1

I tried to install papi on ubutu 16.4 on my desktop. I am facing error I con't find any solution for it. Maybe because this is my first experience with Papi.I download papi version 5.5 from http://icl.cs.utk.edu/papi/software/index.html. I did the first step to install it on my machine by run $ .\configure and them run make, however when I run make test I have the following error:

 make[1]: Leaving directory '/home/kaloon/Downloads/papi-5.5.1/src/ctests'
 ctests/zero
 Test case 0: start, stop.
  -----------------------------------------------
  Default domain is: 1 (PAPI_DOM_USER)
  Default granularity is: 1 (PAPI_GRN_THR)
   Using 20000000 iterations of c += a*b
  -------------------------------------------------------------------------
  Test type    :               1
 PAPI_TOT_INS :         200000222
 PAPI_TOT_CYC :         180014757
 Real usec    :             45126
Real cycles  :      162093114
Virt usec    :          45126
 Virt cycles  :         180504000
 -------------------------------------------------------------------------
 Verification: PAPI_TOT_CYC should be roughly real_cycles
 NOTE: Not true if dynamic frequency scaling is enabled.
 Verification: PAPI_FP_INS should be roughly 40000000
 PAPI_TOT_INS Error of 400.00%
 zero.c                                       FAILED
 Line # 130
 Error: FLOPS validation

  Makefile.inc:225: recipe for target 'test' failed
  make: *** [test] Error 1
kaloon
  • 157
  • 4
  • 15
  • this my machine information Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 8 On-line CPU(s) list: 0-7 Thread(s) per core: 2 Core(s) per socket: 4 Socket(s): 1 NUMA node(s): 1 Vendor ID: GenuineIntel – kaloon Apr 08 '17 at 23:37
  • CPU family: 6 Model: 60 Model name: Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz Stepping: 3 CPU MHz: 3752.490 CPU max MHz: 4000.0000 CPU min MHz: 800.0000 BogoMIPS: 7184.00 Virtualization: VT-x L1d cache: 32K L1i cache: 32K L2 cache: 256K L3 cache: 8192K NUMA node0 CPU(s): 0-7 – kaloon Apr 08 '17 at 23:39

1 Answers1

1

Just posting the answer from the mailing list of papi.

this is a known bug on machines without good FP counters (somewhat recent Intel chips). We fall back to PAPI_TOT_INS but don't change the expected value.

We should probably disable validation in that case, or better yet, make all our general PAPI validation happen with more stable counters (like total_instructions) and move the FP validation to a separate set of utilities.

Vince

Nachshon Cohen
  • 113
  • 2
  • 10