StressCPU v2 : Gromacs based CPU stability tester.

[toTOW]

I'm setting up this topic to be able to discuss about this interesting program.

What it does : stresses CPU to it's maximum, allowing you to verify it's stability, and your cooling capacities. Based on Gromacs software used by F@H it's one of the most accurate stability tester for a machine aimed at Folding .

How it works : it automatically detects the number of CPU's available in your system, and fully loads them.

This program is available for Windows, Linux and OSX, in 32 and 64bits version.

You can get it here : http://folding.stanford.edu/English/DownloadUtils#ntoc1

How to use it :
- start your monitoring software to keep an eye on the temperatures (CoreTemp or Everest are the most accurate for Windows, and lm-sensors is usually provided with Linux)
- start StressCPU v2

If it detects an error, the program will end with a message.
If your system is overclocked and unstable, you could also see a blue screen of death, or a reset of the machine.

There's no rule about how long you should run it ... but the longer, the better. There is an option (-t hh:mm:ss) to run the program for a limited period. If you don't add this option, it will run until it fails, or you press CTRL + C to manually end it.

NOTE : When you are testing your system, always let your cooling system stabilize. It will be fast with heatsinks and fans, and a bit slower on watercooling. It's the time needed for temperature to rise (in metal or water) and for your fan to spin at their optimal speed if they are controlled by temperature. It might be a good idea to set them at maximum speed while you're trying to overclock.

I hope this post is clearer enough and will help you test your machines ... feel free to post comments or suggestions

Mod Edit: Fixed Broken Link - PantherX

[7im]

I'd recommend a thread name change from "CPU stability tester" to "System stability tester"

A CPU might be able to handle the test, but without proper cooling, the system will crash. Same goes for memory. StressCPU primarily stresses the CPU, but if the memory bad, the test will error out as well.

[toTOW]

This program doesn't stress memory a lot ... it's its only defect I will try to contact the developer to see if it is possible to add operations that use memory.

Until that, this test has to be completed by another software that stresses memory. For example in Windows : Orthos in Large or Blend mode, Everest integrated memory benchmark, 3D games, ... Under Linux, the only thing I"m doing to test the RAM is to launch FAH SMP : if the core fails with a Segmentation fault, the memory is guilty.

[7im]

I don't think a thread about 3rd Party software belongs in the Hardware section of the forum.

[toTOW]

It's to test hardware aimed at folding ...

[MstrBlstr]

It is still a discussion on third party software, not about actual hardware. And is mentioned here in the Tools section already. So, it is more appropriate here.

I did leave a pointer to it in the other catagory though.

[7im]

Having a StressCPU thread in both sections would be like double posting, which is against forum rules.

Thanks for moving this MB.

[gwildperson]

This program doesn't stress memory a lot ... it's its only defect I will try to contact the developer to see if it is possible to add operations that use memory.

Another question for the developer: From a hardware perspective, is it possible that Double Precision (SSE2) stresses certain hardware more than Single Precision (SSE). In most cases I expect that there's no difference, but that may not be universally true. (We have at least one person who appears to have a problem only with FahCore_7b, but I don't know what hardware is involved.)

[noorman]

.

Since SSE and SSE2 are physically on another spot of the die, heating up those sections can have a different effect indeed; could be that one of the two has a slightly worse cooling path and could give you FaH crashes.

I know this because of crashes I had with my A64X2 3800+, which I had OC'ed to 2.6 GHz; at that setting, I got one of the cores who did well and never crashed a (single core client WU), but the other core had troubles and regularly crashed. ( that time both running SSE2 WU's )
Just stepping down the CPU clock to 2.5 GHz was enough to make both rock stable !

I think that the thermal contacts are a major cause of working 100% or being unstable ...

So, there will be luck in to the equation to, luck that the die was fixed very well and evenly & that goes for the top plate of the modern packages.
Careful mounting of your HSF is important too, but any thermal defect closer to the processor die will have more impact !


.

[bruce]

Since SSE and SSE2 are physically on another spot of the die, heating up those sections can have a different effect indeed; could be that one of the two has a slightly worse cooling path and could give you FaH crashes.

The degree with which the SSE hardware is separated from SSE2 depends on the CPU design. Perhaps StressCPU2 needs a double-precision version.

In hardware as old as the P4, the ALU (integer) parts of the CPU were separate from the SSE (and SSE2) portions, and StressCPU was an important step forward. It appears that new chip designs have taken the next step and the software needs to catch up again.

[toTOW]

I think I'll try to contact the developper (or maybe uderstand how it works ...) to add SSE2 benchs and memory operations. These are the only thwo thigs missing in this nice program

[Zebulon84]

Does anyone know if such a tool exist for a GPU ?

Comparing a known Gromac calculation to the expected result is certainly more precise that trying to figure out artifacts on a 3D Demo.

[MtM]

toTow, the readme states that there isn't a 64bit version, yet you say there is.

Code: Select all

1. In most cases you can just use one of the precompiled binaries.

2. To compile, copy one of the makefiles to 'Makefile', and 
   issue 'make' to create the stresscpu2 binary.

3. If you can't get threads to work at all, you can disable them with -DNO_THREADS.
   It works just fine to run multiple copies of the program in parallel instead.

4. There is no 64 bit windows binary since I'm not going to go through the effort to 
   buy & install a 64 bit windows system just to test this program and then throw it
   out again. Complain to Microsoft :-) 
   I've created a 64-bit kernel assembly file for masm, but there are likely a couple of
   typos remaining in it. If you have access to a ml64.exe (64-bit masm) you could do me
   a favor by trying to compile kernel_x86_64_masm_syntax.s and send me the error listing.

Have fun,

Erik Lindahl <lindahl@cbr.su.se>

If I run it on vistaX64, it tell me



Isn't using x86 code not going to stress the cpu properly on x64 windows? Just a heads up, think you should change your opening post?

Edit: running ml64 on the assembly I don't get anything usefull but I'm probably using wrong arguments or wrong version ( diffrent version handle delimiters in another way and the error seems to suggest this but I'm just guessing here ).

Code: Select all

c:\Users\Marvin\Desktop\stresscpu2>ml64 kernel_x86_64_masm_syntax.s /Zs /Zd /c /
Gc
Microsoft (R) Macro Assembler (x86-64) Version 8.00.2207
Copyright (C) Microsoft Corporation.  All rights reserved.

 Assembling: kernel_x86_64_masm_syntax.s
kernel_x86_64_masm_syntax.s(1) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(2) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(3) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(4) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(5) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(6) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(7) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(8) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(9) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(10) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(11) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(12) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(13) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(14) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(15) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(16) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(17) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(18) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(19) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(20) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(21) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(22) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(23) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(24) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(25) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(26) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(27) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(28) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(29) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(30) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(31) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(32) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(33) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(34) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(35) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(36) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(37) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(38) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(39) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(40) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(41) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(42) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(43) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(44) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(45) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(46) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(47) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(48) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(49) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(50) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(51) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(52) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(53) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(54) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(55) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(56) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(57) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(58) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(59) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(60) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(61) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(62) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(63) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(64) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(65) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(66) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(67) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(68) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(69) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(70) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(71) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(72) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(73) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(74) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(75) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(76) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(77) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(78) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(79) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(80) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(81) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(82) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(83) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(84) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(85) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(86) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(87) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(88) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(89) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(90) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(91) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(92) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(93) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(94) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(95) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(96) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(97) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(98) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(99) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(100) : error A2044: invalid character in file
kernel_x86_64_masm_syntax.s(101) : fatal error A1012: error count exceeds 100; s
topping assembly

@Zebulon84, Occt has a a cuda based memtest ( which isn't all saying as it's not shader intensive but it's a start ). Doesn't work for Ati cards though ( as it's cuda ).

[John Naylor]

@MtM

64-bit (in the form Intel [excluding Itanium] and AMD use) is just some hardware extensions and extra registers for x86 processors, they work fundamentally the same way as standard x86 processors... as such any x86 application can run natively on x86-64/x64 hardware and any x86 stress testers will push x86-64 CPUs just as hard as x86 CPUs. This applies regardless of the operating system being run, or the number of bits the operating system can use. (in truth, x64 OSes currently only use 48 bits, but I digress...)

[toTOW]

I did some tests under linux 64 bits, which has both 32 and 64 bits binaries, but I didn't find any differences between the two binaries. The CPU was running at the same temperature, and both were showing the same errors.

The only thing I really miss in this program is the memory test. Without a proper memory test, we can't fully validate a system for SMP clients with this program.