Wow. Ok, not many events there. I found the bug. Looks like there is a bug in 'perfmon_events.csv' file. This gets installed to /usr/share/papi I believe (unless you put it elsewhere...)
If you edit that file and find your CPU then change LAST_LEVEL_CACHE_REFERENCES to LAST_LEVEL_CACHE_REFERENCE You'll be good to go. PAPI team, please fix this in the distribution... (unless someone from the perfmon folks tells me it's a bug in libpfm...) Phil On Mar 18, 2009, at 11:22 AM, George Markomanolis wrote: > Hi Phil, > > Thanks for the answer. > I have problem to find the example I just patched my kernel with > perfctr and installed perfctr and papi, there is libpfm.a but no > examples folder (only at the setup files). > > papi_native_avail is: > PAPI Error: pfm_find_full_event(LAST_LEVEL_CACHE_REFERENCES, > 0xbfaedf64): event not found. > PAPI Error: 1 of 7 events in perfmon_events.csv were not valid. > Available native events and hardware information. > -------------------------------------------------------------------------------- > Vendor string and code : GenuineIntel (1) > Model string and code : Intel Core (17) > CPU Revision : 8.000000 > CPU Megahertz : 1000.000000 > CPU Clock Megahertz : 1000 > CPU's in this Node : 2 > Nodes in this System : 1 > Total CPU's : 2 > Number Hardware Counters : 2 > Max Multiplex Counters : 32 > -------------------------------------------------------------------------------- > The following correspond to fields in the PAPI_event_info_t structure. > > Event Code Symbol | Long Description | > -------------------------------------------------------------------------------- > 0x40000000 UNHALTED_CORE_CYCLES | Unhalted core cycles | > -------------------------------------------------------------------------------- > 0x40000001 UNHALTED_REFERENCE_CYCLES | Unhalted reference cycles. > Measures | > | bus cycles | > -------------------------------------------------------------------------------- > 0x40000002 INSTRUCTIONS_RETIRED | Instructions retired | > -------------------------------------------------------------------------------- > 0x40000003 LAST_LEVEL_CACHE_REFERENCE | Last level of cache > references | > -------------------------------------------------------------------------------- > 0x40000004 LAST_LEVEL_CACHE_MISSES | Last level of cache misses | > -------------------------------------------------------------------------------- > 0x40000005 BRANCH_INSTRUCTIONS_RETIRED | Branch instructions retired | > -------------------------------------------------------------------------------- > 0x40000006 MISPREDICTED_BRANCH_RETIRED | Mispredicted branch > instruction re | > | tired | > -------------------------------------------------------------------------------- > 0x40000007 LD_BLOCKS | Load operations delayed due to store buffer > blocks. | > | The preceding store may be blocked due to unknown address, unknow | > | n data, or conflict due to partial overlap between the load and s | > | tore. | > -------------------------------------------------------------------------------- > 0x40000008 SD_DRAINS | Cycles while draining store buffers | > -------------------------------------------------------------------------------- > 0x40000009 MISALIGN_MEM_REF | Misaligned data memory references (MOB > splits | > | of loads and stores) | > -------------------------------------------------------------------------------- > 0x4000000a SEG_REG_LOADS | Segment register loads | > -------------------------------------------------------------------------------- > 0x4000000b SSE_PREFETCH | SSE software prefetch instruction | > 4000100b :NTA | prefetchnta retired | > 4000200b :T1 | prefetcht1 retired | > 4000400b :T2 | prefetcht2 retired | > -------------------------------------------------------------------------------- > 0x4000000c SSE_NT_STORES_RETIRED | SSE streaming store instructions > retired | > -------------------------------------------------------------------------------- > 0x4000000d L2_ADS | L2 address stobes | > 4000100d :ALL_CORES | monitor all cores | > 4000200d :THIS_CORE | monitor this core | > -------------------------------------------------------------------------------- > 0x4000000e DBUS_BUSY | Core cycle during which data bus was busy > (increment | > | s by 4) | > 4000100e :ALL_CORES | monitoring all cores | > 4000200e :THIS_CORE | monitoring this core | > -------------------------------------------------------------------------------- > 0x4000000f L2_LINES_IN | L2 cache lines allocated | > 4000100f :ALL_PREFETCHES_ALL_CORES | monitor all types of prefetches > on a | > | ll cores | > 4000200f :ALL_PREFETCHES_THIS_CORE | monitor all types of prefetches > on t | > | his core | > 4000400f :HW_PREFETCH_ONLY_ALL_CORES | monitor only hardware > prefetches o | > | n all cores | > 4000800f :HW_PREFETCH_ONLY_THIS_CORE | monitor only hardware > prefetches o | > | n this core | > 4001000f :EXCL_HW_PREFETCH_ALL_CORES | monitoring exclude hardware > prefet | > | ches on all cores | > 4002000f :EXCL_HW_PREFETCH_THIS_CORE | monitoring exclude hardware > prefet | > | ches on this core | > -------------------------------------------------------------------------------- > 0x40000010 BUS_DRDY_CLOCKS | External bus cycles while DRDY is > asserted | > 40001010 :ALL_AGENTS | monitoring all agents | > 40002010 :THIS_AGENT | monitoring this agent | > -------------------------------------------------------------------------------- > 0x40000011 BUS_TRANS_RFO | Completed read for ownership (RFO) > transactions | > 40001011 :ALL_CORES_ALL_AGENTS | monitoring all cores and all agents | > 40002011 :ALL_CORES_THIS_AGENT | monitoring all cores and this agent | > 40004011 :THIS_CORE_ALL_AGENTS | monitoring this core and all agents | > 40008011 :THIS_CORE_THIS_AGENT | monitoring this core and this agent | > -------------------------------------------------------------------------------- > -------------------------------------------------------------------------------- > Total events reported: 18 > native_avail.c PASSED > > > and /proc/cpuinfo is: > > processor : 0 > vendor_id : GenuineIntel > cpu family : 6 > model : 14 > model name : Genuine Intel(R) CPU L2400 @ 1.66GHz > stepping : 8 > cpu MHz : 1000.000 > cache size : 2048 KB > physical id : 0 > siblings : 2 > core id : 0 > cpu cores : 2 > fdiv_bug : no > hlt_bug : no > f00f_bug : no > coma_bug : no > fpu : yes > fpu_exception : yes > cpuid level : 10 > wp : yes > flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca > cmov pat clflush dts acpi mmx fxsr sse sse2 ss httm pbe nx > constant_tsc arch_perfmon bts pni monitor vmx est tm2 xtpr > bogomips : 3329.57 > clflush size : 64 > > processor : 1 > vendor_id : GenuineIntel > cpu family : 6 > model : 14 > model name : Genuine Intel(R) CPU L2400 @ 1.66GHz > stepping : 8 > cpu MHz : 1000.000 > cache size : 2048 KB > physical id : 0 > siblings : 2 > core id : 1 > cpu cores : 2 > fdiv_bug : no > hlt_bug : no > f00f_bug : no > coma_bug : no > fpu : yes > fpu_exception : yes > cpuid level : 10 > wp : yes > flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca > cmov pat clflush dts acpi mmx fxsr sse sse2 ss httm pbe nx > constant_tsc arch_perfmon bts pni monitor vmx est tm2 xtpr > bogomips : 3325.11 > clflush size : 64 > > Could I compile showevtinfo? > > Thans a lot, > best regards, > George > > > Philip Mucci wrote: >> Hi George, >> >> Errors like this usually indicate a bug in the PAPI event-table, >> which maps PAPI preset (abstract) events to native event names, >> graciously provided to us by libpfm. It sounds like this event >> table is pretty slim for this CPU. >> >> Now, I'm assuming you have installed perfctr as your patch. >> >> Can you run the 'showevtinfo' from the libpfm/examples directory, >> papi_native_avail and also send us the /proc/cpuinfo? >> >> There could be some stuff missing on the LV cpus... >> >> Phil >> >> On Mar 17, 2009, at 6:05 AM, George Markomanolis wrote: >> >>> Dear all, >>> >>> First of all I should say that I am newbie to PAPI. I have installed >>> PAPI on debian lenny and I think it's ok. >>> How am I sure that the installation is ok? >>> I run papi_avail and it says that only 6 events are yes but no >>> PAPI_FP_OPS. How could I count flops? >>> Events with yes: >>> PAPI_L2_TCM 0x80000007 Yes No Level 2 cache misses >>> PAPI_BR_TKN 0x8000002c Yes No Conditional branch instructions taken >>> PAPI_BR_MSP 0x8000002e Yes No Conditional branch instructions >>> mispredicted >>> PAPI_TOT_INS 0x80000032 Yes No Instructions completed >>> PAPI_BR_INS 0x80000037 Yes No Branch instructions >>> PAPI_TOT_CYC 0x8000003b Yes No Total cycles >>> >>> My laptop has a 32bit low voltage cpum core duo L2400. Is this a >>> problem? >>> >>> >>> Also what's about this error? >>> PAPI Error: pfm_find_full_event(LAST_LEVEL_CACHE_REFERENCES, >>> 0xbfb1a0e4): >>> event not found. >>> The program doesn't crash but I have this error in every program >>> with papi. >>> >>> The kernel is 2.6.25 and I have installed PAPI 3.6.2 from tar file >>> (no cvs). >>> >>> Thanks a lot, >>> best regards, >>> George Markomanolis >>> >>> _______________________________________________ >>> Ptools-perfapi mailing list >>> ptools-perf...@cs.utk.edu >>> http://lists.cs.utk.edu/listinfo/ptools-perfapi >> >> > ------------------------------------------------------------------------------ Apps built with the Adobe(R) Flex(R) framework and Flex Builder(TM) are powering Web 2.0 with engaging, cross-platform capabilities. Quickly and easily build your RIAs with Flex Builder, the Eclipse(TM)based development software that enables intelligent coding and step-through debugging. Download the free 60 day trial. http://p.sf.net/sfu/www-adobe-com _______________________________________________ perfmon2-devel mailing list perfmon2-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/perfmon2-devel
