Geekbench ML Score

7455
Core ML Neural Engine Inference Score
Geekbench ML 0.6.0 for iOS AArch64

Result Information

Upload Date January 07 2024 07:02 PM
Views 1

System Information

System Information
Operating System iOS 17.3
Model iPad14,6
Model IDiPad14,6
Motherboard J621AP
CPU Information
Name ARM
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 15.33 GB
Inference Information
Framework Core ML
Backend Neural Engine
Device Default

Inference Performance

Workload Accuracy Score
Image Classification (F32)
100% 3393
635.0 IPS
 
Image Classification (F16)
100% 6850
1.28 KIPS
 
Image Classification (I8)
99% 7103
1.33 KIPS
 
Image Segmentation (F32)
100% 1330
22.2 IPS
 
Image Segmentation (F16)
100% 1490
24.9 IPS
 
Image Segmentation (I8)
100% 1509
25.2 IPS
 
Pose Estimation (F32)
100% 22757
27.6 IPS
 
Pose Estimation (F16)
100% 81022
98.1 IPS
 
Pose Estimation (I8)
100% 85596
103.7 IPS
 
Object Detection (F32)
100% 2525
188.5 IPS
 
Object Detection (F16)
100% 4106
306.5 IPS
 
Object Detection (I8)
97% 2481
185.2 IPS
 
Face Detection (F32)
100% 9119
108.4 IPS
 
Face Detection (F16)
97% 12279
146.0 IPS
 
Face Detection (I8)
96% 14025
166.8 IPS
 
Depth Estimation (F32)
100% 16220
125.8 IPS
 
Depth Estimation (F16)
100% 47532
368.6 IPS
 
Depth Estimation (I8)
37% 40786
316.3 IPS
 
Style Transfer (F32)
100% 39612
52.1 IPS
 
Style Transfer (F16)
100% 133492
175.6 IPS
 
Style Transfer (I8)
100% 133432
175.5 IPS
 
Image Super-Resolution (F32)
100% 8130
290.3 IPS
 
Image Super-Resolution (F16)
100% 16804
600.1 IPS
 
Image Super-Resolution (I8)
100% 19369
691.7 IPS
 
Text Classification (F32)
100% 1116
1.60 KIPS
 
Text Classification (F16)
100% 717
1.03 KIPS
 
Text Classification (I8)
96% 727
1.04 KIPS
 
Machine Translation (F32)
100% 769
14.2 IPS
 
Machine Translation (F16)
100% 2546
46.8 IPS
 
Machine Translation (I8)
99% 709
13.0 IPS
 

Compare

Set Baseline Geekbench 6 now available