search

ctrlok / macdockergeek

This is a cpu performance comparison inside a docker in laptops (with new M1 pro)
5 stars 3 forks source link

MacBook Air - M1, 2020 results #3

Open realJoshByrnes opened 1 month ago

realJoshByrnes commented 1 month ago

I wanted to compare the results for Docker, with three configurations - aarch64 vs x86 (Rosetta) vs x86 (QEMU) on my MacBookAir10,1 as I've been thinking about moving my home NAS which mostly runs x86_64 containers to this laptop.

Running under Colima on a fresh install of Sonoma, I have assigned 4 cores and 4GB RAM. Using Colima rather than Docker Desktop for MacOS is likely why my comparible scores are a little higher.

Single Core

aarch64 QEMU Rosetta 2
Operating System Ubuntu 20.04.3 LTS 6.8.0-31-generic aarch64 Ubuntu 20.04.3 LTS Ubuntu 20.04.3 LTS
Model Apple Inc. Apple Virtualization Generic Platform Apple Inc. Apple Virtualization Generic Platform Apple Inc. Apple Virtualization Generic Platform
Motherboard N/A N/A N/A

CPU Information

aarch64 QEMU Rosetta 2
Name ARMv8 QEMU TCG version 2.5+ VirtualApple
Topology 1 Processor, 1 Core, 4 Threads 1 Processor, 1 Core, 4 Threads 1 Processor, 1 Core, 4 Threads
Base Frequency 0 MHz 0 MHz 0 MHz
L1 Instruction Cache 64.0 KB x 4 128 KB x 4
L1 Data Cache 64.0 KB x 4 128 KB x 4
L2 Cache 512 KB x 4 8.00 MB x 1

Memory Information

aarch64 QEMU Rosetta 2
Memory 3.81 GB 3.81 GB 3.81 GB

Single-Core Performance

aarch64 QEMU Rosetta 2
Single-Core Score 1690 107 1255
Crypto Score 2683 33 1595
Integer Score 1456 142 1203
Floating Point Score 1863 43 1311
AES-XTS 2683 33 1595
Text Compression 1422 471 1269
Image Compression 1326 97 1130
Navigation 1787 477 1634
HTML5 1697 107 1111
SQLite 1334 123 1022
PDF Rendering 1726 131 1324
Text Rendering 1716 129 1221
Clang 1730 172 1236
Camera 1294 29 990
N-Body Physics 1811 87 1450
Rigid Body Physics 1670 45 959
Gaussian Blur 1402 14 1299
Face Detection 1906 39 1178
Horizon Detection 1906 62 1170
Image Inpainting 3274 73 2503
HDR 2638 104 2213
Ray Tracing 2023 53 1654
Structure from Motion 1501 26 1021
Speech Recognition 1478 143 1204
Machine Learning 1152 5 700

Multi-Core Performance

aarch64 QEMU Rosetta 2
Multi-Core Score 5773 391 4334
Crypto Score 9171 127 6169
Integer Score 5204 522 4095
Floating Point Score 6439 152 4547
AES-XTS 9171 127 6169
Text Compression 4066 1597 3758
Image Compression 4933 366 4236
Navigation 4285 1682 4105
HTML5 6085 398 4086
SQLite 4940 449 3887
PDF Rendering 6348 488 4796
Text Rendering 5549 493 3959
Clang 6379 638 4552
Camera 4821 107 3607
N-Body Physics 5746 308 4674
Rigid Body Physics 6266 166 3559
Gaussian Blur 6085 44 4536
Face Detection 8195 143 4362
Horizon Detection 6993 229 4316
Image Inpainting 10384 267 7850
HDR 9793 384 8140
Ray Tracing 8523 190 6967
Structure from Motion 5449 90 3746
Speech Recognition 3565 485 3036
Machine Learning 3753 16 2391
realJoshByrnes commented 1 month ago

I should note that this isn't an issue, just thought I'd post my findings.

The x86_64 results using Rosetta 2 are still significantly faster than my MacBook Pro 15 Inch, 2016 (2.7 GHz Quad-Core Intel Core i7)