Geekbench AI Score

3404
Single Precision Score
1482
Half Precision Score
4857
Quantized Score
Geekbench AI 1.3.0 Corporate for Windows AVX2

Result Information

User XingGuobin
Upload Date April 15 2025 12:30 PM
Views 2

System Information

AI Information
Framework ONNX
Backend CPU
Device AMD Ryzen 7 5800
System Information
Operating System Microsoft Windows 11 (64-bit)
Model LENOVO 90RC00A5CP
Motherboard LENOVO 3716
Power PlanLenovo
CPU Information
Name AMD Ryzen 7 5800
Topology 1 Processor, 8 Cores, 16 Threads
Identifier AuthenticAMD Family 25 Model 33 Stepping 0
Base Frequency 3.40 GHz
Cluster 18 Cores
Memory Information
Size 64.00 GB

AI Performance

Workload Accuracy Score
Image Classification (SP)
100% 2597
483.0 IPS
 
Image Classification (HP)
100% 571
106.1 IPS
 
Image Classification (Q)
96% 4917
918.2 IPS
 
Image Segmentation (SP)
100% 1863
30.2 IPS
 
Image Segmentation (HP)
100% 811
13.1 IPS
 
Image Segmentation (Q)
97% 2654
43.2 IPS
 
Pose Estimation (SP)
100% 4649
5.42 IPS
 
Pose Estimation (HP)
100% 4032
4.70 IPS
 
Pose Estimation (Q)
86% 11073
13.1 IPS
 
Object Detection (SP)
100% 2403
190.6 IPS
 
Object Detection (HP)
100% 734
58.2 IPS
 
Object Detection (Q)
58% 2971
324.6 IPS
 
Face Detection (SP)
100% 5315
63.2 IPS
 
Face Detection (HP)
100% 1566
18.6 IPS
 
Face Detection (Q)
96% 8272
98.7 IPS
 
Depth Estimation (SP)
100% 6536
50.4 IPS
 
Depth Estimation (HP)
99% 2164
16.7 IPS
 
Depth Estimation (Q)
78% 10356
82.6 IPS
 
Style Transfer (SP)
100% 12393
15.9 IPS
 
Style Transfer (HP)
100% 10440
13.4 IPS
 
Style Transfer (Q)
87% 13484
17.5 IPS
 
Image Super-Resolution (SP)
100% 2260
83.4 IPS
 
Image Super-Resolution (HP)
100% 1861
68.7 IPS
 
Image Super-Resolution (Q)
97% 2937
108.8 IPS
 
Text Classification (SP)
100% 1606
2.14 KIPS
 
Text Classification (HP)
100% 752
1.00 KIPS
 
Text Classification (Q)
97% 1467
1.97 KIPS
 
Machine Translation (SP)
100% 2475
42.6 IPS
 
Machine Translation (HP)
100% 756
13.0 IPS
 
Machine Translation (Q)
68% 3422
66.1 IPS
 

Compare

Set Baseline Geekbench 6 now available