How reliable are the results from hcidesign's Memtest?

[Mark 2007]

Hi all,
My system was running stable for a month, but I decided to try running a memory testing tool to see what the results were. Googling "Memtest" brought up two applications, Memtest86+ and MemTest for windows, from http://hcidesign.com/memtest/

When running the MemTest from hcidesign, I was surprised to see it pop up with an error once or twice during each pass. I lowered my overclocking settings and tried again, and it still reported problems with the ram. I tried several other lower clock settings and I had the same problem.

Then I tried Memtest86+ for DOS and Microsoft's own "Windows Memory Diagnostic". I did that for about 2 or 3 passes, and it reported no problems. I then ram Prime95's torture test, and 6 hours later it reported no problems.

Is MemTest from hcidesign considered a reliable testing tool? Anyone use it regularly? Maybe it's testing method goes much deeper than what a normal app or game would ever use?

Mark

[Methious]

Run the memtest from dos, but run it like 8 hours it's pretty accurate. Inside windows other programs/processes can interfere and generate errors. If your clean in DOS and prime95 runs stable I'd say it's just a program conflicting with the windows version.

[geokilla]

I run Memtest+ Test #5 for 1 to 2 hours, then I deem my RAM stable. Test #5 is the most stressing test according to John from AnandTech.

[elkmanwsu]

I'm pretty sure Memtest+ is more widely used...but I suppose people have their preference for everything.

I have previous experience with memtest+ when I was diagnosing some bad samsung pc-2100 ddr ram and it was able to figure out which stick was going bad.

[Mark 2007]

Thanks for the replies!

I'd be interested in seeing the results from other folk running HCI's Memtest from http://hcidesign.com/memtest/

Anyone want to try? It doesn't take too long to run one pass.

[Athlonite]

well i d/l'd it try it and i got no errors when testing all available ram and i still had programs open and running i also did a few benchies with everest at the same time just see if it would cause an error to occure but nothing happend and memtest just kept on tickin along

[mr_noname111]

I actually use HCI Design as my primary memtest tool, although I do have MEmtest86+ 1.70 burnt onto a disc.

This is because I had issues with memory compatibility with my ram and my motherboard, and had to try out different BIOS revisions to get it stable (insane amounts of corruption). Even when I got tons of corruption, Memtest86+ didnt give me any errors where as HCI gave me literally hundreds.

When I finally landed on F11 rev, HCI gave me no memory errors, and lo and behold, the corruption problems ceased. So I deem HCI a very reliable tool.

[brajko2]

I think that HCI Design's memtest is not that much useful application. I've run it several times, and I was unable to find any error in memory. With memtest I did find errors while I was running it.
The best way to test if my RAM was o.k. (either at default frequency or when overclocking) is Orthos and OCCT. Orthos is especially good, and when I know that I've overclocked my pc too much, Orthos will find error usually 10 seconds after I start it. Soit really a great tool for me.

[Mark 2007]

Thanks everyone for your posts. It seems I've solved my Memtest problems! And now hcidesign's Memtest and Memtest86+ are running flawlessly on my ram.

What was my solution? Well, I was having problems with Source games (Half-Life 2, Counter-Strike Source etc) crashing after about 10 minutes of playing. Textures would disappear before they quit to desktop. After googling, I came across forum posts about changing "PagedPoolSize" in the registry. So I did, and now my games work without any problems. And then I remembered my problems with hcidesign's Memtest so I tried that too. No errors!

So, in retrospect: nothing wrong with hcidesign's Memtest. It's probably more useful for doing "real world" testing, since it checks the RAM while Windows is up and running. I imagine there's a lot of people out there getting the same problem as me though, and I've read that it's more common with people who are running 1gb of ram. If you google that error, you can find the forum posts with step by step instructions for editing the registry.

Thanks
Mark