Thanks Dan And Stephen

I want to sample no. of accesses on TLB (I+D) and L2 cache misses 
simultaneously for a given process / thread.

You said Pentium4 is complex. What's about xeon (single core and quad core) as 
well as core2quad? 

I am giving some more output on Pentium4 as asked by you. Problem is  observed 
in sampling, the counter value remains  zero as shown below:
Kindly help me in fixing it.

Regards
JK

1. Without sampling:
===============

 [EMAIL PROTECTED]:~> pfmon -ePAGE_WALK_TYPE:DTMISS:ITMISS ls /dev/null
/dev/null
1907 PAGE_WALK_TYPE:DTMISS:ITMISS

[EMAIL PROTECTED]:~> pfmon -ePAGE_WALK_TYPE:ITMISS:DTMISS ls /dev/null
/dev/null
1873 PAGE_WALK_TYPE:ITMISS:DTMISS

[EMAIL PROTECTED]:~> pfmon -ePAGE_WALK_TYPE:ITMISS ls /dev/null
/dev/null
920 PAGE_WALK_TYPE:ITMISS

[EMAIL PROTECTED]:~> pfmon -ePAGE_WALK_TYPE:DTMISS ls /dev/null
/dev/null
938 PAGE_WALK_TYPE:DTMISS

2. With sampling:
============

[EMAIL PROTECTED]:~> pfmon --with-header --smpl-module=compact 
--long-smpl-periods=100000 
-eGLOBAL_POWER_EVENTS:RUNNING,PAGE_WALK_TYPE:ITMISS:DTMISS 
--short-smpl-period=100000 ls /dev/null

/dev/null
0            6924     6924  0 0xb7f1dbf8 0x000002470dd40e17   0     0          
-100000 0x0
1            6924     6924  0 0xb7f1dc00 0x000002470dd50850   0     0          
-100000 0x0
2            6924     6924  0 0xc02ac191 0x000002470dd68251   0     0          
-100000 0x0

[EMAIL PROTECTED]:~>pfmon --with-header --smpl-module=compact 
--long-smpl-periods=100000 
-eGLOBAL_POWER_EVENTS:RUNNING,PAGE_WALK_TYPE:DTMISS:ITMISS 
--short-smpl-period=100000 ls /dev/null

/dev/null
0            8892     8892  0 0xb7eb1c17 0x00000ee762d54eb6   0     0          
-100000 0x0
1            8892     8892  0 0xb7eb1c5c 0x00000ee762d64a03   0     0          
-100000 0x0
2            8892     8892  0 0xb7e21f23 0x00000ee762d7c810   0     0          
-100000 0x0


[EMAIL PROTECTED]:~> pfmon --with-header --smpl-module=compact 
--long-smpl-periods=100000 -eGLOBAL_POWER_EVENTS:RUNNING,PAGE_WALK_TYPE:DTMISS 
--short-smpl-period=100000 ls /dev/null

/dev/null
0            7419     7419  0 0xb7e85be7 0x000002b1cc68054a   0     0          
-100000 0x0
1            7419     7419  0 0xb7e85be4 0x000002b1cc68fd87   0     0          
-100000 0x0
2            7419     7419  0 0xb7df5f60 0x000002b1cc6a8460   0     0          
-100000 0x0

[EMAIL PROTECTED]:~> pfmon --with-header --smpl-module=compact 
--long-smpl-periods=100000 -eGLOBAL_POWER_EVENTS:RUNNING,PAGE_WALK_TYPE:ITMISS 
--short-smpl-period=100000 ls /dev/null

/dev/null
0            7466     7466  0 0xb7f4cbe4 0x000002c1224ed59a   0     0          
-100000 0x0
1            7466     7466  0 0xb7f4cbe4 0x000002c1224fcfa4   0     0          
-100000 0x0
2            7466     7466  0 0xc02ac191 0x000002c122514f63   0     0          
-100000 0x0






Dan Terpstra <[EMAIL PROTECTED]> wrote: What happens if you count 
PAGE_WALK_TYPE:DTMISS:ITMISS?
I suspect Stephane's right that the second counter assignment is probably
messed up. Pentium4's a beast...
- d

> -----Original Message-----
> From: [EMAIL PROTECTED] [mailto:perfmon2-devel-
> [EMAIL PROTECTED] On Behalf Of stephane eranian
> Sent: Tuesday, April 01, 2008 3:20 AM
> To: J K Rai
> Cc: [email protected]
> Subject: Re: [perfmon2] Problems encountered on Pentium 4
> 
> Hello,
> 
> 2008/3/27 J K Rai :
> >
> > Problems encountered while using two events together:
> > ========================================================
> >
> **************************************************************************
> ****************************
> > The events mentioned first gets the count values while the next one
> remains
> > zero always as shown below
> >
> **************************************************************************
> ****************************
> >
> > session output:
> > ==============
> > [EMAIL PROTECTED]:~> pfmon -ePAGE_WALK_TYPE:DTMISS,PAGE_WALK_TYPE:ITMISS ls
> > /dev/null
> > /dev/null
> > 967 PAGE_WALK_TYPE:DTMISS
> >   0 PAGE_WALK_TYPE:ITMISS
> >
> Could you try measuring ONLY the second event instead? It may be a bug
> in libpfm assignment of events to counters. That would
> not surprise me on Pentium 4, it is so complicated.
> 
> -------------------------------------------------------------------------
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketpla
> ce
> _______________________________________________
> perfmon2-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/perfmon2-devel



       
---------------------------------
You rock. That's why Blockbuster's offering you one month of Blockbuster Total 
Access, No Cost.
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
perfmon2-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/perfmon2-devel

Reply via email to