Add beam search profile
Browse files- part3-profile/results.json +134 -0
part3-profile/results.json
ADDED
|
@@ -0,0 +1,134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"gpu": "NVIDIA GeForce RTX 2080 Ti",
|
| 3 |
+
"parameters": 162322944,
|
| 4 |
+
"widths": {
|
| 5 |
+
"1": {
|
| 6 |
+
"seconds_per_story": 0.6447052674018778,
|
| 7 |
+
"steps": 640,
|
| 8 |
+
"ms_per_step": {
|
| 9 |
+
"select": 0.23089872383934562,
|
| 10 |
+
"python": 0.004803401225217385,
|
| 11 |
+
"reorder": 0.38361844099199516,
|
| 12 |
+
"forward": 9.336870003789954,
|
| 13 |
+
"forward_gpu": 9.311363458633437
|
| 14 |
+
},
|
| 15 |
+
"forward": {
|
| 16 |
+
"kernels": 324,
|
| 17 |
+
"kernel_ms": 1.9261217500000245,
|
| 18 |
+
"wall_ms": 10.964226949727163
|
| 19 |
+
}
|
| 20 |
+
},
|
| 21 |
+
"2": {
|
| 22 |
+
"seconds_per_story": 0.782631797605427,
|
| 23 |
+
"steps": 640,
|
| 24 |
+
"ms_per_step": {
|
| 25 |
+
"select": 0.2974080234707799,
|
| 26 |
+
"python": 0.005698567656509113,
|
| 27 |
+
"reorder": 0.5211300629525795,
|
| 28 |
+
"forward": 11.301132964035787,
|
| 29 |
+
"forward_gpu": 11.27244565039872
|
| 30 |
+
},
|
| 31 |
+
"forward": {
|
| 32 |
+
"kernels": 359,
|
| 33 |
+
"kernel_ms": 2.458864400000012,
|
| 34 |
+
"wall_ms": 11.192965900409035
|
| 35 |
+
}
|
| 36 |
+
},
|
| 37 |
+
"4": {
|
| 38 |
+
"seconds_per_story": 0.7679494620009791,
|
| 39 |
+
"steps": 626,
|
| 40 |
+
"ms_per_step": {
|
| 41 |
+
"select": 0.2973985340529547,
|
| 42 |
+
"python": 0.006724892391677434,
|
| 43 |
+
"reorder": 0.5206945197903715,
|
| 44 |
+
"forward": 11.29611559582654,
|
| 45 |
+
"forward_gpu": 11.267488924459135
|
| 46 |
+
},
|
| 47 |
+
"forward": {
|
| 48 |
+
"kernels": 359,
|
| 49 |
+
"kernel_ms": 2.5949690500000457,
|
| 50 |
+
"wall_ms": 11.152151101850905
|
| 51 |
+
}
|
| 52 |
+
},
|
| 53 |
+
"8": {
|
| 54 |
+
"seconds_per_story": 0.8190970009018201,
|
| 55 |
+
"steps": 640,
|
| 56 |
+
"ms_per_step": {
|
| 57 |
+
"select": 0.298705698151025,
|
| 58 |
+
"python": 0.008746037474338664,
|
| 59 |
+
"reorder": 0.5198253115850093,
|
| 60 |
+
"forward": 11.813053080641112,
|
| 61 |
+
"forward_gpu": 11.784311142563828
|
| 62 |
+
},
|
| 63 |
+
"forward": {
|
| 64 |
+
"kernels": 431,
|
| 65 |
+
"kernel_ms": 3.4558436999999924,
|
| 66 |
+
"wall_ms": 11.602508201031014
|
| 67 |
+
}
|
| 68 |
+
},
|
| 69 |
+
"16": {
|
| 70 |
+
"seconds_per_story": 0.7960014707001392,
|
| 71 |
+
"steps": 640,
|
| 72 |
+
"ms_per_step": {
|
| 73 |
+
"select": 0.3233663214814442,
|
| 74 |
+
"python": 0.011938456827920163,
|
| 75 |
+
"reorder": 0.5226361700806592,
|
| 76 |
+
"forward": 11.362013666257553,
|
| 77 |
+
"forward_gpu": 11.333163607120516
|
| 78 |
+
},
|
| 79 |
+
"forward": {
|
| 80 |
+
"kernels": 335,
|
| 81 |
+
"kernel_ms": 3.696454650000007,
|
| 82 |
+
"wall_ms": 11.088423698674887
|
| 83 |
+
}
|
| 84 |
+
},
|
| 85 |
+
"32": {
|
| 86 |
+
"seconds_per_story": 0.831490383198252,
|
| 87 |
+
"steps": 640,
|
| 88 |
+
"ms_per_step": {
|
| 89 |
+
"select": 0.3822458850663679,
|
| 90 |
+
"python": 0.018040695067611523,
|
| 91 |
+
"reorder": 0.8000543237358215,
|
| 92 |
+
"forward": 11.562296484044055,
|
| 93 |
+
"forward_gpu": 11.533552107214934
|
| 94 |
+
},
|
| 95 |
+
"forward": {
|
| 96 |
+
"kernels": 359,
|
| 97 |
+
"kernel_ms": 3.757427350000004,
|
| 98 |
+
"wall_ms": 11.353752799914218
|
| 99 |
+
}
|
| 100 |
+
},
|
| 101 |
+
"64": {
|
| 102 |
+
"seconds_per_story": 0.8898728845990262,
|
| 103 |
+
"steps": 640,
|
| 104 |
+
"ms_per_step": {
|
| 105 |
+
"select": 0.5405968356171797,
|
| 106 |
+
"python": 0.02994586611748673,
|
| 107 |
+
"reorder": 0.8986677828033862,
|
| 108 |
+
"forward": 12.20695640795384,
|
| 109 |
+
"forward_gpu": 12.178275310993183
|
| 110 |
+
},
|
| 111 |
+
"forward": {
|
| 112 |
+
"kernels": 359,
|
| 113 |
+
"kernel_ms": 6.012091450000032,
|
| 114 |
+
"wall_ms": 11.445552398799919
|
| 115 |
+
}
|
| 116 |
+
},
|
| 117 |
+
"128": {
|
| 118 |
+
"seconds_per_story": 0.9452142661961261,
|
| 119 |
+
"steps": 618,
|
| 120 |
+
"ms_per_step": {
|
| 121 |
+
"select": 0.911044792692602,
|
| 122 |
+
"python": 0.05660144038404968,
|
| 123 |
+
"reorder": 1.3905601175863622,
|
| 124 |
+
"forward": 12.455254171437714,
|
| 125 |
+
"forward_gpu": 12.426015934126266
|
| 126 |
+
},
|
| 127 |
+
"forward": {
|
| 128 |
+
"kernels": 371,
|
| 129 |
+
"kernel_ms": 8.861494050000026,
|
| 130 |
+
"wall_ms": 11.505910349660553
|
| 131 |
+
}
|
| 132 |
+
}
|
| 133 |
+
}
|
| 134 |
+
}
|