[ofa-general] Problem while running ib tests

[Jason Gunthorpe]

On Wed, Sep 23, 2009 at 08:32:20AM -0700, Ira Weiny wrote:
> On Wed, 23 Sep 2009 09:11:49 -0600
> Jason Gunthorpe <jgunthorpe at obsidianresearch.com> wrote:
> 
> > On Wed, Sep 23, 2009 at 12:01:02PM +0300, Moni Shoua wrote:
> > 
> > > > If I try to run any IB bandwidth test or latency test it end us with warning
> > > > "Conflicting CPU frequency values detected: 2394.000000 != 1596.000000".
> > 
> > > One more option to solve it (besides the -F) is to disable the power
> > > saving of the CPU.  In the power saving mode not all CPUs operate at
> > > the same Hz/sec rate (cycle rate) and this is the reason for the
> > > conflict.  Disabling the power saving mode is done in the BIOS and
> > > it is sometimes called economy mode or performance mode (the other
> > > side of the trade off)
> > 
> > This whole thing should just be removed. x86-64 linux has a VDSO
> > clock_gettime(CLOCK_MONOTONIC) that is fast, doesn't trap into the
> > kernel, and doesn't suffer from these problems.
> 
> What about other architectures?  I would be willing to do a quick patch but...

It varies. 32 bit x86 still doesn't have a VDSO for timekeeping for
some reason.. s390 and ppc are OK.

But this is a fluffy diagnostic - does it matter? More people seem to
get tripped up by the wonky way to keep time than I've ever seen it
benefit.

Jason