Skip to content

Instantly share code, notes, and snippets.

@mrdrozdov
Created November 15, 2018 02:27
Show Gist options
  • Save mrdrozdov/ab5f55a6f76b490aa0b7dac7a005a346 to your computer and use it in GitHub Desktop.
Save mrdrozdov/ab5f55a6f76b490aa0b7dac7a005a346 to your computer and use it in GitHub Desktop.
profiling-diora.txt
--- PROFILING ---
--- INSIDE ---
_unsafe_view 19 127.431us 108.259us
t 19 158.285us 160.697us
split 19 272.784us 276.931us
exp 19 276.274us 283.493us
th_sub 19 282.381us 296.449us
div 19 280.799us 299.139us
max 19 332.444us 343.678us
bmm 19 307.893us 362.781us
chunk 19 459.374us 458.180us
tanh 38 508.321us 520.585us
_mm 38 620.652us 691.228us
sub 19 872.944us 873.760us
sigmoid 76 949.660us 986.720us
_sum 76 1127.296us 1206.638us
expand 224 1209.453us 1238.299us
th_add 95 1155.118us 1283.834us
th_mul 133 1739.922us 1812.060us
sum 76 1888.664us 1883.716us
mm 38 2516.511us 2523.935us
add 133 3334.500us 3351.875us
_th_get_device 1843 3953.165us 3856.774us
matmul 57 4187.487us 4189.667us
mul 133 4254.786us 4244.912us
slice 1140 5505.517us 5625.770us
cat 133 7590.319us 7653.247us
view 1577 8969.422us 9044.264us
--- OUTSIDE ---
_unsafe_view 19 139.717us 72.965us
t 19 164.965us 164.815us
split 19 258.498us 243.075us
exp 19 272.974us 282.912us
th_sub 19 287.134us 319.146us
div 19 284.718us 321.414us
max 19 350.897us 371.041us
chunk 19 448.382us 401.768us
bmm 19 309.697us 440.157us
_mm 38 631.940us 815.973us
sub 19 873.228us 874.941us
sigmoid 76 970.515us 1008.928us
mm 38 1012.242us 1132.846us
_sum 76 1143.593us 1228.904us
expand 226 1238.692us 1266.144us
_th_get_device 703 1473.330us 1436.067us
th_add 95 1232.884us 1441.943us
th_mul 133 1767.056us 1869.358us
sum 76 1905.297us 1905.830us
tanh 38 2212.632us 2233.576us
matmul 57 2715.424us 2819.307us
view 439 4379.442us 4048.011us
mul 133 4295.157us 4292.817us
_cast_Long 1254 4988.312us 4382.346us
cat 133 4228.074us 4543.700us
add 133 5219.103us 5335.887us
tensor 1255 6621.567us 6660.574us
index_select 1254 22025.449us 27701.063us
--- LOSS ---
alias 1 4.818us 4.096us
_cast_Float 3 9.309us 8.224us
div 1 14.514us 15.999us
transpose 2 20.205us 19.520us
_cast_Long 20 56.237us 49.660us
tensor 24 131.325us 132.379us
th_add 19 190.357us 210.302us
view 25 226.669us 228.996us
add 21 483.372us 483.553us
t 100 780.738us 625.602us
cat 24 752.587us 766.047us
multi_margin_loss_forward 20 1078.269us 1086.881us
th_mul 80 1191.997us 1284.161us
select 240 1260.051us 1288.415us
multi_margin_loss 20 1309.145us 1297.050us
_th_get_device 618 1345.314us 1304.337us
_sum 80 1218.254us 1304.636us
_mm 80 1220.897us 1570.946us
sum 80 2066.277us 2054.274us
embedding 80 1287.647us 2067.939us
mm 80 2159.630us 2378.108us
mul 100 2664.762us 2668.190us
--- BACKWARD ---
torch::autograd::GraphRoot 1 20.036us 8.192us
DivBackward0 1 26.579us 26.623us
MeanBackward0 1 64.247us 64.510us
torch::autograd::AccumulateGrad 11 137.073us 149.292us
AddBackward 76 206.771us 214.085us
transpose 74 480.161us 362.587us
MulBackward 20 458.010us 456.921us
multi_margin_loss_backward 20 505.517us 527.935us
ThAddBackward 205 558.362us 537.445us
select 80 527.227us 539.230us
_unsafe_view 89 889.172us 636.948us
MultiMarginLossBackward 20 701.385us 701.504us
scatter_ 37 732.471us 866.539us
neg 74 966.164us 1011.360us
SumBackward0 1 1081.920us 1073.311us
UnsafeViewBackward 37 1076.161us 1076.111us
div 76 1008.065us 1114.780us
_tanh_backward 76 1051.795us 1120.216us
ThSubBackward 37 1350.994us 1353.027us
ExpBackward 37 1388.481us 1382.881us
cat 37 1418.365us 1565.613us
TanhBackward 76 1576.318us 1578.766us
bmm 74 1168.337us 1608.604us
expand 228 1711.590us 1744.743us
SplitBackward 37 1740.573us 1807.831us
clone 89 1465.725us 1998.753us
_sigmoid_backward 150 1923.866us 2067.032us
unsqueeze 301 2121.912us 2166.077us
BmmBackward 37 2310.235us 2438.614us
MaxBackward0 37 2563.568us 2580.338us
SigmoidBackward 150 2939.758us 2924.019us
TBackward 139 3211.354us 3188.156us
t 489 3574.080us 3213.207us
_sum 183 3308.187us 3671.783us
SelectBackward 80 4144.497us 4249.165us
DivBackward1 37 4920.302us 4934.395us
SumBackward1 226 5030.638us 5017.960us
_mm 231 3683.775us 5294.044us
sum 183 5365.974us 5474.426us
ExpandBackward 438 7028.565us 7071.922us
mm 231 6273.584us 7320.141us
th_mul 789 11421.374us 12267.498us
MmBackward 156 13018.102us 13316.200us
tensor 2324 13506.104us 13773.186us
zero_ 2324 18725.438us 18763.668us
slice 3315 19590.575us 19702.797us
index_add_ 1214 15218.748us 23564.720us
mul 846 29786.829us 29879.585us
ThMulBackward 339 30428.454us 30368.980us
as_strided 3614 36248.612us 36421.127us
narrow 2322 38189.874us 37601.418us
th_add 2539 35993.263us 38433.193us
SliceBackward 993 42471.537us 43544.861us
CatBackward 278 45162.544us 44409.542us
reshape 1896 51060.396us 51124.851us
ViewBackward 1859 64920.213us 64799.984us
_th_get_device 23324 69167.414us 67426.092us
IndexSelectBackward 1214 63099.901us 68575.462us
add 2539 84678.916us 84306.908us
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment