Memtest, Memtest86 & Memtest86+

If you've never had problems with your hardware you must have a contract with God. Share your pain and knowledge with other members.
Post Reply
User avatar
Book
Sargeant at Arms
Posts: 249
Joined: Sat Dec 09, 2006 4:16 am

Memtest, Memtest86 & Memtest86+

Post by Book » Tue Sep 22, 2009 1:07 am

It appears that there are a lot of people that are wondering which "memtest" to use for testing their memory, and when talking about "memtest" different people may refer to different programs (I've seen that happening on at least one forum). So, I though I'd post this, since I couldn't register on the forum in question, due to a recaptcha problem.

As of now (22 Sept 2009), there are 3 "memtests" that I know of

a) Memtest
b) Memtest86
c) Memtest86+

a) [http://hcidesign.com/memtest/]
This is a program that runs from within windows and tests memory. You can get the free version for free obviously. It's a nice tool to have, but I don't know how accurate it is, even though there's been somebody on a forum who said that this program found errors in his memory when memtest86+ failed to (this actually seems to happen to one of my PCs as well, but I haven't tested it extensively). Also when people refer to "memtest" they sometimes do not refer to this utility, but rather "memtest86" or "memtest86+".

b) [http://www.memtest86.com/]
This is the original memtest86, and the program that most people refer to, for testing memory. You can boot off of it (either from a CDrom or floppy), and it runs a bunch of tests to evaluate your memory. There has been a period when this program was not updated, for 2 years or so, and memtest86+ was developed (by another team) to compensate for that. This does not hold true right now, though, since the last version as of today is v3.5 (released at 3 Jan 2009).

c) [http://www.memtest.org/]
This is the program that was developed when memtest86 was not being actively updated. However, both are now being updated, and have their own version numbers which can be confusing. As far as I know, these programs run the same tests (as of today, 22 Sept 2009), and therefore it doesn't really matter which one you choose (aside from possible bugs and/or compatibility issues). Also keep in mind that the memory tests themselves shouldn't change too much with every version, what changes is hardware compatibility and bugs. Latest version of memtest86+ is 2.11, released at 22/12/2008.

Here's a quote from wikipedia:
Memtest86 was developed by Chris Brady. After Memtest86 remained at v3.0 for two years, Memtest86+ was created by Samuel Demeulemeester to add support for newer CPUs and chipsets. Both are now (as of 2009) being actively worked on. The bootloading code was originally derived from Linux 1.2.1. Memtest86 is written in C and x86 assembly. The source code is released under the GNU General Public License (GPL). The current version of Memtest86 is v3.5, released on January 3, 2009 and v2.11 of Memtest86+, released on December 22, 2008. Both versions now support current dual- and quad-core-CPU's and the corresponding chipsets. The latest release of Memtest86+ supports Intel-based Macintosh computers.
Also, if you're using either memtest86 or memtest86+ it would be a good idea to leave them running for at least 15 hours, because in my exprerience I have seen the first error after the 30th pass of the standard set of tests (1-8), which translates to around 12 hours (depending on hardware configuration). I know, though, that this is an extreme exception (and probably occured due to excessive heat, as I was running on a laptop). Generally (as mentioned in the readme), the first pass should detect all but the most obscure errors.

Good luck!

User avatar
Book
Sargeant at Arms
Posts: 249
Joined: Sat Dec 09, 2006 4:16 am

Re: Memtest, Memtest86 & Memtest86+

Post by Book » Thu Nov 11, 2010 7:38 am

I know this is a year old thread, and that I'm only replying to myself, but I felt like expressing a relevant experience I had with Memtest86+

I got a hold of two DDR memory sticks of undisclosed origin the other day, so I set them up, booted up my handy UBCD disk, and launched Memtest86+ v4.10 (the current version as of this post). After letting it run its standard tests (Tests 1-8, which is the standard test pack) for a little over 24 hours, no errors were detected. Since I had the option, I also let it run the "Bit fade test" (Test #9, takes 3 hours for one pass) for another 24 hours, yet still no errors were detected. Just to make sure, I let Memtest86 (version 3.5 which is the current as of this post) to run a pass of its standard tests as well. It didn't detect anything either, but that was to be expected since they are pretty similar to each other.

Confident that the sticks are in good working condition, I took one of them and used it on my workstation PC. Within one day strange little errors started popping around with no evident cause, and then a sudden restart (no BSOD). So, logically acting, I launched my third memory diagnostic option (option #1 in the above post), and surprisingly enough errors were detected within seconds! I really wasn't expecting that, so I restarted everything and began the test again, only to get same results. I took off my "new" stick and the errors were gone (along with the random windows errors). (Using the other stick hasn't generated any problems, and it so far passes the stress test, so I shouldn't be having any problems with that)

Now, Memtest86+ which is supposed to be a comprehensive, bootable tool for memory testing couldn't detect faulty memory in 48 HOURS, that a not-so-known windows application detects within seconds? I am still rather surprised by this, but that's what's happened to me. Bottom line is, don't rely on Memtest86(+) alone on your next memory troubleshooting

Not to mention that Memtest86 (v3.5) has a bug which will make most systems with 4GB of memory restart within the first few seconds. And it's not even mentioned on their home page! I had to refer to a wikipedia talk page (now it's also in the main article) to find out that other users are having the same problem! That issue (which has caused me trouble in the past) doesn't exist in Memtest86+ though.

As a last note I should mention that I'm not affiliated with any company, just sharing an experience that I think may be helpful for other users.

User avatar
Tech_Junkie
Your Senior
Posts: 871
Joined: Wed Apr 14, 2004 8:07 pm

Re: Memtest, Memtest86 & Memtest86+

Post by Tech_Junkie » Fri Nov 12, 2010 1:01 pm

Some food for thought.
"Confident that the sticks are in good working condition, I took one of them and used it on my workstation PC. Within one day strange little errors started popping around with no evident cause, and then a sudden restart (no BSOD)"

You could have a bad or incompatible memory controller on the PC that kicked out the errors. You may also be able to change the memory timings on that machine manually in the BIOS and use it without issues.

Cooling may also be an issue.

Did you check the memory again in the machine that gave no errors?

You may be able to load the same operating system and tools on the 1st machine and run error free because of the difference in supporting hardware or timings.

Each type of memory is meant to run at certain timings and voltages. If you are running 2 different types/brands in the same machine they may not work well together.

It most likely isn't a problem with memtest.

User avatar
Book
Sargeant at Arms
Posts: 249
Joined: Sat Dec 09, 2006 4:16 am

Re: Memtest, Memtest86 & Memtest86+

Post by Book » Tue Nov 16, 2010 4:12 pm

Which program are you referring to? :grin:

I did some re-testing and I have even some more interesting points to make:

In the machine that gave the errors, two DDR memory sticks were used of different capacity (and as I later realised different CAS latency, 3 & 2.5) but of the same manufacturer and of the same frequency. This has worked for me before, the manufacturer is approved and they were the same sticks that were used in the error-free machine (tested with memtest86+ for 48 hours). Memtest86+ showed the CAS latency as 2.5 so I can only assume this reason for the malfunction, since it's beyond the one stick's specifications. Memtest86+ was indeed able to detect the malfunction too, but see below

On the error free machine, Memtest86+ did not show any numbers for the latencies. I checked the BIOS, and it reported (in a not so clear menu though), that the CAS latency was 2.5 for this configuration too! This overruled the last assumption, but maybe this reading was not really correct, because placing just one DIMM with CAS latency 3 would not change it! Anyway...

1) In the first case I would have assumed that the latency would be automatically set to the highest of the individual DIMMs that were in place, but it wasn't. I searched all the BIOS settings, but there was no such option to configure manually (apparently the MB is on the lowest end of the "enthusiast" scale).

2) I don't know why the BIOS in the other machine reports 2.5 CAS latency but still manages to work everything out perfectly (or even if that is true). I haven't tried using the simple memtest yet (no 86+-), but so far all tests have been successful (I runed another I think 18-19 passes)

3) In the machine that gave errors, Memtest86+ (in one instance) completed 2 passes, without giving any errors (in previous instanced errors were detected in Test #5), but there were in fact errors just like before which were detected with memtest (not so fast though, either). Maybe they would have been detected after a few more passes but that just shows you something right there. (Either let the program run longer or even use other programs as well)

User avatar
SLaX
Apprentice
Posts: 44
Joined: Fri Aug 17, 2007 2:13 pm
Location: Somewhere
Contact:

Re: Memtest, Memtest86 & Memtest86+

Post by SLaX » Mon Apr 18, 2011 10:22 pm

Cool, didn't know there were different versions. Would be nice if the program would tell you which stick is acting up though.

User avatar
Cool_Fire
Not a sandwich
Posts: 1912
Joined: Fri May 09, 2003 1:20 pm
Location: 41 6d 73 74 65 72 64 61 6d
Contact:

Re: Memtest, Memtest86 & Memtest86+

Post by Cool_Fire » Tue Apr 19, 2011 2:33 am

If you really want to, you can infer this from the memory address any errors show up in.
If we're breaking the rules, then how come you can't catch us? You can't find us? I know why. Cause, it's ... MAGIC!
Hackerthreads chat, where the party is going 24/7.

twister
n00b
Posts: 8
Joined: Fri Jul 15, 2011 12:45 pm

Re: Memtest, Memtest86 & Memtest86+

Post by twister » Fri Jul 15, 2011 5:06 pm

Memtest86 and Memtest86+ are open source memory test software programs designed to test and stress test an x86 architecture computer's random access memory (RAM) for errors. Each tries to verify that the RAM will accept and correctly retain arbitrary patterns of data written to it, and that there are no errors where different bits of memory interact, and there are no conflicts between memory addresses.
:grin: :grin: :grin:
"A compliment is something like a kiss through a veil"
(Victor Hugo)

MJCRO
n00b
Posts: 2
Joined: Fri Dec 23, 2011 3:59 pm
Location: N.Y.

Re: Memtest, Memtest86 & Memtest86+

Post by MJCRO » Fri Dec 30, 2011 9:14 pm

I love using MemTest86+. Great little program to put on a CD and run at POST to test memory. :grin:
Intel i7 2600K @ 5.1Ghz / HT On / 1.45v
16GB G.Skill Sniper 1866Mhz CAS9 RAM @ 2133Mhz
Gigabyte Z68X-UD7-B3 Rev. 1.0
EVGA GTX 580 Classified 3GB @ Stock [1Ghz+]
PC Power & Cooling Silencer MK.II 950W

Post Reply