Upload Date | May 08 2024 10:14 PM |
Views | 1 |
System Information | |
---|---|
Operating System | Android 13 |
Model | Google nissa |
Model ID | Google nissa |
Motherboard | joxer |
CPU Information | |
---|---|
Name | placeholder |
Topology | 1 Processor, 8 Cores |
Identifier | ARM implementer 78 architecture 8 variant 2 part 0 revision 1 |
Base Frequency | 0 MHz |
Cluster 1 | 8 Cores |
Memory Information | |
---|---|
Size | 6.83 GB |
Inference Information | |
---|---|
Framework | TensorFlow Lite |
Backend | GPU |
Device | Default |
Workload | Accuracy | Score | |
---|---|---|---|
Image Classification (F32)
|
100% |
252
47.1 IPS |
|
Image Classification (F16)
|
100% |
379
70.9 IPS |
|
Image Classification (I8)
|
97% |
379
71.0 IPS |
|
Image Segmentation (F32)
|
100% |
298
4.98 IPS |
|
Image Segmentation (F16)
|
100% |
500
8.35 IPS |
|
Image Segmentation (I8)
|
98% |
414
6.91 IPS |
|
Pose Estimation (F32)
|
100% |
384
0.46 IPS |
|
Pose Estimation (F16)
|
100% |
721
0.87 IPS |
|
Pose Estimation (I8)
|
100% |
716
0.87 IPS |
|
Object Detection (F32)
|
100% |
215
16.1 IPS |
|
Object Detection (F16)
|
100% |
318
23.7 IPS |
|
Object Detection (I8)
|
65% |
302
22.6 IPS |
|
Face Detection (F32)
|
100% |
700
8.33 IPS |
|
Face Detection (F16)
|
98% |
1149
13.7 IPS |
|
Face Detection (I8)
|
85% |
1076
12.8 IPS |
|
Depth Estimation (F32)
|
100% |
542
4.20 IPS |
|
Depth Estimation (F16)
|
100% |
958
7.43 IPS |
|
Depth Estimation (I8)
|
94% |
938
7.27 IPS |
|
Style Transfer (F32)
|
100% |
1250
1.64 IPS |
|
Style Transfer (F16)
|
100% |
2228
2.93 IPS |
|
Style Transfer (I8)
|
98% |
2194
2.89 IPS |
|
Image Super-Resolution (F32)
|
100% |
266
9.50 IPS |
|
Image Super-Resolution (F16)
|
100% |
481
17.2 IPS |
|
Image Super-Resolution (I8)
|
98% |
452
16.2 IPS |
|
Text Classification (F32)
|
100% |
103
148.4 IPS |
|
Text Classification (F16)
|
100% |
103
148.1 IPS |
|
Text Classification (I8)
|
92% |
44
63.5 IPS |
|
Machine Translation (F32)
|
100% |
212
3.89 IPS |
|
Machine Translation (F16)
|
100% |
209
3.84 IPS |
|
Machine Translation (I8)
|
64% |
78
1.43 IPS |