Geekbench AI Score

5406
Single Precision Score
30578
Half Precision Score
33661
Quantized Score
Geekbench AI 1.3.0 for macOS AArch64

Result Information

User marcisme
Upload Date May 16 2025 05:16 AM
Views 7

System Information

AI Information
Framework Core ML
Backend Neural Engine
Device Apple M3 Ultra
System Information
Operating System macOS 15.5 (Build 24F74)
Model Mac15,14
Model IDMac15,14
Motherboard Mac15,14
CPU Information
Name Apple M3 Ultra
Topology 1 Processor, 32 Cores
Identifier Apple M3 Ultra
Base Frequency 3.99 GHz
Cluster 124 Cores
Cluster 28 Cores
Memory Information
Size 512.00 GB

AI Performance

Workload Accuracy Score
Image Classification (SP)
100% 1694
315.0 IPS
 
Image Classification (HP)
100% 15778
2.93 KIPS
 
Image Classification (Q)
97% 17478
3.26 KIPS
 
Image Segmentation (SP)
100% 2085
33.8 IPS
 
Image Segmentation (HP)
100% 22103
358.3 IPS
 
Image Segmentation (Q)
99% 26764
433.9 IPS
 
Pose Estimation (SP)
100% 23390
27.3 IPS
 
Pose Estimation (HP)
100% 121003
141.2 IPS
 
Pose Estimation (Q)
95% 119461
140.0 IPS
 
Object Detection (SP)
100% 2136
169.4 IPS
 
Object Detection (HP)
100% 18947
1.50 KIPS
 
Object Detection (Q)
91% 18964
1.52 KIPS
 
Face Detection (SP)
100% 4477
53.2 IPS
 
Face Detection (HP)
100% 38339
455.5 IPS
 
Face Detection (Q)
100% 50432
599.2 IPS
 
Depth Estimation (SP)
100% 7807
60.1 IPS
 
Depth Estimation (HP)
99% 103784
799.6 IPS
 
Depth Estimation (Q)
85% 120378
941.2 IPS
 
Style Transfer (SP)
100% 38005
48.9 IPS
 
Style Transfer (HP)
100% 218891
281.4 IPS
 
Style Transfer (Q)
96% 216820
279.8 IPS
 
Image Super-Resolution (SP)
100% 6965
257.2 IPS
 
Image Super-Resolution (HP)
100% 41811
1.54 KIPS
 
Image Super-Resolution (Q)
99% 50966
1.88 KIPS
 
Text Classification (SP)
100% 2699
3.60 KIPS
 
Text Classification (HP)
100% 4936
6.59 KIPS
 
Text Classification (Q)
93% 4882
6.55 KIPS
 
Machine Translation (SP)
100% 4839
83.4 IPS
 
Machine Translation (HP)
100% 4973
85.7 IPS
 
Machine Translation (Q)
98% 5381
93.0 IPS
 

Compare

Set Baseline Geekbench 6 now available