGE_OFFSET are from year 2006.

    CPU is Intel P4 2.80GHz with disabled HT (kernel has not enabled CONFIG_SMP). Memory is 4x 512 MB dual-channel 400MHz CL3. Used kernel was 3.2.1 with CONFIG_TRANSPARENT_HUGEPAGE_ALWAYS.

    Diff of config:
    -# CONFIG_NOHIGHMEM is not set
    +# CONFIG_HIGHMEM4G is not set
     # CONFIG_HIGHMEM64G is not set
    +# CONFIG_VMSPLIT_3G is not set
    +# CONFIG_VMSPLIT_3G_OPT is not set
    +# CONFIG_VMSPLIT_2G is not set
    +# CONFIG_VMSPLIT_1G is not set
    $ mplayer -fs -nosound -benchmark
    Movie (codec) HIGHMEM (-vo null) VMSPLIT_2G_OPT (-vo null) VMSPLIT performance % (-vo null) HIGHMEM (-vo xv) VMSPLIT_2G_OPT (-vo xv) VMSPLIT performance % (-vo xv)
    Avatar 1080p (ffh264) 111.479s 114.822s 97.0% 123.826s 126.070s 98.2%
    nVidia PureVideo 1080p (wmv9dmo) 165.938s 166.637s 99.6% 181.632s 184.528s 98.4%
    Star Trek 11 1080p (wmv9dmo) 69.056s 68.384s 101.0% 78.598s 80.052s 98.2%

  2. #2
    Linux Guru Rubberman's Avatar
    Join Date
    Apr 2009
    I can be found either 40 miles west of Chicago, in Chicago, or in a galaxy far, far away.
    This sounds like a good question for Intel and nVidia engineers. I doubt that any of us here have the expertise required to answer your well-put question. In my case, the only rational question is "is it fast enough", and the only rational answer is either "yes" or "no".
    Sometimes, real fast is almost as good as real time.
    Just remember, Semper Gumbi - always be flexible!

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts