Hello, On Fri, Apr 4, 2008 at 6:12 AM, J K Rai <[EMAIL PROTECTED]> wrote: > Thanks Stephane > > What I mean by cumulative values is the values of additional counters / > events output by pfmon while sampling. > > As shown below in last column: > > ~> pfmon --with-header --smpl-module=compact --long-smpl-periods=100000 > -eGLOBAL_POWER_EVENTS:RUNNING,PAGE_WALK_TYPE:DTMISS,PAGE_WALK_TYPE:ITMISS > --short-smpl-period=100000 ls /dev/null >
Ok, now I understand. In this case you want to add the --reset-non-smpl option and pfmon will print the delta for the other events. > /dev/null > 0 6132 6132 0 0xb7e8abe0 0x0000029137d7c537 0 0 > -100000 0x0 0x65 > 1 6132 6132 0 0xb7e8ac5c 0x0000029137d8c557 0 0 > -100000 0x0 0x65 > 2 6132 6132 0 0xb7e48250 0x0000029137da989e 0 0 > -100000 0x0 0xc3 > 3 6132 6132 0 0xb7e04ae3 0x0000029137dbbeba 0 0 > -100000 0x0 0xe4 > 4 6132 6132 0 0xb7e0380a 0x0000029137dcdb30 0 0 > -100000 0x0 0x11b > 5 6132 6132 0 0xb7dff171 0x0000029137de411f 0 0 > -100000 0x0 0x172 > 6 6132 6132 0 0xb7e03e20 0x0000029137df8280 0 0 > -100000 0x0 0x195 > 7 6132 6132 0 0xb7e054d6 0x0000029137e09a14 0 0 > -100000 0x0 0x1c2 > 8 6132 6132 0 0xb7dbf695 0x0000029137e1d847 0 0 > -100000 0x0 0x1f3 > 9 6132 6132 0 0xb7dbf8c1 0x0000029137e30bfa 0 0 > -100000 0x0 0x20e > 10 6132 6132 0 0xb7e0381c 0x0000029137e45233 0 0 > -100000 0x0 0x243 > 11 6132 6132 0 0xb7db957e 0x0000029137e59276 0 0 > -100000 0x0 0x25e > 12 6132 6132 0 0xb7f16aee 0x0000029137e6d097 0 0 > -100000 0x0 0x291 > 13 6132 6132 0 0xb7dba9b0 0x0000029137e8274a 0 0 > -100000 0x0 0x2d1 > 14 6132 6132 0 0xb7dff506 0x0000029137ed8138 0 0 > -100000 0x0 0x357 > 15 6132 6132 0 0xb7f0fcdd 0x0000029137eecd40 0 0 > -100000 0x0 0x3c4 > # total samples : 16 > # total buffer overflows : 0 > > Can we get the values of these additional events / counters for that > sampling period only instead of cumulative one. The first event is for > deciding sampling period and rest of the events are additional, which I want > to sample for the sampling period. > > Regards > JK > > > > stephane eranian <[EMAIL PROTECTED]> wrote: > JK, > > > > > On Wed, Apr 2, 2008 at 3:23 PM, J K Rai wrote: > > Hi all > > > > It seems that sampling option with pfmon gives the cumulative values of > the > > events to be sampled. > > > You mean it gives you the IP-based profiles at the end. Samples are > aggregated > by IP-address. > > > Can we have it non cumulative i.e. like the no. of event observed during > > that sampling period. > > > You can have a sequential dump of the samples if you use a different > sampling > module. > > $ pfmon --smpl-module=compact --long-smpl-periods=2400000 my_test_program > > 0 19069 19069 1 0x400701 0x000020bc6cfe559d 17 0 > -2400000 > 1 19069 19069 1 0x400701 0x000020bc6d15b975 17 0 > -2400000 > 2 19069 19069 1 0x400701 0x000020bc6d2cbb54 17 0 > -2400000 > 3 19069 19069 1 0x400701 0x000020bc6d43ccc1 17 0 > -2400000 > 4 19069 19069 1 0x400701 0x000020bc6d5accff 17 0 > -2400000 > 5 19069 19069 1 0x400701 0x000020bc6d71dbd4 17 0 > -2400000 > 6 19069 19069 1 0x400701 0x000020bc6d88dbe9 17 0 > -2400000 > 7 19069 19069 1 0x400701 0x000020bc6d9fea2b 17 0 > -2400000 > 8 19069 19069 1 0x400701 0x000020bc6db6f7f6 17 0 > -2400000 > 9 19069 19069 1 0x400701 0x000020bc6dcdf825 17 0 > -2400000 > > > Regards > > JK > > > > > > > > can we have sampling option > > > > perfmon2-devel@lists.sourceforge.net > > > > stephane eranian wrote: > > Hello, > > > > On Wed, Apr 2, 2008 at 1:04 PM, J K Rai wrote: > > > 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? > > > > > > > > If you have an Intel Core 2, then use this. It is much better and > > easier to program > > and the libpfm support is in much better shape than for Pentium 4. > > Note that you'd > > have to find the equivalent events. > > > > > > > > ________________________________ > > 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 > > perfmon2-devel@lists.sourceforge.net > > 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 > perfmon2-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/perfmon2-devel > > ------------------------------------------------------------------------- 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 perfmon2-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
