Upload Date | July 26 2024 09:52 AM |
Views | 1 |
System Information | |
---|---|
Operating System | Android 14 |
Model | realme RMX3870 |
Model ID | realme RMX3870 |
Motherboard | RM6877 |
CPU Information | |
---|---|
Name | ARM MT6877V/TTZA |
Topology | 1 Processor, 8 Cores |
Identifier | ARM implementer 65 architecture 8 variant 1 part 3393 revision 0 |
Base Frequency | 2.00 GHz |
Cluster 1 | 6 Cores @ 2.00 GHz |
Cluster 2 | 2 Cores @ 2.60 GHz |
Memory Information | |
---|---|
Size | 7.36 GB |
Inference Information | |
---|---|
Framework | TensorFlow Lite |
Backend | CPU |
Device | Default |
Workload | Accuracy | Score | |
---|---|---|---|
Image Classification (F32)
|
100% |
165
30.9 IPS |
|
Image Classification (F16)
|
100% |
165
30.9 IPS |
|
Image Classification (I8)
|
97% |
371
69.4 IPS |
|
Image Segmentation (F32)
|
100% |
280
4.67 IPS |
|
Image Segmentation (F16)
|
100% |
264
4.41 IPS |
|
Image Segmentation (I8)
|
98% |
483
8.07 IPS |
|
Pose Estimation (F32)
|
100% |
417
0.51 IPS |
|
Pose Estimation (F16)
|
100% |
399
0.48 IPS |
|
Pose Estimation (I8)
|
100% |
1781
2.16 IPS |
|
Object Detection (F32)
|
100% |
187
14.0 IPS |
|
Object Detection (F16)
|
100% |
184
13.7 IPS |
|
Object Detection (I8)
|
61% |
488
36.4 IPS |
|
Face Detection (F32)
|
100% |
398
4.73 IPS |
|
Face Detection (F16)
|
100% |
403
4.79 IPS |
|
Face Detection (I8)
|
86% |
860
10.2 IPS |
|
Depth Estimation (F32)
|
100% |
428
3.32 IPS |
|
Depth Estimation (F16)
|
100% |
408
3.17 IPS |
|
Depth Estimation (I8)
|
95% |
1005
7.79 IPS |
|
Style Transfer (F32)
|
100% |
689
0.91 IPS |
|
Style Transfer (F16)
|
100% |
713
0.94 IPS |
|
Style Transfer (I8)
|
98% |
1943
2.56 IPS |
|
Image Super-Resolution (F32)
|
100% |
230
8.23 IPS |
|
Image Super-Resolution (F16)
|
100% |
229
8.18 IPS |
|
Image Super-Resolution (I8)
|
98% |
685
24.4 IPS |
|
Text Classification (F32)
|
100% |
210
301.8 IPS |
|
Text Classification (F16)
|
100% |
257
369.0 IPS |
|
Text Classification (I8)
|
92% |
184
264.8 IPS |
|
Machine Translation (F32)
|
100% |
369
6.78 IPS |
|
Machine Translation (F16)
|
100% |
408
7.50 IPS |
|
Machine Translation (I8)
|
64% |
289
5.32 IPS |