Stress testing

Running an overclocked PC is fine as long as it is stable and that the temperature of its components do not exceed their acceptable range. There are several programs available to assess system stability through stress testing the system and thereby the overclock level. The steps of overclocking a PC are beyond the scope of this article, but there is pretty inclusive guide written by graysky on the topic: Overclocking guide.

Note: The linked guide is a bit dated. More contemporary guides are recommended for modern hardware.

Stressing Programs

These common stressing programs are listed in two categories: higher demand voltage and medium demand voltage:

Voltage Demand	Program	Description
Medium
	Cc/Gcc	Both cc/gcc compilation is a great method of stress testing. Both are available in the base-devel group.
	HandBrake-cli	handbrake-cli can be used to encode using high quality settings.
	Systester	systesterAUR Systester is a multithreaded piece of software capable of deriving values of pi out to 128,000,000 decimal places. It has built in check for system stability.
High	stress	stress is a simple CPU, memory, I/O, and disk workload generator implemented in C.
	mprime	mprime-binAUR factors large numbers and is an excellent way to stress CPU and memory.
	linpack	linpackAUR - Linpack makes use of the BLAS (Basic Linear Algebra Subprograms) libraries for performing basic vector and matrix operations. and is an excellent way to stress CPUs for stability.

It is recommended to use programs in both categories to assess the overall system stability. It can happen that a system is more sensitive to a test from the medium than from the high demand category. Higher demand voltage programs require the most CPU core voltage (V_CORE) due to intense hardware usage to perform their tasks. Medium demand voltage programs do not always call for the highest V_CORE when running and as such can be more prone to throwing errors for systems that are undervolted relative to the clock speed requested.

Example on an overclocked i7-3770K (4.50 GHz); V_CORE is +0.020 V in offset mode with all power-saving features enabled. This machine running with a V_CORE of +0.005 in offset mode remains stable with both #MPrime and #Linpack for hours, but throws errors under both x264 and gcc after only several minutes:

Idle: 0.7440 V - 0.8320 V (varies).
Mprime small FFTs: 1.2880 V (steady).
Mprime large FFTs: 1.3040 V (steady).
Mprime blend: 1.2960 V (steady).
Linpack: 1.2320 V - 1.2720 V (varies).
x264 encoding: 1.2320 V - 1.2720 V (varies).
gcc compiling: 1.2720 V (steady).

Tip: To ensure the stability of a system, it is recommended to run such tests for a long period of time, from a few hours to a few days, in different temperature and humidity conditions. If the room temperature can for example vary significantly between winter and summer time, this is something to be considered.

Stressing CPU and Memory

stress

stress performs a loop that calculates the square root of a random number in order to stress the CPU. It can run simultaneously several workers to load all the cores of a CPU for example. It can also generate memory, I/O or disk workload depending on the parameters passed. The FAQ provides examples and explanations.

To spawn 4 workers spinning on sqrt(), use the command:

$ stress --cpu 4

MPrime

MPrime (also known as Prime95 in its Windows and MacOS implementation) is recognized universally as one defacto measure of system stability. MPrime under torture test mode will perform a series of very CPU intensive calculations and compare the values it gets to known good values.

The Linux implementation is called mprime^AUR and is available in the AUR.

Warning: Before proceeding, it is HIGHLY recommended that users have some means to monitor the CPU temperature. Packages such as Lm_sensors can do this.

To run mprime, simply open a shell and type "mprime"

$ mprime

Note: If using a cpu-frequency scaler such as cpufrequtils or powernowd sometimes, users need to manually set the processor to run with its highest multiplier because mprime uses a nice value that doesn't always trip the step-up in multiplier.

When the software loads, simply answer 'N' to the first question to begin the torture testing:

Main Menu

1.  Test/Primenet
2.  Test/Worker threads
3.  Test/Status
4.  Test/Continue
5.  Test/Exit
6.  Advanced/Test
7.  Advanced/Time
8.  Advanced/P-1
9.  Advanced/ECM
10.  Advanced/Manual Communication
11.  Advanced/Unreserve Exponent
12.  Advanced/Quit Gimps
13.  Options/CPU
14.  Options/Preferences
15.  Options/Torture Test
16.  Options/Benchmark
17.  Help/About
18.  Help/About PrimeNet Server

There are several options for the torture test (menu option 15).

• Small FFTs (option 1) to stress the CPU
• In-place large FFTs (option 2) to test the CPU and memory controller
• Blend (option 3) is the default and constitutes a hybrid mode which stresses the CPU and RAM.

Errors will be reported should they occur both to stdout and to ~/results.txt for review later. Many do not consider a system as 'stable' unless it can run the Large FFTs for a 24 hour period.

Example ~/results.txt; note that the two runs from 26-June indicate a hardware failure. In this case, due to insufficient vcore to the CPU:

[Sun Jun 26 20:10:35 2011]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Sat Aug 20 10:50:45 2011]
Self-test 480K passed!
Self-test 480K passed!
[Sat Aug 20 11:06:02 2011]
Self-test 128K passed!
Self-test 128K passed!
[Sat Aug 20 11:22:10 2011]
Self-test 560K passed!
Self-test 560K passed!
...

Note: Users suspecting bad memory or memory controllers should try the blend test first as the small FFT test uses very little memory.

Linpack

linpack^AUR makes use of the BLAS (Basic Linear Algebra Subprograms) libraries for performing basic vector and matrix operations. It is an excellent way to stress CPUs for stability (only Intel CPUs are supported). After installation, users should copy /usr/share/linpack/linpack.conf to ~/.config/linpack.conf and adjust it according to the amount of memory on the system.

Systester (AKA SuperPi for Windows)

Systester^AUR is available in the AUR in both cli and gui version. It tests system stability by calculating up to 128 millions of Pi digits and includes error checking. Note that one can select from two different calculation algorithms: Quadratic Convergence of Borwein and Gauss-Legendre. The latter being the same method that the popular SuperPi for Windows uses.

A cli example using 8 threads is given:

$ systester-cli -gausslg 64M -threads 8

Intel Processor Diagnostic Tool

The Intel Processor Diagnostic Tool is a tool that verifies the functionality of an Intel Microprocessor by stress testing the CPU. A Fedora Linux LiveUSB ISO images are available. The LiveUSB image allows you to stress test your machine without using your main operating system; such method might be useful in extreme cases especially when dealing with cold reboots/crashes.

Burn the image to a USB stick by using dd or Gnome Disks and then boot the Live CD. Once booted, open the terminal and type the following command to install Intel Processor Diagnostic Tool for 64-bit machines:

$ install64

Once it is installed, you can run the Diagnostic Tool by clicking on the IPDT Icon that is located on the desktop.

Stressing Memory

Originally a very good program for memory failure detection, Memtest86+ can also be used to stress test the memory. It is based on the well-known original memtest86 written by Chris Brady. Memtest86+ is, like the original, released under the terms of the GNU General Public License (GPL).

Running Memtest86+

Either download and burn the ISO to a CD and boot from it, or install memtest86+ from [extra] and update GRUB which will auto-detect the package and allow users to boot directly to it (only for BIOS). If you are using EFI, use memtest86-efi^AUR (non-free) instead or boot grub with memtest86+ from Arch Linux install image.

Tip: Allowing Memtest86+ to run for >10 cycles without errors is usually sufficient.

Discovering Errors

Some stressing applications like #MPrime or #Linpack have built in consistency checks to discover errors due to non-matching results. A more general and simple method for measuring hardware instabilities can be found in the kernel itself. To use it, simply watch the output from the kernel ring buffer by this command:

# cat /proc/kmsg

The key error to watch for looks like this:

[Hardware Error]: Machine check events logged

The kernel can throw these errors while the stressing application is running, before it ends the calculation and reports the error, thus providing a very sensitive method to assess stability.