Geekbench AI Score

448
Single Precision Score
430
Half Precision Score
797
Quantized Score
Geekbench AI 1.1.0 for Android AArch64

Result Information

User MNS_UA
Upload Date October 10 2024 06:03 AM
Views 10

System Information

AI Information
Framework TensorFlow Lite
Backend CPU
Device ARM MT6785V/CD
System Information
Operating System Android 13
Model Xiaomi M2101K7BNY
Model IDXiaomi M2101K7BNY
Motherboard rosemary
CPU Information
Name ARM MT6785V/CD
Topology 1 Processor, 8 Cores
Identifier ARM implementer 65 architecture 8 variant 3 part 3339 revision 0
Base Frequency 2.00 GHz
Cluster 16 Cores @ 2.00 GHz
Cluster 22 Cores @ 2.05 GHz
Memory Information
Size 5.46 GB

AI Performance

Workload Accuracy Score
Image Classification (SP)
100% 195
36.3 IPS
 
Image Classification (HP)
100% 198
36.8 IPS
 
Image Classification (Q)
100% 529
98.5 IPS
 
Image Segmentation (SP)
100% 241
3.91 IPS
 
Image Segmentation (HP)
100% 243
3.94 IPS
 
Image Segmentation (Q)
98% 642
10.4 IPS
 
Pose Estimation (SP)
100% 950
1.11 IPS
 
Pose Estimation (HP)
100% 946
1.10 IPS
 
Pose Estimation (Q)
84% 1341
1.59 IPS
 
Object Detection (SP)
100% 193
15.3 IPS
 
Object Detection (HP)
100% 201
16.0 IPS
 
Object Detection (Q)
83% 551
44.5 IPS
 
Face Detection (SP)
100% 646
7.68 IPS
 
Face Detection (HP)
100% 511
6.07 IPS
 
Face Detection (Q)
95% 1198
14.3 IPS
 
Depth Estimation (SP)
100% 851
6.56 IPS
 
Depth Estimation (HP)
99% 659
5.08 IPS
 
Depth Estimation (Q)
64% 1343
12.3 IPS
 
Style Transfer (SP)
89% 1569
2.04 IPS
 
Style Transfer (HP)
89% 1627
2.11 IPS
 
Style Transfer (Q)
98% 3315
4.27 IPS
 
Image Super-Resolution (SP)
100% 433
16.0 IPS
 
Image Super-Resolution (HP)
100% 433
16.0 IPS
 
Image Super-Resolution (Q)
97% 568
21.1 IPS
 
Text Classification (SP)
100% 240
320.1 IPS
 
Text Classification (HP)
100% 239
319.2 IPS
 
Text Classification (Q)
88% 442
595.8 IPS
 
Machine Translation (SP)
100% 428
7.38 IPS
 
Machine Translation (HP)
100% 417
7.18 IPS
 
Machine Translation (Q)
50% 309
11.0 IPS
 

Compare

Set Baseline Geekbench 6 now available