Which tests and for how long ?

M

MrEd

Weaksauce
Joined
Dec 15, 2006
Messages
72
So my dream machine is built, couldn't be happier and everything seems to be running good. Now I'm looking to see just how stable it is before I dabble with overclocking. I've never ran any of the stress stuff before so it's all new to me. I've saw in other posts about different things people metioning running this test or that one for so long and stuff, but it was never really the subject of the post itself. So what is the most common and accepted ones to use and how how long should they be run ? I see all the ones listed in the sticky "Benchmarking/Overclocking", I just don't know which ones to pick and how long to run them.

I downloaded Orthos and have it running now, not sure how long it should go, and I guess memtest86 should be done as well ? I'm looking for two things actually. One is the stuff I should run and for how long to check complete system stability, and two is, the stuff I should run to get scores I guess for benchmarking, mainly to compare to the new ones I would get after trying some overclocking. I know of 3DMark06 for benchmarking, are there others that should be run along with that one ? Thanks in advance for the education, any opinions are appreciated as well !
 
Oh, one more thing. If you guys can hang with me for a bit I'll be posting any info from the tests on this thread so you can tell me if there's something I should be concerned about. Right now Orthos has been running in the background for a couple hours and my CPU core temps are between 47C and 51C. That's using the Blend option and checking it with Core Temp Beta 0.94.
 
I can help. Either here or on MSN. My profile has my msn address.

Orthos should be run at least 6 (preferrably 18) hours to determine stability. Memtest is good overnight. Your temps are either good (if you have a conroe) or a bit high (if you're using an AMD proc).

For benching, I recommend SuperPi 1M (the standard test) to compare your OCing results (provides instant gratification and a good CPU/MEM benchmark) and 3Dmark for your 3D benching needs.
 
Arcygenical:

Thanks friend, that's the ones I'll use then. Orthos has been at it for just over four hours now, I'll let it go thru the night which should get me to about 18 hours. If I have time in the morning I'll try and download memtest and let it run while I'm at work. My chip is the X6800 Conroe underneath a Tuniq Tower 120 so I guess the temps are good which is nice to know. Once the stability stuff is done I'll do the bench stuff hopefully tomorrow night. It just so happens that the MB disc came with a free version of 3DMark 06 which I ran a couple days ago. It came up with 13002. I'm not sure if that's good or bad, I have 2 EVGA 8800GTX's in SLI. I had an issue with temps on them that I was able to help a little but they still seem kinda high to me, I'm still looking for an answer in that area. So anyhow thanks, I'll let you know how things go :)
 
Well, I'm just over 16 hours on Orthos. I don't have time to download and figure out Memtest since I hit the snooze button a couple too many times :D, so since I'm not at 18 hours either, I'll let it go while I'm at work and that should get me to 24.
 
If you have a dual core you need to run 2 instances at the same time, one for each core - I'm sure you can do that in Orthos.
 
Sweet, 26 hours, 23 minutes and no errors ! So I read Windows Memtest was good as Memtest86 and could be run from within windows, is this true or should I stick with Memtest86 ?
 
it's better to run memory diagnostics outside of windows (since nothing will be loaded into memory)


there is a windows memory diagnostic that can be run out of windows. Couple techs I work with like it a lot.

There's also Gold Memory or something that's pretty good

Not sure which one is the best though. I usually just use memtest86
 
Heya ChingChang,

I found the windows one that runs out of windows. I think I'll try them both just for the experience of it, I guess it never hurts to know what folks are talking about. I should have some results by tomorrow.
 
Hmm, windows memtest is running now but when I booted off of the floppy I got a message basically saying I had too much memory and it was not going to test it all. I'm assuming because I have 4 gigs in the comp, I know Windows XP can't use almost half of it, but I bought them in anticipation of Vista for whenever it becomes available and the bugs get worked out of it. The memory, Corsair PC2 8888 was kinda hard to find so when I found it I bought it. Since I got that message would it be best to take a matched pair out and do them one set at a time ?
 
Oh, the other question I had if someone knows was, will Memtest86 test 4 gigs at once or are they both the same in that area.
 
I doubt you're going to be able to test 4 GB of memory all at once. As I understand it, this is not a limitation of any memory testing software, but rather a limitation of CPU/system architecture. Basically there are only so many registers in a CPU where it can store memory addresses, and once you run out of those, your system cannot make use of any more memory because it has no way to label the extra RAM and track its usage.

This is where the big deal with 64-bit computing comes from (or part of it, anyway). If the CPU's registers are 64 bits rather than 32, that's WAY more unique memory addresses it can store and hence way more memory it can use. More complete info at http://en.wikipedia.org/wiki/64-bit.

I'm not 100% sure -- maybe someone who is better-informed can correct or elaborate on this. You might want to check your System Properties (right-click My Computer and choose Properties) and make sure Windows recognizes all 4 GB of your memory.

Even if I'm wrong about all that, I'd still check the modules in pairs. It will take a lot less time for each test iteration, and if you do run into an error, you've got only two modules to check individually to find the culprit rather than four.
 
Oh, another thing, and I'm much more confident about this one... :)

When running MemTest86, set it to run Test #5 for a while (at least 10-20 minutes) before you move on to other tests. Most bad RAM will barf on Test #5 pretty promptly, which might save you a lot of time. Of course, here's hoping you get no errors!
 
Thanks Crooked Paul, that all makes sense. I kinda thought I needed to test them in pairs just because of the message I got, I hadn't thought about the fact that if there was an issue I wouldn't know which set had it though. I came home from work to no power last night so I couldn't do much of nothing, it's back on this morning but It will have to wait until tonight.
 
Crooked Paul said:
I'm not 100% sure -- maybe someone who is better-informed can correct or elaborate on this. You might want to check your System Properties (right-click My Computer and choose Properties) and make sure Windows recognizes all 4 GB of your memory.
Click to expand...
correct http://www.hardforum.com/showthread.php?t=1035670


but I don't think that will matter if he's booting up to memtest since the OS is not running.
 
Crooked Paul said:
Oh, another thing, and I'm much more confident about this one... :)

When running MemTest86, set it to run Test #5 for a while (at least 10-20 minutes) before you move on to other tests. Most bad RAM will barf on Test #5 pretty promptly, which might save you a lot of time. Of course, here's hoping you get no errors!
Click to expand...

Agreed. I usually test with test #5 first for at least 30 passes then move onto test #8 for another 30 passes, if they both pass, I reboot, and let memtest run all tests overnight.
 
NordicRX8 said:
Agreed. I usually test with test #5 first for at least 30 passes then move onto test #8 for another 30 passes, if they both pass, I reboot, and let memtest run all tests overnight.
Click to expand...

Well, that sounds like a plan to me, I'll start that tonight on one pair and then the other pair the night after, thanks guys !
 
You must log in or register to reply here.
Back
Top