Re: oltp ~10% regression with 2.6.27-rc5 on stoakley machine

[Gregory Haskins <ghaskins@novell.com>]

Hi Peter,

Peter Zijlstra wrote:
>
>
> Also, we used to have some auto-tuning in there, which dissapeared some
> time ago, gregory brought it back to live recently, perhaps he likes to
> share? :-)
>

Per your request, I brought those "revert" patches I had in the PQ branch
forward to 2.6.27-rc5.  They are completely unbuilt and untested, but here
for your experimentation.

-Greg

---

Gregory Haskins (4):
      sched: make task_hot() once again use sd->cache_hot_time
      Revert "sched: zap the migration init / cache-hot balancing code"
      Revert "[PATCH] sched: remove cache_hot_time"
      revert "sched: sched_cacheflush is now unused"


 Documentation/kernel-parameters.txt |   43 +++
 arch/ia64/kernel/setup.c            |   15 +
 arch/mips/kernel/smp.c              |   11 +
 arch/sparc/kernel/smp.c             |   10 +
 arch/sparc64/kernel/smp.c           |   27 ++
 arch/x86/kernel/smpboot.c           |   12 +
 include/asm-m32r/system.h           |   10 +
 include/asm-mips/system.h           |   10 +
 include/asm-parisc/system.h         |   11 +
 include/asm-x86/system.h            |    9 +
 include/linux/sched.h               |    7 
 include/linux/topology.h            |    1 
 kernel/sched.c                      |  499 ++++++++++++++++++++++++++++++++++-
 13 files changed, 656 insertions(+), 9 deletions(-)

-- 
Signature