Geekbench AI Score

1180
Single Precision Score
555
Half Precision Score
1022
Quantized Score
Geekbench AI 1.5.0 for Windows AVX2

Result Information

User iambjlu
Upload Date November 06 2025 04:45 PM
Views 8

System Information

AI Information
Framework ONNX
Backend CPU
Device Intel Xeon E5-2650 v4
System Information
Operating System Microsoft Windows 10 (64-bit)
Model VMware, Inc. VMware7,1
Motherboard Intel Corporation 440BX Desktop Reference Platform
CPU Information
Name Intel Xeon E5-2650 v4
Topology 2 Processors, 4 Cores
Identifier GenuineIntel Family 6 Model 79 Stepping 0
Base Frequency 2.20 GHz
Cluster 14 Cores
Memory Information
Size 10.00 GB
Type DRAM

AI Performance

Workload Accuracy Score
Image Classification (SP)
100% 863
160.5 IPS
 
Image Classification (HP)
100% 228
42.4 IPS
 
Image Classification (Q)
96% 837
156.2 IPS
 
Image Segmentation (SP)
100% 545
8.83 IPS
 
Image Segmentation (HP)
100% 328
5.32 IPS
 
Image Segmentation (Q)
97% 779
12.7 IPS
 
Pose Estimation (SP)
100% 1739
2.03 IPS
 
Pose Estimation (HP)
100% 1423
1.66 IPS
 
Pose Estimation (Q)
86% 1820
2.15 IPS
 
Object Detection (SP)
100% 800
63.4 IPS
 
Object Detection (HP)
100% 283
22.4 IPS
 
Object Detection (Q)
58% 577
63.1 IPS
 
Face Detection (SP)
100% 2206
26.2 IPS
 
Face Detection (HP)
100% 657
7.81 IPS
 
Face Detection (Q)
96% 1858
22.2 IPS
 
Depth Estimation (SP)
100% 2162
16.7 IPS
 
Depth Estimation (HP)
99% 790
6.09 IPS
 
Depth Estimation (Q)
78% 2053
16.4 IPS
 
Style Transfer (SP)
100% 4086
5.25 IPS
 
Style Transfer (HP)
100% 3062
3.94 IPS
 
Style Transfer (Q)
87% 3161
4.11 IPS
 
Image Super-Resolution (SP)
100% 783
28.9 IPS
 
Image Super-Resolution (HP)
100% 607
22.4 IPS
 
Image Super-Resolution (Q)
97% 914
33.9 IPS
 
Text Classification (SP)
100% 620
827.7 IPS
 
Text Classification (HP)
100% 320
427.0 IPS
 
Text Classification (Q)
97% 210
281.0 IPS
 
Machine Translation (SP)
100% 850
14.6 IPS
 
Machine Translation (HP)
100% 303
5.22 IPS
 
Machine Translation (Q)
68% 791
15.3 IPS
 

Compare

Set Baseline Geekbench 6 now available