Geekbench ML Score

3378
Core ML GPU Inference Score
Geekbench ML 0.6.0 for iOS AArch64

Result Information

Upload Date May 08 2024 02:37 PM
Views 1

System Information

System Information
Operating System iOS 17.5
Model iPad Pro (11-inch 3rd generation)
Model IDiPad13,4
Motherboard J517AP
CPU Information
Name Apple M1
Topology 1 Processor, 8 Cores
Identifier ARM
Base Frequency 3.19 GHz
Cluster 14 Cores
Cluster 24 Cores
L1 Instruction Cache128 KB x 1
L1 Data Cache64.0 KB x 1
L2 Cache4.00 MB x 1
Memory Information
Size 7.33 GB
Inference Information
Framework Core ML
Backend GPU
Device Default

Inference Performance

Workload Accuracy Score
Image Classification (F32)
100% 1891
353.8 IPS
 
Image Classification (F16)
100% 1201
224.8 IPS
 
Image Classification (I8)
99% 1230
230.2 IPS
 
Image Segmentation (F32)
100% 833
13.9 IPS
 
Image Segmentation (F16)
100% 883
14.7 IPS
 
Image Segmentation (I8)
100% 881
14.7 IPS
 
Pose Estimation (F32)
100% 16041
19.4 IPS
 
Pose Estimation (F16)
100% 17896
21.7 IPS
 
Pose Estimation (I8)
100% 17629
21.3 IPS
 
Object Detection (F32)
100% 1386
103.5 IPS
 
Object Detection (F16)
100% 1545
115.3 IPS
 
Object Detection (I8)
97% 1521
113.6 IPS
 
Face Detection (F32)
100% 5918
70.4 IPS
 
Face Detection (F16)
98% 7260
86.3 IPS
 
Face Detection (I8)
97% 7224
85.9 IPS
 
Depth Estimation (F32)
100% 11901
92.3 IPS
 
Depth Estimation (F16)
100% 13588
105.4 IPS
 
Depth Estimation (I8)
37% 12449
96.5 IPS
 
Style Transfer (F32)
100% 27502
36.2 IPS
 
Style Transfer (F16)
100% 30738
40.4 IPS
 
Style Transfer (I8)
100% 31731
41.7 IPS
 
Image Super-Resolution (F32)
100% 4812
171.8 IPS
 
Image Super-Resolution (F16)
100% 5005
178.7 IPS
 
Image Super-Resolution (I8)
100% 5836
208.4 IPS
 
Text Classification (F32)
100% 1173
1.69 KIPS
 
Text Classification (F16)
100% 767
1.10 KIPS
 
Text Classification (I8)
96% 762
1.09 KIPS
 
Machine Translation (F32)
100% 529
9.74 IPS
 
Machine Translation (F16)
100% 535
9.85 IPS
 
Machine Translation (I8)
99% 534
9.83 IPS
 

Compare

Set Baseline Geekbench 6 now available