• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

MemTest64 crashes even though other memory tests don’t

Joined
Mar 5, 2023
Messages
158 (0.19/day)
I’m using a G.Skill Flare X5 6000 CL36 with EXPO enabled. Build: https://pcpartpicker.com/list/fpfDLs

I ran MemTest64 with Task Manager and HWiNFO. Both TestMem5 and Task Manager crashed. I also got an error message later saying "MemTest64.exe - System Error: Exception Processing Message 0xc0000005 - Unexpected parameters." I had run it for 8 hours at this point, and there weren't actually any memory errors reported in MemTest64 even with this message. There was one time my entire screen went black for some reason. I'm not sure if there were any memory errors here since I couldn't see anything. Here's a pic of that:

I'm also getting error popups when I try to open other programs, but I guess that's expected if I have 100% CPU usage and almost 100% RAM usage.

I troubleshooted by running other tests.The other tests I ran include 24 hours of HCI MemTest, 4 passes of MemTest86 (I don't have the paid version), 3 cycles each of TestMem5 Absolut, Extreme, and 1usmus configurations (I'll run more cycles though), 1 hour of OCCT memory stress tests, AIDA64 memory benchmarks, and 24 hours of Y-Cruncher stress tests which cycle between all of the available tests. I tried to do 30 GB in Linpack Xtreme, but that made my SSD usage jump up to 100%; from what I understand from running HCI MemTest, that's not a very accurate memory test then. How should I run that then? I'll try to install Linux to use GSAT, and maybe I'll pay for Karhu. Additionally, are there any other memory stress tests or benchmarks I should be running?

I've only run into issues with MemTest64; is something just wrong with the program itself then, or is my RAM unstable?
 
This been discuss the number of times. Some people think that no matter the program if it crashes your memories unstable.

Personally I only trust three programs and that's it

Memtest5 is the best free one.
 
Last edited:
This been discuss the number of times. Some people think that no matter the program if it crashes your memories unstable.

Personally I only trust reprograms and that's it

Memtest5 is the best free one.
What do you mean by reprograms? I have tried MemTest5 Extreme, Absolut, and 1usmus configs for 3 cycles each, and they all worked without any errors. I’ll try even more cycles though.

additionalyl, I forgot to mention that I can’t disable Power Down Mode. Maybe it’s different for DDR5 RAM, but I thought it was generally a good idea to disable that. If I disable that, my PC doesn’t boot up. Is that a RAM, mobo, or BIOS issue? I also can’t enable memory context restore and EXPO at the same time, though that’s understandable.
 
Last edited:
Is your system stable otherwise? If yes, no worries.
 
Is your system stable otherwise? If yes, no worries.
It’s stable as long as I don’t disable Power Down mode or enable Memory Context Restore and EXPO at the same time. The latter case is understandable, but I don’t see why Power Down mode isn’t working. For DDR4 RAM, I thought disabling it is recommended, and it worked fine for my AM4 build; maybe it’s different for DDR5 RAM. Does that mean my mobo, RAM, or any other parts are faulty? Or is it likely just a BIOS issue?
 
run y cruncher VST and testmem5 anta777 extreme. (several hours)
 
run y cruncher VST and testmem5 anta777 extreme. (several hours)
What’s Y Cruncher VST? I ran the stress test option with all tests enabled for 24 hours. I also only ran TM5 Anta777 extreme but only for 3 cycles. I’ll try it again for several hours.
 
Y-Cruncher is a stress testing utility.
VST is a mode in Y Cruncher.
 
What do you mean by reprograms?
Auto correct error on my phone. 3 programs is what I meant to say.

Remember this is AMD first attempt at DDR5. Not surprised by things like GDM not working. I haven't done much overclocking besides seeing what the max freq is.
 
Y-Cruncher is a stress testing utility.
VST is a mode in Y Cruncher.
Oh okay. I’ll look for VST mode then. Thanks!

Do you have any issue without EXPO? Make sure you CPU is 100% stable by its own.
I don’t have issues when I enable Memory Context Restore without EXPO. I haven’t tried Power Down mode without EXPO yet, so I’ll try that out and see if that works. I’ll try these with one stick too; my previous RAM module worked with MCT for one stick but didn’t work for the other which is why I thought it was defective and returned it. It seems like it’s not a memory issue since it’s happening with my second one though I’ll try that again.


Auto correct error on my phone. 3 programs is what I meant to say.

Remember this is AMD first attempt at DDR5. Not surprised by things like GDM not working. I haven't done much overclocking besides seeing what the max freq is.
That makes sense. So, stuff like Power Down mode and Memory Clntext Restore are expected to not work. I did hear that my ASUS TUF GAMING B650-Plus WiFi has bad BIOS versions too which seems true. I wish the official reviews I read mentioned that! Still, if that’s all normal, I don’t really care. It persisted for both of the memory modules I tried, so it’s either a mobo or BIOS issue at this point. What are the three programs that I should rely on?
 
Last edited:
Same here, as today, on my MSI x670e Carbon, "Memory Context Restore" has to be "AUTO". Neither "Enable" or "Disabled" works. (BSOD)
It seems that setting is just buggy then. Have you tried disabling Power Down mode yet?
 
Yes, in the DRAM menu, I always set "Power Down Mode = Disabled". I never had issue regarding this.
I see. I don’t know why that isn’t working for me then. Do you think that’s a problem with my motherboard and memory or with my BIOS since AM5 ASUS has bad BIOS apparently. Memory Context Restore didn’t work with any BIOS versions, but I’ll try disabling Power Down Mode with every BIOS version to see if any of them work. I think the latest Bios Version wasn’t even working with EXPO alone, so I guess it won’t be that one.
 
That makes sense. So, stuff like Power Down mode and Memory Clntext Restore are expected to not work. I did hear that my ASUS TUF GAMING B650-Plus WiFi has bad BIOS versions too which seems true. I wish the official reviews I read mentioned that! Still, if that’s all normal, I don’t really care. It persisted for both of the memory modules I tried, so it’s either a mobo or BIOS issue at this point.
Memory Context Restore is a add-on after people complained. GDM is part of the design. GDM Off is like Command Rate 1, On is like 1.5. 1T is picky on Intel DDR5 with only a few MBs being able to run it. On AMD DDR4, if you turned off GDM depending on the motherboard either it would default to 1T or 2T. So it's possible your is defaulting to 1T when it should be 2T instead for DDR5.
 
I see. I don’t know why that isn’t working for me then. Do you think that’s a problem with my motherboard and memory or with my BIOS since AM5 ASUS has bad BIOS apparently. Memory Context Restore didn’t work with any BIOS versions, but I’ll try disabling Power Down Mode with every BIOS version to see if any of them work. I think the latest Bios Version wasn’t even working with EXPO alone, so I guess it won’t be that one.
It is hard to say what goes wrong. I personally got a bad ddr5 kit when I built my am5 rig. Expo could boot but not stable. It took me months to figure out it was the bad ram. However, from MSI forum, I saw someone with similar problems but only replacing the cpu fixed the issue. Recently, the YouTuber JayzTwoCents encountered bad motherboard on his am5 build. After he replaced the motherboard, he still could not clock ram over 5200MT/s. You may check out his channel. Thus, nothing is guaranteed.
 
It is hard to say what goes wrong. I personally got a bad ddr5 kit when I built my am5 rig. Expo could boot but not stable. It took me months to figure out it was the bad ram. However, from MSI forum, I saw someone with similar problems but only replacing the cpu fixed the issue. Recently, the YouTuber JayzTwoCents encountered bad motherboard on his am5 build. After he replaced the motherboard, he still could not clock ram over 5200MT/s. You may check out his channel. Thus, nothing is guaranteed.
I realized yesterday that setting PBO to Enabled alongside disabling Power Down mode makes my PC fail to boot up. I read that Auto and Enabled are basically the same thing for PBO, but it seems they're not exactly the same thing; still, sicne theyre the same and my CPU is indeed boosting to 5.13GHz, I guess I'll just go with Auto for now. I'll have to test these two alongside memory context restore. I've tested EXPO on its own for about a month, and I haven't noticed any issues with that. It seems that EXPO on its own is at least stable. I thought my old memory kit was bad since Memory Context Restore worked on one stick but didn't work on the other; I guess it's not a memory issue though since it's not working on my new memory kit either. That makes sense though; it could be any number of reasons why it's not working.

Memory Context Restore is a add-on after people complained. GDM is part of the design. GDM Off is like Command Rate 1, On is like 1.5. 1T is picky on Intel DDR5 with only a few MBs being able to run it. On AMD DDR4, if you turned off GDM depending on the motherboard either it would default to 1T or 2T. So it's possible your is defaulting to 1T when it should be 2T instead for DDR5.
I see, so Memory Context Restore probably isn't the most stable setting. As for GDM, is that different from Power Down mode or is that the same thing? I'm not able to find GDM in my BIOS for some reason although HWiNFO64 does say that my command rate is 1T. If I can find where to change that, I'll try to change that to 2T. Does changing the command rate from 1T to 2T hinder performance at all?
 
Context restore keeps the previously trained values for the next boot. I haven't played with it that much, but so far I noticed higher frequency memory and the more memory you have, the less stable it is to use that function. Advanced memory training is part of the design. Why fight it?

As for GDM, I see it in software, but not in later BIOS either. 1T and 2T impact is really only seen in benchmark stuff. Nothing that affects normal uses.
 
Context restore keeps the previously trained values for the next boot. I haven't played with it that much, but so far I noticed higher frequency memory and the more memory you have, the less stable it is to use that function. Advanced memory training is part of the design. Why fight it?

As for GDM, I see it in software, but not in later BIOS either. 1T and 2T impact is really only seen in benchmark stuff. Nothing that affects normal uses.
That makes sense. I can see why I’m having issues then with high frequency 6000MHz RAM. I was just worried that something was wrong with my parts, but it’s good to hear that this is normal. I’ll just keep that disabled then.

Is there another way to change that to 2T then? If it’s more stable, I’ll definitely go with 2T.
 
Is there another way to change that to 2T then? If it’s more stable, I’ll definitely go with 2T.
I'm looking at the ASUS X670E Hero BIOS and don't see it at all. My guess it's already 2T. But idk for certain without poking around more.

If you pass y-cruncher 2.5b and memtest5, I wouldn't even worry about it. As long as the apps you actually don't crash your good.
 
I'm looking at the ASUS X670E Hero BIOS and don't see it at all. My guess it's already 2T. But idk for certain without poking around more.

If you pass y-cruncher 2.5b and memtest5, I wouldn't even worry about it. As long as the apps you actually don't crash your good.
I wonder why HWINFO says it’s 1T. Maybe changing that to 2T will make these other BIOS settings work though I don’t see an option for that. Regardless, I am passing both of those tests, so I’m glad that all seems good then. None of the apps I actually use have crashed yet either. Thanks!
 
It could very well be 1T for AMD and DDR5. The platform is only a 4 months old a this point. Intel to AMD doesn't translate always. Just a obversion I seen with Intel and 1T for DDR5.

Memtest5 and Karhu as the two programs I trust the most and tend to be the most "stressful" in my opinion. Y-cruncher is just icing on top. The prime95 of memory if you will.
 
It could very well be 1T for AMD and DDR5. The platform is only a 4 months old a this point. Intel to AMD doesn't translate always. Just a obversion I seen with Intel and 1T for DDR5.

Memtest5 and Karhu as the two programs I trust the most and tend to be the most "stressful" in my opinion. Y-cruncher is just icing on top. The prime95 of memory if you will.
That makes sense. I've been looking into it a bit more, and I've also heard that 1T is less stable. I'd like to change since it doesn't make a difference, but it doesn't seem that there's an option to do that. I would run Karhu, but I don't want to pay for it lol. As for MemTest5, I ran the ABSOLUT test for 20 hours which is probably excessive, but there at least weren't any errors. I will try to run the Extreme test for about 20 hours as well.
 
I do a lot of memory testing. I've been doing it for at lease 10+ years & this is what I know.

Each memory tester out there including different version of the same program maybe different. ie different test pattern are added or taken away. So when one version may past all test, another version may fail.

All memory testers require a little bit of system memory to work. So let't look at this carefully. Depending on where the program has allocate a little bit of system memory to run the program, you must remember the program itself can be corrupted (normally not permanent) due to the high level of errors in the memory module. ...ie what i'm saying here is, the program itself may not have loaded correctly due to memory. fault(s). Testing within windows is a bad idea because windows itself may not have loaded correctly. Besides this windows OS is occupying some of system memory which can't be tested.

The best way to test your memory is to use an external standalone bootable device that loads the program & uses little as possible system memory. This is about the program not loading or hitting a bad part of your system memory.

Memory faults are sometimes difficult to detect, but if you have the right professional hardware, "all faults" are capture. .& when I say "all faults", I mean every single bit of fault, there is no escape or hiding place. Understanding memory faults is very difficult at times. It's not always easy to understand what's going on.. It took me many years to understand what's going on & how test programs work, which test pattern triggered certain type of fault(s) within the memory module.

At the end of the day, if you system is unstable & you suspect it's the memory module, you need to know what type of fault you are dealing with. Unfortunately most end user(s) will not know the answer to this.
 
Last edited:
Back
Top