[molpro-user] Benchmark timing on linux raid

Jacek Klos jklos at umd.edu
Wed Nov 24 23:24:29 GMT 2010


Thanks Gerald,

I ran 30 jobs with 12 CPUs after fresh reboot on linux Dell machine  
and the first job n12_1 is the fastest, subsequent jobs
are  more or less 160 seconds longer. Looks indeed that bigger jobs  
fills some disk cache, smaller jobs like small_normal_ccsd
are more or less fast until some  bigger job was running that used  
disk more extensively , after that these small benchmarks
  suffer  slightly longer execution time too.

mpp_verybig_normal_mrci.n12_10.out: TOTAL           0.0000000    
1438.67   260.21   1698.88   1794.16    80.19      94.69
mpp_verybig_normal_mrci.n12_11.out: TOTAL           0.0000000    
1425.08   265.75   1690.83   1793.75    79.45      94.26
mpp_verybig_normal_mrci.n12_12.out: TOTAL           0.0000000    
1426.57   266.03   1692.60   1781.57    80.07      95.01
mpp_verybig_normal_mrci.n12_13.out: TOTAL           0.0000000    
1426.60   267.82   1694.42   1802.77    79.13      93.99
mpp_verybig_normal_mrci.n12_14.out: TOTAL           0.0000000    
1435.97   266.42   1702.39   1807.21    79.46      94.20
mpp_verybig_normal_mrci.n12_15.out: TOTAL           0.0000000    
1434.03   259.73   1693.76   1801.96    79.58      94.00
mpp_verybig_normal_mrci.n12_16.out: TOTAL           0.0000000    
1416.84   278.66   1695.50   1794.82    78.94      94.47
mpp_verybig_normal_mrci.n12_17.out: TOTAL           0.0000000    
1435.78   281.07   1716.85   1807.26    79.45      95.00
mpp_verybig_normal_mrci.n12_18.out: TOTAL           0.0000000    
1426.61   257.95   1684.56   1787.10    79.83      94.26
mpp_verybig_normal_mrci.n12_19.out: TOTAL           0.0000000    
1439.22   275.68   1714.90   1803.59    79.80      95.08
FIRST->mpp_verybig_normal_mrci.n12_1.out: TOTAL           0.0000000    
1398.61   160.33   1558.94   1639.61    85.30      95.08
mpp_verybig_normal_mrci.n12_20.out: TOTAL           0.0000000    
1435.07   262.59   1697.66   1800.39    79.71      94.29
mpp_verybig_normal_mrci.n12_21.out: TOTAL           0.0000000    
1441.47   267.42   1708.89   1800.25    80.07      94.93
mpp_verybig_normal_mrci.n12_22.out: TOTAL           0.0000000    
1440.82   263.13   1703.95   1802.00    79.96      94.56
mpp_verybig_normal_mrci.n12_23.out: TOTAL           0.0000000    
1442.43   270.11   1712.54   1803.42    79.98      94.96
mpp_verybig_normal_mrci.n12_24.out: TOTAL           0.0000000    
1427.74   280.92   1708.66   1813.78    78.72      94.20
mpp_verybig_normal_mrci.n12_25.out: TOTAL           0.0000000    
1435.17   277.26   1712.43   1818.69    78.91      94.16
mpp_verybig_normal_mrci.n12_26.out: TOTAL           0.0000000    
1431.74   271.63   1703.37   1801.21    79.49      94.57
mpp_verybig_normal_mrci.n12_27.out: TOTAL           0.0000000    
1425.27   284.95   1710.22   1803.83    79.01      94.81
mpp_verybig_normal_mrci.n12_28.out: TOTAL           0.0000000    
1425.87   285.76   1711.63   1819.90    78.35      94.05
mpp_verybig_normal_mrci.n12_29.out: TOTAL           0.0000000    
1440.25   272.53   1712.78   1810.78    79.54      94.59
mpp_verybig_normal_mrci.n12_2.out: TOTAL           0.0000000    
1433.19   271.42   1704.61   1845.38    77.66      92.37
mpp_verybig_normal_mrci.n12_30.out: TOTAL           0.0000000    
1427.30   275.63   1702.93   1819.76    78.43      93.58
mpp_verybig_normal_mrci.n12_3.out: TOTAL           0.0000000    
1432.39   269.17   1701.56   1847.98    77.51      92.08
mpp_verybig_normal_mrci.n12_4.out: TOTAL           0.0000000    
1440.61   270.70   1711.31   1853.01    77.74      92.35
mpp_verybig_normal_mrci.n12_5.out: TOTAL           0.0000000    
1434.48   271.24   1705.72   1844.74    77.76      92.46
mpp_verybig_normal_mrci.n12_6.out: TOTAL           0.0000000    
1424.29   273.93   1698.22   1829.11    77.87      92.84
mpp_verybig_normal_mrci.n12_7.out: TOTAL           0.0000000    
1429.79   270.34   1700.13   1811.49    78.93      93.85
mpp_verybig_normal_mrci.n12_8.out: TOTAL           0.0000000    
1433.15   262.41   1695.56   1804.51    79.42      93.96
mpp_verybig_normal_mrci.n12_9.out: TOTAL           0.0000000    
1429.19   271.79   1700.98   1819.42    78.55      93.49

Best
Jacek Klos
On Nov 24, 2010, at 3:16 AM, Gerald Knizia wrote:

> On Wednesday 24 November 2010 00:49, Jacek Klos wrote:
>> I have noticed some strange behavior on our linux Dell machine.
>> Machine is dual-6core 3.33GHz 24GB RAM, and RAID disk composed
>> of 3 SAS 15k RPM drives making logical volume of about 1TB.
>>
>> When I run   small_normal_ccsd benchmarks with 12 cpus using parallel
>> molpro (GA4.2 Open-MPI) and I repeat it let's say 20 times after  
>> the fresh
>> reboot of the machine the fastest elapsed time is 76 seconds.
>> The range of elapsed times is more or less +/- 5 sek.
>>
>> But when I run again the same series of 12cpus jobs after machine is
>> on for a day or so I get elapsed timings like 30 seconds longer:
>
> It may be possible that this is related to the 'system cache', i.e.,  
> the
> amount of free physical RAM not committed to any other running  
> processes.
> Operating systems tend to use all of that memory for disk caching,  
> and since
> these CCSD jobs are still rather small and you have lots of memory,  
> maybe
> after a fresh reboot the OS never actually writes the integral data  
> to disk
> in the first place (or never reads it back). AOINT, AOSORT, HF and  
> TRANSFORM
> all depend heavily on the IO performance usually, so if for some  
> reason there
> is lots of spare RAM for disk caching, these would become much faster.
>
> Why this stops being the case after the system has run for a longer  
> time is
> another question. Maybe in the meantime more files have been opened  
> and the
> OS determined these as more cache-worthy than the integral files.
>
> There may be tools in linux to tell you about the system cache  
> usage, but I'm
> not an expert and could not tell which. One thing you could try is  
> to test of
> this difference of execution speed also occurs if (a) you run  
> smaller jobs,
> which always fit completely into memory or (b) you run larger jobs  
> (with
> integral files >> 24 GB), which never fit completely into memory.
> -- 
> Gerald Knizia




More information about the Molpro-user mailing list