[openib-general] Re: cycles_to_units is incorrect in rdma_lat, rdma_bw.

[Michael S. Tsirkin]

Quoting r. Shirley Ma <xma at us.ibm.com>:
> Subject: Re: cycles_to_units is incorrect in rdma_lat, rdma_bw.
> 
> > in case someone's firmware is less accurate than HP's.
> > Shirley, did the patch to get_clock.c fix the problem?
> 
> Cycles is not equal to the clock rate. If you have a processor with internal
> doubling or pipelining or whatever, the clock rate isn't the same.

The name get_cycles was chosen to match linux kernel API.
(get_cycles / cpu_mhz) shall return time in microseconds.
If it doesnt get_cycles can just be fixed/renamed.

Do you see an actual machine where thats not the case?

> > Maybe there shall be a separate utility to compare
> > get_cycles and gettimeofday results, may be useful for debug.
> >
> > Anyway, if get_cpu_mhz isnt reliable people can just dump raw
> > cycles data and do the math outside the tool.
> 
> Agree if the rdma_lat, and rdma_bw are only used for debugging/regression
> testing not for performance measurement.

Its just a work-around in case someone has a problem.
But no one complained yet.

-- 
MST