memtest for windows

[Kingfisher]

OK,

I have two instances set -up to run this app, do you need to set affinity like you do in prime 95? Both programs have checkmarks in 0 and 1 by default.

Praz can you give a noob some help?

[Praz]

You don't have to do anything. Double click on the exe and put in the amount of memory to test. Double click on the same exe a second time and put in the same amount of memory.

[Kingfisher]

Alright, thank you.

[Praz]

I wish I was at the computer at the house. I have a little something I wrote up for this but haven't posted it yet. I'll get it done.

[Guest Kobalt]

That memtest isn't as good as the self booting memtest. (from memtest86.com) It can't test all the memory, since windows is using it.

[Praz]

That's why it needs to be run for 2000%-3000% coverage minimum. Through memory swapping practically all memory will be tested.

 

As a rule of thumb, Win9X uses about 32MB of RAM when nothing is open, and Windows2000/XP uses about 64MB. Even though you cannot directly check this RAM, Windows dynamically moves the location of most of its subsystems, so most of your RAM will be checked eventually.

For verifying ram at it's stated primary timings and speed there probably isn't anything better then Memtest86. But as most know what's Memtest86 stable isn't necessarily Windows stable. Memtest For Windows is excellent for verifying secondary timing compatibility with Windows. It's just one more tool to help us reach our overclocking goals.

[soundx98]

I wish I was at the computer at the house. I have a little something I wrote up for this but haven't posted it yet. I'll get it done.

 

While I use memtest all the time I wonder if I really am making maximum use of the program. Little tutorial action "Praz style" would be OK with me.

 

That memtest isn't as good as the self booting memtest. (from memtest86.com) It can't test all the memory, since windows is using it.

 

But the corollary to that is it IS testing the memory in windows INSTEAD of DOS.

I'm sure there a benefits and drawbacks in both programs.

 

Let's see what Praz got up his sleeve over the next couple days.

[Guest Kobalt]

But the corollary to that is it IS testing the memory in windows INSTEAD of DOS.

I'm sure there a benefits and drawbacks in both programs.

Memtest don't use DOS. Heck, even MS's own memory test program don't use DOS. It is a custom loader.

[Praz]

I'm sure that was used as a generic term to mean not running under an operating system.

[soundx98]

I'm sure that was used as a generic term to mean not running under an operating system.

 

My apologies for any possible disinformation to all.

And I certainly hope it didn't detract from the gist of my post.

 

Questions on the Memtest programs, how to correctly use them, which tests to use, what the tests do, and the length of time they should be run are quite common on the forum.

It is quite common for users with issues to be asked if they have run Memtest without much detail on the procedure.

 

I had suggested to Praz in a PM that a tutorial/discussion thread on the Memtest programs could be helpful to many and encouraged him to look into it.

Sorry for any thread derailment or confusion I may have caused.

[hawkeyefan]

That memtest isn't as good as the self booting memtest. (from memtest86.com) It can't test all the memory, since windows is using it.

 

Not true. HCI Memtest and SnM Memory Test are each better tests than memtest86+ for DOS, as they place the additional strain of OS overhead and background processes on the memory.

 

memtest86+ is a good for running quick tests of #5 and #8 to check for any major instability, and should be run on new sticks for 12-24 hours for reasons you point out. But if you are tinkering with known good sticks, HCI and SnM are going to show errors faster than memtest86+. They will also error out at settings that appear perfectly stable according to memtest86+. Several different times I've had memtest86+ pass for 12+ hours, only to have HCI or SnM pick up errors in 15 minutes.

 

So there is definitely a huge tradeoff for relying on only memtest86+. It proves that you don't have any bad individual ICs, but it doesn't prove that the sticks are stable in Windows or 3D apps.

[jaz]

Not true. HCI Memtest and SnM Memory Test are each better tests than memtest86+ for DOS, as they place the additional strain of OS overhead and background processes on the memory.

 

memtest86+ is a good for running quick tests of #5 and #8 to check for any major instability, and should be run on new sticks for 12-24 hours for reasons you point out. But if you are tinkering with known good sticks, HCI and SnM are going to show errors faster than memtest86+. They will also error out at settings that appear perfectly stable according to memtest86+. Several different times I've had memtest86+ pass for 12+ hours, only to have HCI or SnM pick up errors in 15 minutes.

 

So there is definitely a huge tradeoff for relying on only memtest86+. It proves that you don't have any bad individual ICs, but it doesn't prove that the sticks are stable in Windows or 3D apps.

 

Could you please post links to these programs? I can't configure memtest, although i finally passed over 8 hours at stock settings with a64info settings manually entered into bios and broken in slowly with demos and benchies.

Thanks in advance.