Geekbench ML Score

8954
Core ML Neural Engine Inference Score
Geekbench ML 0.6.0 for iOS AArch64

Result Information

Upload Date August 15 2024 07:02 PM
Views 2

System Information

System Information
Operating System iOS 17.6.1
Model iPad16,3
Model IDiPad16,3
Motherboard J717AP
CPU Information
Name ARM
Topology 1 Processor, 9 Cores
Identifier ARM
Base Frequency 4.41 GHz
Cluster 13 Cores
Cluster 26 Cores
L1 Instruction Cache128 KB x 1
L1 Data Cache64.0 KB x 1
L2 Cache4.00 MB x 1
Memory Information
Size 7.28 GB
Inference Information
Framework Core ML
Backend Neural Engine
Device Default

Inference Performance

Workload Accuracy Score
Image Classification (F32)
100% 4247
794.6 IPS
 
Image Classification (F16)
100% 8543
1.60 KIPS
 
Image Classification (I8)
99% 8835
1.65 KIPS
 
Image Segmentation (F32)
100% 1843
30.8 IPS
 
Image Segmentation (F16)
100% 2306
38.5 IPS
 
Image Segmentation (I8)
100% 2242
37.4 IPS
 
Pose Estimation (F32)
100% 26799
32.5 IPS
 
Pose Estimation (F16)
100% 99132
120.0 IPS
 
Pose Estimation (I8)
100% 100299
121.5 IPS
 
Object Detection (F32)
100% 3541
264.3 IPS
 
Object Detection (F16)
100% 5600
418.1 IPS
 
Object Detection (I8)
97% 3293
245.9 IPS
 
Face Detection (F32)
100% 11233
133.6 IPS
 
Face Detection (F16)
97% 13579
161.5 IPS
 
Face Detection (I8)
96% 13624
162.0 IPS
 
Depth Estimation (F32)
100% 19621
152.2 IPS
 
Depth Estimation (F16)
100% 45449
352.5 IPS
 
Depth Estimation (I8)
37% 45527
353.1 IPS
 
Style Transfer (F32)
100% 52388
68.9 IPS
 
Style Transfer (F16)
100% 154787
203.6 IPS
 
Style Transfer (I8)
100% 151824
199.7 IPS
 
Image Super-Resolution (F32)
100% 9055
323.4 IPS
 
Image Super-Resolution (F16)
100% 19893
710.4 IPS
 
Image Super-Resolution (I8)
100% 20746
740.9 IPS
 
Text Classification (F32)
100% 1486
2.14 KIPS
 
Text Classification (F16)
100% 1437
2.06 KIPS
 
Text Classification (I8)
96% 1058
1.52 KIPS
 
Machine Translation (F32)
100% 661
12.2 IPS
 
Machine Translation (F16)
100% 1815
33.4 IPS
 
Machine Translation (I8)
99% 652
12.0 IPS
 

Compare

Set Baseline Geekbench 6 now available