[ale] HPC Node Test Util or Suite

[Ron Frazier]

Greg,

I just happen to run a prime number test most of the time which will do 
quite nicely.  You can use it just for stress testing, or you can sign into 
the research network I'm going to refer you to and help find world record 
prime numbers by running their process in the background.  They support: 
WINDOWS,  MAC OS X,  LINUX,  WINDOWS 64-BIT,  LINUX 64-BIT.

Go to: http://www.mersenne.org/

Click on download software and select the appropriate file and download and 
save it.

Here is the link to the download page, which also contains some instructions.

http://www.mersenne.org/freesoft/

The Windows version has it's own GUI interface, but the Linux version runs 
in a terminal.

I'm running Ubuntu 10.04.  It's been a while since I installed it.  It 
doesn't really tell you what to do.  I believe you just unpack the tar ball 
into its own directory.  Make it executable.  Then create a launcher on 
your desktop to activate it.  Tell the launcher you want an application in 
terminal.  I used the following command line, which you can adapt: 
/home/ron/Prime95/mprime -m

When you run that, a terminal window should pop up with a text menu.  Item 
15 will give some options for torture testing.  The terminal window also 
displays ongoing status messages.  To get back to the menu, just hit 
enter.  Once the test is running, you should be able to use option 4 to 
stop the test, and option 5 to quit the program.  This will stress cpu and 
ram, not so much disk.  You can find out more about the features by running 
the Windows version and playing with the menu options.

NOTE: I found out during my research that every cpu and graphics card has 
different temperature tolerances.  The only way to know is to check the 
manufacturer's data sheet.  The computer I'm typing on has an AMD Athlon X2 
chip, which will take up to about 70 deg C.  It's hovering around 46 deg C 
at the moment with this prime number test running.  My old computer, a 
Toshiba laptop of 2002 vintage, with a Pentium 4, will take 70 deg C.  My 
newer Dell laptop, with a Pentium Dual Core, will take 100 deg C.  You have 
to find out what your system will take.  You may need the exact model 
number of the chip.  Some of the system info type utilities may be able to 
get that electronically.  Make sure all the fans are running.  Also, make 
sure there are no dust globs in the heat pipes and radiators.

Not long ago, I was having a problem with my laptop overheating and locking 
up.  I took the cover off and removed the heat pipe very carefully, and 
reluctantly.  I found a large glob of dust on the INSIDE facing side of the 
radiator, blocking about 3/4 of the air flow.  I cleaned out all 1" of the 
(not) huge radiator.  Now my laptop runs at 75 deg C at full throttle, 25 
deg C away from its limit.

You should be able to use lm-sensors to do your temp readings.  On Ubuntu, 
you can put an icon on your display panel to show the temp.  Just make sure 
you read the right sensor.  Some motherboards have a digital display on 
them.  My MSI motherboard reads the CPU temp after it gets through with the 
POST stuff during boot.  That way, I can compare my on screen reading with 
my on motherboard reading.

Hope this helps.  Feel free to contact me personally (see my signature) and 
I'll help you walk through the procedures.  Ubuntu is the only Linux I've 
used however.

PS: You may want to put the machines through a memory test.  Check out 
ultimatebootcd.com and look for memtest86+.  You want to see at least 24-48 
solid hours of memory checking with no errors.  I think the Ubuntu disk may 
have this as well.

Sincerely,

Ron

At 9/14/2010 12:37 PM -0400, you wrote:
>Hey ALErs,
>
>We have a few 1U Dual Xeon HPC nodes going to a National Lab soon and I 
>was wanting to give them a good exercise to see what kind of CPU temps 
>we'll get when they're running flat out. I'm not sure what flavor of Linux 
>the lab runs, but I was thinking of tossing on Rocks. Any suggestions on 
>what to use to put a pair of E-5620 Quad Core CPUs w/ 48GB of RAM through 
>a stress test?
>
>Thanks,
>GC


--------------------------
(PS - If you email me and don't get a quick response, you might want to 
call on the phone.  I get about 300 emails per day from alternate energy 
mailing lists and such.  I don't always see new messages very quickly.)

Ron Frazier

770-205-9422 (O)   Leave a message.
linuxdude AT c3energy.com