Geekbench AI Score

2641
Single Precision Score
12850
Half Precision Score
13612
Quantized Score
Geekbench AI 1.0.0 for macOS AArch64

Result Information

User mprimeaux
Upload Date August 15 2024 05:27 PM
Views 6

System Information

AI Information
Framework Core ML
Backend Neural Engine
Device Apple M2 Ultra
System Information
Operating System macOS 14.6.1 (Build 23G93)
Model Mac Studio (2023)
Model IDMac14,14
Motherboard Mac14,14
CPU Information
Name Apple M2 Ultra
Topology 1 Processor, 24 Cores
Identifier Apple M2 Ultra
Base Frequency 3.58 GHz
Cluster 116 Cores
Cluster 28 Cores
Memory Information
Size 192.00 GB

AI Performance

Workload Accuracy Score
Image Classification (SP)
100% 1119
208.1 IPS
 
Image Classification (HP)
100% 8211
1.53 KIPS
 
Image Classification (Q)
99% 8661
1.62 KIPS
 
Image Segmentation (SP)
100% 780
12.6 IPS
 
Image Segmentation (HP)
100% 5122
83.0 IPS
 
Image Segmentation (Q)
100% 5167
83.8 IPS
 
Pose Estimation (SP)
100% 16326
19.0 IPS
 
Pose Estimation (HP)
100% 93315
108.9 IPS
 
Pose Estimation (Q)
94% 88831
104.2 IPS
 
Object Detection (SP)
100% 1273
101.0 IPS
 
Object Detection (HP)
99% 9687
770.6 IPS
 
Object Detection (Q)
93% 10222
815.6 IPS
 
Face Detection (SP)
100% 2078
24.7 IPS
 
Face Detection (HP)
100% 11940
141.9 IPS
 
Face Detection (Q)
100% 13978
166.1 IPS
 
Depth Estimation (SP)
100% 4201
32.4 IPS
 
Depth Estimation (HP)
99% 56793
437.6 IPS
 
Depth Estimation (Q)
97% 61778
477.7 IPS
 
Style Transfer (SP)
100% 14616
18.8 IPS
 
Style Transfer (HP)
100% 147423
189.5 IPS
 
Style Transfer (Q)
100% 147751
189.9 IPS
 
Image Super-Resolution (SP)
100% 3667
135.4 IPS
 
Image Super-Resolution (HP)
100% 21007
775.7 IPS
 
Image Super-Resolution (Q)
98% 28119
1.04 KIPS
 
Text Classification (SP)
100% 998
1.33 KIPS
 
Text Classification (HP)
100% 725
967.4 IPS
 
Text Classification (Q)
93% 709
951.5 IPS
 
Machine Translation (SP)
100% 1951
33.6 IPS
 
Machine Translation (HP)
96% 2122
36.7 IPS
 
Machine Translation (Q)
98% 2113
36.5 IPS
 

Compare

Set Baseline Geekbench 6 now available