Memtest86

[Ashley]

it's for a UEFI boot PC running normally Windows, its owner does not want to install Ubuntu. "memtest86+" is completely free (unlike "memtest86"), but is quite old, and does not offer a UEFI boot - this is why having it within the Ubuntu options makes things easier)

Memtest86+ version 6 (and newer) can be downloaded as a standalone zip file, which contains an iso file. This iso file can be cloned directly into a USB drive to make a bootable memtest86+ system and it works both in UEFI mode and the old BIOS mode alias legacy mode.

There is another version, the 'original' but now not FOSS version memtest86, and it works in UEFI mode. There is a commercial version and a free version (no cost, but not open source code).

Use PCMemTest. Since Ubuntu 22.04 the PCMemTest is available from the Ubuntu repository. PCMemTest is a fork of MemTest86+ and it works fine on UEFI too. As it is part of Ubuntu it is the easiest to install:

Diagnosing RAM with tools most of the time does not work. I ran Memtest86 for 5/6 hours. It reported that the RAMs do not have any issues. Then I took out the RAMs i doubted and the system started running smoothly.

Just keep one RAM and remove others. Then run your machine for a while and check if there is any issue or not (remember, the issues can be very subtle, for example keyboard does not work for 1-5 seconds every once in a while). If there is no issue then it is a good RAM. Then check ram 2,3...n.

Please note that all RAMs and slots can be good but if there is a miss match (bus speed or brand or the ram) then it can cause issues too. So, you have to make sure that combinations also work. For example, use same brand and same bus speed RAMs together and test the machine for a while (remember, the issues can be very subtle). Then attach another brand / bus speed RAM and test for a while. Then maybe enable XMP and test for a while.

According to this discussion it can't run in UEFI mode because it's a 16-bit program.

If available, booting the live USB (or DVD) in Legacy/CSM (AKA "BIOS mode") should bring the memtest86+ option you want to the live menu.

If you're talking about memtest86/memtest86+, as in the bootable programs, sure. Interrupting it won't do anything, since it never writes any persistent data. In fact, the tests are intentionally endless - it'll just keep running passes until you stop it.

memtest is structured as a number of tests, each of a specific pattern. A single completed run through all selected tests is known as a pass. As mentioned before, you can safely abort at any time, simply by switching off the machine.

There is no optimal number of passes - an obvious failure will be caught in the first pass, while intermittent failures might take a hundred passes to appear (at which point you might as well get ECC RAM). I used to recommend running at least 10 passes to catch the more common intermittent failures, though with larger RAM capacities these days it could take too long.

You are spot on, I actually managed to interrupt the auto-start of the test and navigate to settings and manually change the resolution. Once changed to either 1920x1080 or the native 2256x1504, the test screen now displays it properly.

PassMark suggests that this is a firmware issue, and further recommends purchasing the configurable version of memtest86+ to remediate it. If it really is a firmware issue, it may be worth pursuing a fix on the Framework side.

I'm installing FreeDOS by floppy and CD now, and so far there has been no issues. It also boots FreeDOS from floppy disk without issues. This makes me believe that this issue with memtest86 is not related to an actual hardware problem but probably an old mainboard with some quirks and maybe a BIOS with some bugs.

There's a DOS version of Memtest86+, too. If you're able to boot into DOS, you could try running the exe from there.

It's no longer available for the latest version, but older versions are still online: (get the "Pre-Compiled EXE file")

I expect you guys know that 4.20 introduces "failsafe mode", but also there was at least one version of 4.20 that was buggy. The one that shipped with Ubuntu 12.10 LiveISO was false-positive happy. I don't know how it was limited to that release of 4.20, or if it actually extended further.

While ago, I had an customer with notebook that would boot and go BSOD, thinking either bad hard drive or software. I knew better and got memtest86+ on usb and booted with which promptly detected one bad memory module out of two and replaced it with exact match when special order came in. Fixed perfectly.

After reading through the change logs...

I don't think any past v2.11 add anything new that would help with pre-DDR2 vintage equipment's.

v4.00 and up seem to only add things for DDR2 equipment and newer so for retro I don't see any benefit in using 4.10 over 4.00.

.

GRUMPY OLD FART - On Hiatus, sort'a

Mann-Made Global Warming. - We should be more concerned about the Intellectual Climate.

You can teach a man to fish and feed him for life, but if he can't handle sushi you must also teach him to cook.

Detecting the type of memory depends on the chipset. Reading chipset configuration is a feature memtest got around Pentium III time, if I remember correctly, so around 2000. No one cared about implementing getting configuration from old 486 chipsets at that time, so the feature is just not implemented. One reason for this is that for many chipsets of that time, datasheets are publicly not available, although IIRC your LS board uses a SiS 496/497 chipset which has a publicly available datasheet.

The cache size is determined by doing timing measurement, I think. memtest doesn't implement any classic timing methods, but relies on the cycle counter (time stamp counter) that is only provided from Pentium onwards, so you won't get cache size and performance data on 486 class computers.

Pretty old topic but let's pull it up. Does anybody have a working floppy image that can test the ram on my FIC PIO-2 with a dx4 which they can provide? So far I did not find any working memtest. ctramtest complains that it cannot detect my host bridge and exits, memtest86 just reboots and memtest86+ in either 4.0 or 2.x are iso's only and too large to convert them to a floppy image.

Thanks a lot, that worked nicely! Simply copying the images to the Ventoy drive was enough. I tried both the propietary (memtest86) and the FOSS (memtest86+) version successfully. For the FOSS version, I downloaded the ISO from the project homepage.

Normally the CPU is not what determines supported memory speed. That is usually controlled by the mobo and bios firmware. My CPU is a ryzen 5 3600 on a B550M Pro mobo and the mobo supports the 3600 RAM.

The memory speed should match the infinity fabric speed, anything else will have significant performance impact. Since memory is DDR (double data rate) you want to divide the memory speed by 2 to match the infinity fabric. So for 3200 your infinity fabric speed needs to be able to reach 1600.

Underclocking memory is generally fine and you could spend hours cranking down the timings but as you pointed out, the slower speed memory is cheaper, so it just depends on where you want to spend time/$$$.

Most of the better motherboards have a bios that is able to set memory and CPU timing factors as well as adjusting voltages for both memory and CPU. Adjusting the memory settings is usually in the same area as adjusting values for the CPU.

What other issues could cause it?

Could it be the fact that the old ram is 1333 and the new ram is 1600?

Could it be a hair or dust on one of the pins?

Should I return it and get all new RAM?

We need to get a bit specific here. The memtest86 program has different tests that it does, each testing different types of memory failures. Once it loops through all the tests it does a second pass of all the tests, then a third, etc. The failure will be in a specific test, and knowing which one would help us narrow down potential causes.

Please notice that there are two versions of memtest, one FOSS that works only in BIOS mode and one free as in free beer, that works also in UEFI mode. This link may help explaining the details.

If I understand and remember correctly from , the bleeding edge version is memtest86+ v5.31b, which is a beta version. Last September I downloaded and tested it, and it worked in BIOS mode but did not work in UEFI mode.

You probably called the kernel "memtest86.bin". If the file extension is ".bin", PXELinux will try to load the file as a MBR instead of a kernel file. Rename the file and change the configuration, everything will work fine then.

I have a dual-boot system with WIndows 10 and Arch. Both are using UEFI boot with an ESP partition. I used efibootmgr according to the EFISTUB wiki article to add Arch as a boot option. This works perfectly fine, with the output from efibootmgr being:

3a8082e126