Geekbench ML Score

1158
Core ML Neural Engine Inference Score
Geekbench ML 0.6.0 for iOS AArch64

Result Information

Upload Date April 26 2024 08:37 AM
Views 1

System Information

System Information
Operating System iOS 17.4.1
Model iPhone XS Max
Model IDiPhone11,6
Motherboard D331pAP
CPU Information
Name Apple A12 Bionic
Topology 1 Processor, 6 Cores
Identifier ARM
Base Frequency 2.49 GHz
Cluster 12 Cores
Cluster 24 Cores
L1 Instruction Cache48.0 KB x 1
L1 Data Cache32.0 KB x 1
L2 Cache2.00 MB x 1
Memory Information
Size 3.67 GB
Inference Information
Framework Core ML
Backend Neural Engine
Device Default

Inference Performance

Workload Accuracy Score
Image Classification (F32)
100% 191
35.7 IPS
 
Image Classification (F16)
100% 1996
373.5 IPS
 
Image Classification (I8)
99% 1995
373.3 IPS
 
Image Segmentation (F32)
100% 169
2.82 IPS
 
Image Segmentation (F16)
100% 259
4.32 IPS
 
Image Segmentation (I8)
100% 257
4.30 IPS
 
Pose Estimation (F32)
100% 1570
1.90 IPS
 
Pose Estimation (F16)
100% 22725
27.5 IPS
 
Pose Estimation (I8)
100% 22118
26.8 IPS
 
Object Detection (F32)
100% 455
33.9 IPS
 
Object Detection (F16)
100% 421
31.4 IPS
 
Object Detection (I8)
97% 440
32.9 IPS
 
Face Detection (F32)
100% 1730
20.6 IPS
 
Face Detection (F16)
98% 2265
26.9 IPS
 
Face Detection (I8)
97% 2228
26.5 IPS
 
Depth Estimation (F32)
100% 1750
13.6 IPS
 
Depth Estimation (F16)
100% 17773
137.8 IPS
 
Depth Estimation (I8)
99% 17614
136.6 IPS
 
Style Transfer (F32)
100% 4398
5.79 IPS
 
Style Transfer (F16)
100% 5326
7.01 IPS
 
Style Transfer (I8)
100% 5287
6.96 IPS
 
Image Super-Resolution (F32)
100% 854
30.5 IPS
 
Image Super-Resolution (F16)
100% 3627
129.5 IPS
 
Image Super-Resolution (I8)
100% 3690
131.8 IPS
 
Text Classification (F32)
100% 161
231.2 IPS
 
Text Classification (F16)
100% 211
303.0 IPS
 
Text Classification (I8)
96% 209
300.9 IPS
 
Machine Translation (F32)
100% 144
2.65 IPS
 
Machine Translation (F16)
100% 143
2.63 IPS
 
Machine Translation (I8)
99% 145
2.67 IPS
 

Compare

Set Baseline Geekbench 6 now available