Geekbench ML Score

299
TensorFlow Lite NNAPI Inference Score
Geekbench ML 0.6.0 for Android AArch64

Result Information

Upload Date April 13 2024 10:38 AM
Views 1

System Information

System Information
Operating System Android 14
Model samsung SM-P620
Model IDsamsung SM-P620
Motherboard s5e8825
Governorenergy_aware
CPU Information
Name ARM ARMv8
Topology 1 Processor, 8 Cores
Identifier ARM implementer 65 architecture 8 variant 1 part 3393 revision 1
Base Frequency 2.00 GHz
Cluster 16 Cores @ 2.00 GHz
Cluster 22 Cores @ 2.40 GHz
Memory Information
Size 3.64 GB
Inference Information
Framework TensorFlow Lite
Backend NNAPI
Device Default

Inference Performance

Workload Accuracy Score
Image Classification (F32)
100% 120
22.5 IPS
 
Image Classification (F16)
100% 119
22.3 IPS
 
Image Classification (I8)
97% 323
60.5 IPS
 
Image Segmentation (F32)
100% 193
3.22 IPS
 
Image Segmentation (F16)
100% 284
4.74 IPS
 
Image Segmentation (I8)
98% 362
6.05 IPS
 
Pose Estimation (F32)
100% 667
0.81 IPS
 
Pose Estimation (F16)
100% 2201
2.67 IPS
 
Pose Estimation (I8)
100% 806
0.98 IPS
 
Object Detection (F32)
100% 127
9.52 IPS
 
Object Detection (F16)
100% 128
9.52 IPS
 
Object Detection (I8)
61% 365
27.2 IPS
 
Face Detection (F32)
100% 311
3.70 IPS
 
Face Detection (F16)
100% 308
3.66 IPS
 
Face Detection (I8)
86% 715
8.50 IPS
 
Depth Estimation (F32)
100% 262
2.03 IPS
 
Depth Estimation (F16)
100% 259
2.01 IPS
 
Depth Estimation (I8)
95% 771
5.98 IPS
 
Style Transfer (F32)
100% 560
0.74 IPS
 
Style Transfer (F16)
100% 581
0.76 IPS
 
Style Transfer (I8)
98% 378
0.50 IPS
 
Image Super-Resolution (F32)
100% 256
9.14 IPS
 
Image Super-Resolution (F16)
100% 375
13.4 IPS
 
Image Super-Resolution (I8)
98% 470
16.8 IPS
 
Text Classification (F32)
100% 76
109.1 IPS
 
Text Classification (F16)
100% 87
124.6 IPS
 
Text Classification (I8)
92% 67
96.6 IPS
 
Machine Translation (F32)
100% 292
5.38 IPS
 
Machine Translation (F16)
100% 293
5.40 IPS
 
Machine Translation (I8)
64% 387
7.12 IPS
 

Compare

Set Baseline Geekbench 6 now available