Re: [openib-general] [PATCH] udapl: PPC64 cpuinfo change

2005-10-07 Thread Grant Grundler
On Thu, Oct 06, 2005 at 11:01:21PM -0500, Troy Benjegerdes wrote:
 Oh boy is there some reason 'gettimeofday' does not work?

In general, it doesn't work as well.

 Trying to infer timebase/clock/rtsc frequency is going to be a mess.

Using cycle counters is quite portable today and provides
accurate results (with caveats on it's use). I'm open to
using the next best thing once it's clear the cycle counters
do NOT work.

 Think cpus that dynamically change frequency.. Laptops do now..
 how long before something with infiniband does and breaks this
 code horribly? (think embedded systems)

I don't buy this argument. Most of the tests load the CPU and it
essentially runs at a fixed frequency.

A better argument is how to benchmark under virtualized environment.
I think that is totally broken today regardless of what method
one uses to measure time.

 There are a couple of implementations of gettimeofday fully in userspace
 that hide the details and still read the high-res hardware counters. Google
 for 'vDSO gettimeofday'.

Well, I'm sure Michael is open to patches on this for userspace/perftest
stuff and like wise for James Lentini for uDAPL.

grant
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general


[openib-general] [PATCH] udapl: PPC64 cpuinfo change

2005-10-06 Thread Todd Bowman
This patch in addition to PPC64 atomic function additions provides udapl support on PPC64 platform.

/proc/cpuinfo on PPC64 prints different label for processor speed.

Signed-off-by: Todd Bowman [EMAIL PROTECTED]

Index: userspace/dapl/test/dapltest/mdep/linux/dapl_mdep_user.c
===
--- userspace/dapl/test/dapltest/mdep/linux/dapl_mdep_user.c (revision 3547)
+++ userspace/dapl/test/dapltest/mdep/linux/dapl_mdep_user.c (working copy)
@@ -186,7 +186,12 @@
 void )
{
#define DT_CPU_MHZ_BUFFER_SIZE 128
+
+#if defined (__PPC64__)
+#define DT_CPU_MHZ_MHZ clock
+#else
#define
DT_CPU_MHZ_MHZ
cpu MHz
+#endif
#define DT_CPU_MHZ_DELIMITER :

 FILE *fp;

___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general

Re: [openib-general] [PATCH] udapl: PPC64 cpuinfo change

2005-10-06 Thread Grant Grundler
On Thu, Oct 06, 2005 at 11:48:02AM -0600, Todd Bowman wrote:
 /proc/cpuinfo on PPC64 prints different label for processor speed.
...

ISTR the clock value in cpuinfo is NOT the same as the CPU MHz.
Can you remind me if clock value * mtfb results in
wall clock time units?

If not, then use of DT_CPU_MHZ_MHZ needs to be reviewed since
it typically makes that assumption.

Also, if someone cares about sparc (hey Tom! :^) ),
then might leverage the get_clock.c code on:
http://svn.gnumonks.org/trunk/mmio_test/


hth,
grant
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general


Re: [openib-general] [PATCH] udapl: PPC64 cpuinfo change

2005-10-06 Thread Troy Benjegerdes
On Thu, Oct 06, 2005 at 02:14:08PM -0700, Grant Grundler wrote:
 On Thu, Oct 06, 2005 at 11:48:02AM -0600, Todd Bowman wrote:
  /proc/cpuinfo on PPC64 prints different label for processor speed.
 ...
 
 ISTR the clock value in cpuinfo is NOT the same as the CPU MHz.
 Can you remind me if clock value * mtfb results in
 wall clock time units?
 
 If not, then use of DT_CPU_MHZ_MHZ needs to be reviewed since
 it typically makes that assumption.
 
 Also, if someone cares about sparc (hey Tom! :^) ),
 then might leverage the get_clock.c code on:
   http://svn.gnumonks.org/trunk/mmio_test/

Oh boy is there some reason 'gettimeofday' does not work? Trying to
infer timebase/clock/rtsc frequency is going to be a mess. Think cpus
that dynamically change frequency.. Laptops do now.. how long before
something with infiniband does and breaks this code horribly? (think
embedded systems)

There are a couple of implementations of gettimeofday fully in userspace
that hide the details and still read the high-res hardware counters. Google
for 'vDSO gettimeofday'.
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general