Geekbench AI Score

1843
Single Precision Score
732
Half Precision Score
1925
Quantized Score
Geekbench AI 1.2.0 for Windows AVX2

Result Information

User TiggerTH
Upload Date March 05 2025 01:37 PM
Views 13

System Information

AI Information
Framework ONNX
Backend CPU
Device Intel Xeon E5-2670 v3
System Information
Operating System Microsoft Windows 10 Pro (64-bit)
Model Intel X99
Motherboard Intel X99H
Power PlanBitsum Highest Performance
CPU Information
Name Intel Xeon E5-2670 v3
Topology 1 Processor, 12 Cores, 24 Threads
Identifier GenuineIntel Family 6 Model 63 Stepping 2
Base Frequency 2.30 GHz
Cluster 112 Cores
Memory Information
Size 24.00 GB

AI Performance

Workload Accuracy Score
Image Classification (SP)
100% 2261
420.5 IPS
 
Image Classification (HP)
100% 258
48.0 IPS
 
Image Classification (Q)
96% 2145
400.5 IPS
 
Image Segmentation (SP)
100% 931
15.1 IPS
 
Image Segmentation (HP)
100% 444
7.20 IPS
 
Image Segmentation (Q)
97% 1151
18.7 IPS
 
Pose Estimation (SP)
100% 3115
3.63 IPS
 
Pose Estimation (HP)
100% 2447
2.86 IPS
 
Pose Estimation (Q)
86% 5249
6.20 IPS
 
Object Detection (SP)
100% 1213
96.2 IPS
 
Object Detection (HP)
100% 325
25.7 IPS
 
Object Detection (Q)
58% 1323
144.6 IPS
 
Face Detection (SP)
100% 2283
27.1 IPS
 
Face Detection (HP)
100% 731
8.68 IPS
 
Face Detection (Q)
96% 3604
43.0 IPS
 
Depth Estimation (SP)
100% 4129
31.8 IPS
 
Depth Estimation (HP)
99% 994
7.66 IPS
 
Depth Estimation (Q)
78% 4802
38.3 IPS
 
Style Transfer (SP)
100% 7438
9.56 IPS
 
Style Transfer (HP)
100% 5995
7.71 IPS
 
Style Transfer (Q)
87% 6635
8.63 IPS
 
Image Super-Resolution (SP)
100% 1274
47.0 IPS
 
Image Super-Resolution (HP)
100% 1040
38.4 IPS
 
Image Super-Resolution (Q)
97% 1380
51.2 IPS
 
Text Classification (SP)
100% 751
1.00 KIPS
 
Text Classification (HP)
100% 344
458.8 IPS
 
Text Classification (Q)
97% 227
304.6 IPS
 
Machine Translation (SP)
100% 849
14.6 IPS
 
Machine Translation (HP)
100% 312
5.38 IPS
 
Machine Translation (Q)
68% 1136
21.9 IPS
 

Compare

Set Baseline Geekbench 6 now available