Skip to content

Instantly share code, notes, and snippets.

@EricRahm
Last active August 29, 2015 14:16
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save EricRahm/dae275b3dbed2d13b65a to your computer and use it in GitHub Desktop.
Save EricRahm/dae275b3dbed2d13b65a to your computer and use it in GitHub Desktop.
AWSY - real hardware vs virtualized

On an 8 core, 32GB machine

  • 2015/03/04 Nightly build
  • 16 iterations, 8 at a time 2 loops
  • numbers in KiB
   iteration,       Start,    Start+30,         Max,      Max+30,  MaxForceGC,         End,      End+30,  EndForceGC
           0,       99971,       87024,      392673,      377293,      345905,      311185,      203429,      171741
           1,      100135,       86956,      390233,      381165,      348737,      328053,      182673,      171661
           2,       99507,       87104,      394609,      380941,      348617,      327945,      204901,      172209
           3,      100055,       86994,      385889,      381097,      348249,      328069,      184811,      172205
           4,       98303,       85664,      384445,      376821,      347433,      326301,      206145,      173069
           5,       99467,       87088,      396853,      383741,      351197,      330921,      210041,      172621
           6,       99391,       87072,      385769,      380309,      347453,      327309,      185077,      173885
           7,       99887,       87108,      394709,      379605,      347829,      327601,      183401,      172549
           8,       99371,       87016,      385481,      381241,      348881,      328305,      185223,      171673
           9,       99571,       87100,      393749,      380173,      347313,      327069,      205595,      173041
          10,       97499,       85684,      391617,      378589,      348457,      327181,      186001,      171189
          11,       99479,       87104,      389257,      380953,      349833,      329497,      187485,      173929
          12,       99823,       87580,      389121,      380929,      349433,      319861,      205909,      172485
          13,       98331,       85772,      381309,      375833,      346181,      325217,      205061,      172541
          14,       97689,       85734,      384989,      379045,      348169,      327049,      206413,      171113
          15,       99915,       86956,      393241,      378961,      347641,      327661,      205833,      172657
        mean,       99275,       86747,      389621,      379793,      348208,      326201,      196750,      172410
      stddev,     816.486,     612.512,    4405.561,    1922.438,    1272.050,    4485.651,   10525.114,     801.188
 stddev/mean,     0.00822,     0.00706,     0.01131,     0.00506,     0.00365,     0.01375,     0.05349,     0.00465

On a 2 core, 4GB VM

  • 2015/03/04 Nightly
  • 16 iterations, 2 at a time, 8 loops
  • numbers in KiB
   iteration,       Start,    Start+30,         Max,      Max+30,  MaxForceGC,         End,      End+30,  EndForceGC
           0,       96188,       84684,      380381,      372357,      343209,      321637,      182189,      168029
           1,       96843,       85092,      378041,      371173,      341221,      319337,      182077,      170257
           2,       98820,       85964,      381253,      375417,      344109,      323725,      181497,      169505
           3,       96891,       84740,      373757,      373685,      343945,      323133,      180617,      167289
           4,       98999,       86356,      378473,      373101,      342013,      321081,      182157,      169517
           5,       97319,       84740,      378953,      369577,      340629,      319301,      182373,      170165
           6,       98236,       85984,      380801,      373993,      344077,      323429,      179929,      169001
           7,       97035,       84608,      377505,      370265,      341521,      316357,      182009,      168241
           8,       98787,       86720,      386613,      376925,      345273,      324757,      181865,      169725
           9,       98931,       86484,      379461,      374701,      343561,      322377,      180389,      169821
          10,       96959,       84684,      382065,      370525,      341689,      314001,      198921,      169137
          11,       98540,       86124,      383113,      373705,      343629,      323069,      183691,      171241
          12,       96238,       84962,      378129,      371301,      341657,      320793,      182435,      170917
          13,       99191,       86396,      379545,      374005,      342093,      321777,      184309,      171101
          14,       96700,       84632,      379149,      363829,      338793,      318013,      180993,      169137
          15,       98863,       86284,      383081,      375765,      344933,      323573,      181601,      171225
        mean,       97784,       85528,      380020,      372520,      342647,      321022,      182940,      169644
      stddev,    1061.946,     787.592,    2814.318,    3032.572,    1675.894,    2860.900,    4264.603,    1133.702
 stddev/mean,     0.01086,     0.00921,     0.00741,     0.00814,     0.00489,     0.00891,     0.02331,     0.00668

Deltas

real_hardware - VM

  • - indicates VM had a higher value
            Start,  Start+30,       Max,    Max+30,MaxForceGC,       End,   End+30,EndForceGC
mean,        1491,      1219,      9601,      7273,      5561,      5179,    13810,      2766
stddev,  -245.46 ,  -175.08 ,  1591.243, -1110.134,  -403.844,  1624.751, 6260.511,  -332.514
dev/mn,  -0.00264,  -0.00215,   0.00390,  -0.00308,  -0.00124,   0.00484,  0.03018,  -0.00204

Conclusions

  • VM had lower average values for every measure, best guess is that GC type stuff kicked in more often
  • VM had higher variance in 5/8 measures, only one was > 1MiB and only for the second highest value Max+30, so that's a 0.3% drift
  • VM had much better variance on End+30
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment