Aside from Memtest86+ v 2.x are there any other reliable testing programs that i can run to stress test my ram and cpu at the same time ?
I've passed the memtest86+ test # over 100 times.
Printable View
Aside from Memtest86+ v 2.x are there any other reliable testing programs that i can run to stress test my ram and cpu at the same time ?
I've passed the memtest86+ test # over 100 times.
Have you ran any of the actually burn in utilities and benchmarks yet like Prime95 and Orthos. Those will stress the whole system to check for stability in your OC. Memtest is only there to test the memory for errors.
Yes, they cant fill up more than 4gb since they are 32bit applications.. infact they are limited to 1.82gb under windows taskmanager and only show 40% usage on my end.
EDIT:
Would this work;
Prime95 v25.6 -2 threads ( CPU affinity 0 & 1 ) Blend test
OCCT 2.0.0a - ( CPU affinity 2& 3 ) RAM test
OR
Dual Orthos ( CPU affinity 0 & 1 ) blend
Dual Orthos ( CPU affinity 2 & 3 ) blend
that should cover about 4gb usage.. since im using 6gb there would be a problem in running out of CPU pairs
Yes that is what you have to do in order to load up a multi core CPU and all the RAM with those applications.
Hmm, i think the best balance would be a dual orthos with 1 CPU affinity per program and 1,447.5Gb ram per orthos..
But im open to more recommendations.
HCI MemTest plus Prime95 is what I use too.
HCI MemTest shows errors really quickly, you should definitely give it a try! :up:
Some people don't like it because it runs in Windows, which in theory adds a bit more into the stability equation.
But considering most of us run our PCs in a Windows environment, i'd say it's a fair way to stress a system ;)
Actually, for stressing RAM & the northbridge, it's by far the best way IMO.
I've had my system run Memtest86+ all night long error-free only to have it error in HCI within 15 minutes.
Memtest86+ is crippled by the fact it's 32-bit & obviously as such, cannot stress all RAM simultaneously on systems with a lot of RAM (think 4+ GB).
With HCI, you can run as many instances as needed to stress all available RAM (two instances is good for 4 GB; four for 8 GB), & it puts a beating on the NB as well.
While i do still use Memtest86+ for initial basic testing, i much prefer to use HCI to actually verify stability, as quite frankly, neither Memtest86+, P95, OCCT, or pretty much all other stresstesting apps out there can truely stress 4+ GB.
I'd say that other than LinPack, there's nothing that will test RAM quite as hard as HCI.
http://hcidesign.com/memtest/
Of course Memtest86+ can test more than 8 GB.
My point was that it cannot test all the RAM simultaneously.
It does it in chunks.
So while it can check massive amounts of RAM, any instabilities that are due to the stress the NB takes with all the RAM being utilized, or the added stress of all 8 GB being utilized at once vs. the chunks Memtest does, don't tend to show up unfortunately.
Hence why i rely more on HCI now that i have 8 GB.
HCI also tests the memory in chunks, a lot smaller if i may add
LOL.
I know it's small amounts that HCI is limited to.
That's why i said multiple instances!
I did this a while back to test my system. I do blend of OCCT for overall stability, but to narrow down to ram stability I do memtest.
I like this picture very much! Serious business... :D
Any benefit from doing 512's instead of multiple 1536's ?
No, for some reason it kept telling me it couldn't assign more than 512 per instance and so that is exactly what I gave it.
Weird.
On my system, it usually lets me assign up to around 1450 MB each, though it sometimes is more; sometimes less.
mine allowed a max of 1197 for a single instance on winxp pro sp2
Im starting to see the weakness of this program for machines that have 6Gb or higher, The concept of its memory testing is to stress test the said memory space as fast as possible and as stressing as possible the faster the coverage goes over every 100% the "stressful" it gets. seeing as we have a quadcore we are then limited to 4 instances to have an optimal testing speed, the program is limited to 1400Mb in memory testing size so having 4 instances tests 5.6Gb only w/c is reasonable since the 400mb is usually used by the OS, for 8GB machines having quadcores this would be a problem as you are forced to use 5-6 instances but no dedicated CPU for those extra instance forcing the program to steal cycles from the other instances and thus creating a slower coverage rounds, but then again there is no other effective tool aside from running 4 instances of dual orthos..
Take a look at Linpack if you have an Intel CPU.
http://www.xtremesystems.org/forums/...d.php?t=175729