A benchmark review that thoroughly compares Apple's 'M2' chip and AMD's 'Ryzen 7 PRO 6850U' is released



Prepare the ``

MacBook Air '' equipped with the next-generation silicon `` M2 '' chip announced by Apple in June 2022 and the ThinkPad X13 Gen 3 equipped with AMD's `` Ryzen 7 PRO 6850U '', and combine the M2 chip and Ryzen Technology media Phoronix has published a review comparing the benchmark scores of the 7 PRO 6850U.

Apple M2 vs. AMD Ryzen 7 PRO 6850U Performance In Nearly 200 Benchmarks - Phoronix
https://www.phoronix.com/review/apple-m2-amd-ryzen

Phoronix adopted Arch Linux , which is Linux that runs on Apple Silicon, in order to match the conditions of the M2-equipped MacBook Air and ThinkPad X13 Gen 3 as much as possible when performing benchmark tests. Also, in order to maximize the performance of each CPU, set the ACPI profile, which is a standard for power management and hardware configuration management, to 'performance mode' and CPU frequency scaling to 'performance priority'. doing. In addition, since the M2-equipped MacBook Air does not support Linux power / temperature sensors, ``We cannot accurately compare the performance per watt of the M2 and Ryzen 7 PRO 6850U,'' Phoronix wrote. .

The specifications of the M2-equipped MacBook Air and ThinkPad X13 Gen 3 that ran the benchmark test are as follows. The difference in specifications is that the M2-equipped MacBook Air has 8 GB of memory (RAM) and 251 GB of storage (ROM), while the ThinkPad X13 Gen 3 has 16 GB of RAM and 512 GB of ROM.



◆Etcpak 1.0
The score when benchmarking with

Etcpak 1.0 , an open source benchmark software, is as follows. The higher the number, the better the score, and 'Max Perf' means 'maximum performance'.

Ryzen 7 PRO 6850U wins in multithread performance comparison



Ryzen 7 PRO 6850U records more than double the score difference even in single thread performance



◆ Leela Chess Zero 0.28
A graph that scores performance when running version 0.28 of

Leela Chess Zero , an open source neural network-based chess engine, on each device. A higher number means better performance.

Here the MacBook Air with M2 recorded a better score



◆LAMMPS Molecular Dynamics Simulator
The graph below scores the performance of LAMMPS , a molecular dynamics program at Sandia National Laboratories. A higher number means better performance.

This is the overwhelming victory of Ryzen 7 PRO 6850U



◆WebP Image Encode 1.1
The graph below scores the time it takes to encode an image into

WebP , an image format created by Google. A lower number means better performance.

As a result, Ryzen 7 PRO 6850U is superior, although it is a small difference.



◆simdjson 2.0
The graph below scores the data transfer speed when using

simdjson , a high-speed JSON analysis library. A higher number means better performance.

Although we tested it in two ways, the result was that the Ryzen 7 PRO 6850U was superior in both formats.




◆Java Gradle Build
The graph below scores the time it takes to build a Java application with

Gradle . A lower number means better performance.

MacBook Air with M2 wins by a small margin



◆DaCapo Benchmark 9.12-MR1
The graph below shows the score when using the benchmark software '

DaCapo Benchmark '. A lower number means better performance.

Java Test: H2



Java Test: Jython



Java Test: Tradebeans



◆ Renaissance 0.14
The graph below shows the score when using the benchmark software '

Renaissance '. A lower number means better performance.

Random Forest



Apache Spark ALS



Finagle HTTP Requests



◆Zstd Compression 1.5.0
The graph below shows the data compression speed when using the real-time compression algorithm

Zstd . A higher number means better performance.

Compression speed at compression level 19



Decompression speed at compression level 19



◆GNU Radio
The graph below shows the data transfer speed when using '

GNU Radio ' which is free software for providing signal processing blocks for implementing software radio. A higher number means better performance.

When passing through 5 FIR filters in succession



When tested with a signal source



When passed through an FIR filter



◆Vp9 libvpx Encoding 1.10.0
The graph below shows the data compression speed when compressing a video using VP9 , a video compression codec developed by Google. A higher number means better performance.

When compressing 4K video



When compressing 1080p video



x265 3.4
The graph below shows the data compression speed when compressing 4K video using the video compression standard

H.265 . A higher number means better performance.



◆Coremark 1.0
The graph below shows the score when using the benchmark software '

Coremark '. A higher number means better performance.



◆libavifavifenc 0.10
The graph below summarizes the time it takes to create an AVIF file using

libavif . A lower number means better performance.

When the setting is encoder speed 6



If the settings are encoder speed 6 and lossless



When the setting is encoder speed 10 and lossless



◆Build2 0.13
The graph below summarizes the time required to compile using

build2 , an open source cross-platform build tool. A lower number means better performance.



◆POV-Ray 3.7.0.7
The graph below summarizes the time required for tracing when using the ray tracing software '

POV-Ray '. A lower number means better performance.



◆Prime Sieve 8.0
The graph below summarizes the processing time when using

Primesieve , a C++ library for building Python. A lower number means better performance.



◆ Numpy Benchmark
The graph below summarizes the

NumPy Benchmark score of the airspeed benchmark. A higher number means better performance.



◆Ngspice 34
The graph below summarizes the processing time of the open source electric and electronic circuit simulator

Ngspice . A lower number means better performance.

Circuit: C2670



Circuit: C7552



◆ OpenSSL 3.0
The graph below summarizes the data transfer speed when calculating the hash value of SHA256 with the open source library OpenSSL version 3.0. A higher number means better performance.



◆Node.js V8 Web Tooling Benchmark
The graph below summarizes the scores of the benchmark software

Node.js V8 Web Tooling Benchmark that can test the JavaScript performance of systems using Node.js. A higher number means better performance.



◆Liquid-DSP 2021.01.31
The graph below summarizes the data processing speed when using the open source signal processing library

Liquid-DSP for software radio written in C language. A higher number means better performance.

Thread: 2, Buffer length: 256, Filter length: 57



Thread: 8, Buffer Length: 256, Filter Length: 57



◆TensorFlow Lite 2022-05-18
The graph below summarizes the time taken to process data when using

TensorFlow Lite , a lightweight version of the TensorFlow model for mobile. A lower number means better performance.

Model: SqueezeNet



Model: Mobilenet Quant



Model: Inception ResNet V2



◆Darktable 4.0.0
The graph below summarizes the time taken for data processing when using

Darktable , an open source photo workflow application. A lower number means better performance.

Test: Boat



Test: server room



◆GEGL
The graph below summarizes the time required for data processing when using the programming library

GEGL developed for image processing applications. A lower number means better performance.

Execution processing: Crop



Execution processing: cartoon conversion



Execution processing: Reflection


Execution processing: anti-aliasing



Execution processing: color enhancement



◆ GIMP 2.10.32
The graph below summarizes the time taken for data processing when using the photo retouching software

GIMP . A lower number means better performance.

Execution processing: Resize



Execution processing: rotation



Execution processing: automatic level correction



◆Inkscape
The graph below summarizes the time it takes to convert an SVG file to a PNG file using

Inkscape , a vector image editing software. A lower number means better performance.



◆GNU Octave Benchmark 7.2.0
The graph below summarizes the results of measuring the time it takes to call a C ++ executable file from GNU Octave using the benchmark software

GNU Octave Benchmark . A lower number means better performance.



◆ PyBench 2018-02-16
The graph below summarizes the results of measuring the average test time of various functions with Python benchmark software

PyBench . A lower number means better performance.



Among the 190 types of benchmark tests that were conducted, the ones with the highest scores were classified as follows. Ryzen 7 PRO 6850U (yellow-green) 20 types (10.5%), M2 (orange) 26 types (13.7%), M2 maximum performance (purple) 67 types (35.3%), Ryzen 7 PRO 6850U maximum performance ( Blue-green) is 77 types (40.5%).



Taking the average value of all benchmark tests is as follows. A higher number means better performance.



In addition, Phoronix wrote that how close the M2-equipped MacBook Air can be to the score of Ryzen 7 PRO 6850U, ``whether the target software is optimized for AArch64 '' and ``depends heavily on a subset of workloads.'' , and points out that the low score is partly due to the limited Linux support in Apple Silicon.

in Software,   Hardware, Posted by logu_ii