Server Performance Evaluation Tool

Server Performance Evaluation Tool allows you to easily test the performance of x86_64 processors on Linux.

Features

• Automatically install system prerequisites using the operating system's package manager.
• Automatic detection of system information.
• Automatic downloading, setting up, and running of all performance tests.
• Overriding available for general tuning of tests and changing test versions.
• Results in JSON and text form.
• All major commands documented in the results text file.

Getting Started

Requirements

1. x86_64 system
2. Linux (Ubuntu, Debian, CentOS, openSUSE) Operating System
   • Mostly tested on Ubuntu
3. Root access
4. Python 3

Setup

pip3 install -r requirements.txt

Run

Once everything is set up, you can simply run Server Performance Evaluation Tool by executing the spet command:

Note: You need root access to run Server Performance Evaluation Tool.

spet

If for some reason, the CLI does not work, you can call Server Performance Evaluation Tool module with the following:

python3 -m spet

Results

Results will be outputted to the console (stdout) as they are gathered. All result files will be located in $HOME/spet_results under your specific run's directory.

If $HOME was not detected for whatever reason, the result files will be located in the Server Performance Evaluation Tool's spet/spet_results directory.

Additional Options

Option	GNU long option	Meaning
-v	--verbose	Show additional information on SPET progress.
-d	--debug	Show debugging information.
-e [..]	--exclude [..]	Exclude the desired benchmark(s). Available options: lmbench, mlc, openssl, compilation, zlib, linpack, stream, nosql, sql, and docker.
-avx512	--avx512	Enable AVX-512 for High-Performance Linpack.

Performance Tests

Purpose	Description
Cryptography	Uses OpenSSL's built-in speed command for testing encryption and decryption on 128- and 256-bit AES-GCM.
Cache Latency	Uses LMbench for measuring L1-, L2-, and L3-cache latencies using lat_mem_rd by measuring memory read latency for varying memory sizes and strides.
Memory Latency	Uses Intel(R) Memory Latency Checker (MLC) for measuring node-to-node latency.
Memory Bandwidth	Uses STREAM for measuring sustainable memory bandwidth and the corresponding computation rate for simple vector kernels in parallel using Open MPI.
Floating-point / Math	Uses High-Performance Linpack for measuring the floating-point rate of execution of the system by running a program that solves a system of linear equations.
Compression / Decompression	Uses zlib for testing the performance of compression and decompression on a 2 GB text file.
Software Development / Compute	Uses the system's build utilities to compile the Linux kernel.
Database SQL	Uses Yahoo! Cloud Serving Benchmark (YCSB) to measure read and update performance on MySQL databases.
Database NoSQL	Uses Yahoo! Cloud Serving Benchmark (YCSB) to measure read and update performance on Cassandra databases.
Containers	Uses Docker to test the performance of compiling the Linux kernel (same test as the software development/compute) on 100 concurrent containers.

Documentation

Visit the docs/ directory for additional documentation.

License

This project is licensed under the MIT License - see the LICENSE file for details.