Geekbench ML Score

485
TensorFlow Lite GPU Inference Score
Geekbench ML 0.6.0 for Android AArch64

Result Information

Upload Date June 17 2025 09:34 AM
Views 3

System Information

System Information
Operating System Android 12
Model DTEN Bar_DCR200
Model IDDTEN Bar_DCR200
Motherboard R1-003
Governorperformance
CPU Information
Name ARM ARMv8
Topology 1 Processor, 8 Cores
Identifier ARM implementer 65 architecture 8 variant 4 part 3339 revision 0
Base Frequency 1.80 GHz
Cluster 14 Cores @ 1.80 GHz
Cluster 24 Cores @ 2.35 GHz
Memory Information
Size 15.33 GB
Inference Information
Framework TensorFlow Lite
Backend GPU
Device Default

Inference Performance

Workload Accuracy Score
Image Classification (F32)
100% 326
60.9 IPS
 
Image Classification (F16)
100% 483
90.3 IPS
 
Image Classification (I8)
97% 465
87.0 IPS
 
Image Segmentation (F32)
100% 461
7.70 IPS
 
Image Segmentation (F16)
100% 624
10.4 IPS
 
Image Segmentation (I8)
98% 608
10.2 IPS
 
Pose Estimation (F32)
100% 317
0.38 IPS
 
Pose Estimation (F16)
100% 433
0.52 IPS
 
Pose Estimation (I8)
100% 432
0.52 IPS
 
Object Detection (F32)
100% 261
19.5 IPS
 
Object Detection (F16)
100% 382
28.5 IPS
 
Object Detection (I8)
62% 382
28.5 IPS
 
Face Detection (F32)
100% 994
11.8 IPS
 
Face Detection (F16)
98% 1252
14.9 IPS
 
Face Detection (I8)
86% 1187
14.1 IPS
 
Depth Estimation (F32)
100% 610
4.73 IPS
 
Depth Estimation (F16)
100% 731
5.67 IPS
 
Depth Estimation (I8)
95% 726
5.63 IPS
 
Style Transfer (F32)
100% 1435
1.89 IPS
 
Style Transfer (F16)
100% 1769
2.33 IPS
 
Style Transfer (I8)
98% 1770
2.33 IPS
 
Image Super-Resolution (F32)
100% 321
11.5 IPS
 
Image Super-Resolution (F16)
100% 397
14.2 IPS
 
Image Super-Resolution (I8)
98% 394
14.1 IPS
 
Text Classification (F32)
100% 155
222.3 IPS
 
Text Classification (F16)
100% 155
222.9 IPS
 
Text Classification (I8)
92% 255
366.1 IPS
 
Machine Translation (F32)
100% 277
5.09 IPS
 
Machine Translation (F16)
100% 280
5.16 IPS
 
Machine Translation (I8)
64% 265
4.87 IPS
 

Compare

Set Baseline Geekbench 6 now available