Memtest freezing with 4 sticks but each set tests fine on their own

ryzen-5600x-build-freezes-windows-10-af ... d-2530779/
^
In this thread over here, everyone is suspecting Team Group RAM could be the problem someone else is having with their build. This is my 3rd read already about RAM issues and its TG DDR4 ram. Your thread, the one above and on another board. Any chance you can get return/refund the Team Group RAM and get 64GB kit of another brand? Corsair or G.Skill?

Turn down your settings in the BIOS until you get a stable system. Maybe start with standard 3200 settings and see if it passes Memtest with all sticks. If it does, then increase timings, etc. slowly until it locks up again.

What is your objective with the tests? Are you still within the return period and want to return it if it can't run all four @ 3600?

Or are you past the return period, but just want to fully max everything out for overclocking? Do you get lockups with all four under normal use and trying to narrow down the problem?

Don't have another system with 4 ram slots to test? Could rule out the ram if you had another motherboard to try it on.

If the ram tests fine individually it could be some bizarre motherboard incompatability, or possibly a faulty slot.

I tend to steer away from the TeamGroup, Adata/XPG and other lower cost ram brands that dont manufacture their own chips for this reason, just not worth the hassle and having that suspicion hanging over head if weird things happen IMHO.

I would rma this if possible. If it isnt then I would increase the memory timing or very slightly increase the voltage to the ram.

Just because you have two sets of the same memory kit doesn't mean they'll always play nice together. Also, combined kits aren't often tested by motherboard manufacturers as part of their compatibility lists.

i got an asus x570 dark hero and 2 32 gig kits of corsair ram they worked fine together for several months then started crashing. memtest would reboot when running it and prime95 would come up with rounding errors. when running just 2 sticks though it stopped. corsair exchange it for a kit of 4 matched 16 gig sticks and it still didn't fix the issue so they exchanged it again for 2 32 gig sticks and everything is working as it should they are also going to send me one of there light kits with 2 dummy sticks to fill the other slots at no charge.

my only guess is that some bios update along the way broke compatibility with 4 sticks because i setup everything back in April 2021 and everything was smooth till just a few weeks ago.

heyyahblah wrote: ↑ryzen-5600x-build-freezes-windows-10-af ... d-2530779/
^
In this thread over here, everyone is suspecting Team Group RAM could be the problem someone else is having with their build. This is my 3rd read already about RAM issues and its TG DDR4 ram. Your thread, the one above and on another board. Any chance you can get return/refund the Team Group RAM and get 64GB kit of another brand? Corsair or G.Skill?

I might be able to return them, they are from Newegg and they seem to be all friendly now about returns, because of that scandal I guess. What sucks though is that I used a $100 off promo that I wouldn't be able to use again.

jm1 wrote: ↑Turn down your settings in the BIOS until you get a stable system. Maybe start with standard 3200 settings and see if it passes Memtest with all sticks. If it does, then increase timings, etc. slowly until it locks up again.

What is your objective with the tests? Are you still within the return period and want to return it if it can't run all four @ 3600?

Or are you past the return period, but just want to fully max everything out for overclocking? Do you get lockups with all four under normal use and trying to narrow down the problem?

I am not overclocking, just turned on XMP that's it. I am doing initial testing of a new PC, just on the bench so to speak, not yet assembled. I haven't actually tested all 4 in normal use such as running Windows. I did run 4 other sticks from 2 different kits, all 4 memtest'd fine and ran Windows OK but all I did was run some temperature tests.

M1K3Z0R wrote: ↑Don't have another system with 4 ram slots to test? Could rule out the ram if you had another motherboard to try it on.

If the ram tests fine individually it could be some bizarre motherboard incompatability, or possibly a faulty slot.

I tend to steer away from the TeamGroup, Adata/XPG and other lower cost ram brands that dont manufacture their own chips for this reason, just not worth the hassle and having that suspicion hanging over head if weird things happen IMHO.

Sorry I completely forgot to check if my other boards can handle 4x32GB, I will check and if so test. But the slots are OK because I memtest'd with 4 other sticks from 2 different kits before this. I understand TeamGroup is lower end, which of course also makes it even more attractive when buying 128GB for the prices. I have lots of TeamGroup RAM working fine in other PCs, but I have seen some boards that just don't like certain sticks, from multiple brands as well.

Devhux wrote: ↑Just because you have two sets of the same memory kit doesn't mean they'll always play nice together. Also, combined kits aren't often tested by motherboard manufacturers as part of their compatibility lists.

Yeah, as a last resort I could ask for an exchange.



I forgot to mention that all 4 slots were previously tested together with different RAM so the slots are OK. The RAM was slower though. The motherboard was purchased used so I am stuck with it.

Interested to see how this plays out OP. I'm betting this is down to motherboard shenanigans (either with this specific ram, perhaps an odd UEFI memory bug being a newer 12th gen mobo), the RAM is probably fine.

I glossed over the fact you are running 128GB. Though your motherboard supports that capacity, I'm thinking it's just not playing nice with this brand and the mixed kits. Try installing one stick from each kit so you have a half and half mix just 64GB capacity and run the test, if it crashes there then solid chance it's down to ram revisions or something not getting on with your mobo.

Perhaps try turning multithreading off before running the test to see if it still hangs, I have heard of some server boards having such issues with high thread count CPUs and memtest.

Does it crash with XMP off? I had similar issues with TG RAM. Would crash with the default XMP settings, but worked fine without. Returned it the next day.

You do realize XMP is overclocking the ram? Running 4 sticks is highly stressing the memory controller. You can run without xmp or you can bump up the voltage. Team Group isn't exactly top tier brand. I doubt they're getting the best chips. But just as likely your motherboard isn't providing the stable voltages needed to run that spec.

M1K3Z0R wrote: ↑Interested to see how this plays out OP. I'm betting this is down to motherboard shenanigans (either with this specific ram, perhaps an odd UEFI memory bug being a newer 12th gen mobo), the RAM is probably fine.

I glossed over the fact you are running 128GB. Though your motherboard supports that capacity, I'm thinking it's just not playing nice with this brand and the mixed kits. Try installing one stick from each kit so you have a half and half mix just 64GB capacity and run the test, if it crashes there then solid chance it's down to ram revisions or something not getting on with your mobo.

Perhaps try turning multithreading off before running the test to see if it still hangs, I have heard of some server boards having such issues with high thread count CPUs and memtest.

kramer1 wrote: ↑Does it crash with XMP off? I had similar issues with TG RAM. Would crash with the default XMP settings, but worked fine without. Returned it the next day.

shutterbug wrote: ↑ You do realize XMP is overclocking the ram? Running 4 sticks is highly stressing the memory controller. You can run without xmp or you can bump up the voltage. Team Group isn't exactly top tier brand. I doubt they're getting the best chips. But just as likely your motherboard isn't providing the stable voltages needed to run that spec.

Yeah I know what you mean but when people say "overclocking" they usually mean something more involved than just running RAM at the manufacturer XMP settings, that's what I meant sorry if that didn't make sense. You're correct that I should test it without XMP, I forgot to do so. That brings up another question, will running the RAM at base 2400Mhz hurt the CPU (i9-12900K) performance a lot?



I tested the RAM on another board I happen to have that supports 4x32GB, an AMD system with a B550. I could not get it to boot with XMP on so I settled for memtest'ing without it, hasn't frozen yet and it's well into the second pass. On the Z690 it would freeze very quickly, 20 seconds or so, to maybe 60 seconds max. So yeah it's looking like "shenanigans". I MIGHT be able to get the motherboard replaced by asking the person I purchased it from, but I don't have high hopes of that solving it anyway, it would just be a what the heck let's try it kinda thing.

People keep referring to the fact that it's TeamGroup, but is that more a concern when you get into larger amounts like 128GB? I have lots of TeamGroup RAM in other PCs or ones that are gone now, it doesn't seem especially bad. That B550 I'm using to test has 4x16GB TeamGroup RAM, I have a 2x8GB 3733Mhz and a 2x8GB 4000Mhz that are working OK in other PCs.

Another option here is to just ignore the fact that is freezes with 4 sticks as a potential memtest quirk, and just run Windows and do a bunch of tests and see what happens. Does that sound reasonable?

Oh and if someone brings up the usual question, why am I building an i9 and not buying really expensive RAM and such to go with it, I got the CPU+board used, just happened upon a good deal, true RFD style , it wasn't a planned build. I was going to re-use RAM to save money but then I thought what the heck I'll get 128GB.

If its working with XMP OFF then that means it is usable. However I personally want what I paid for. If it claims 3200Mhz, then it better work at that, else it is going back.

Only reason I mentioned TG that it is they are the only brand I have ever had such an issue. Must mean they are using bottom of the barrel silicon.

Most other RAMs will even OC past XMP, but TG, from my limited exp, can't even manage the bare minimum.

Don’t ignore the hanging as a quirk. It will probably manifest as random errors during use.

Also 128GB is not an exceptional amount of RAM. Many servers have TBs of RAM.

From my experience it is just not worth bothering with bad hardware. Sure frugality is imp but so is the peace of mind that your device will work as intended when powered on. Also an error in one device can cause seeming unrelated errors in an another.

I have a network card that will cause the CPU to throw errors.

I can't remember if I used Memtest86 or Memtest86+, but forcing it to use a single CPU core fixed the freezing.

https://www.memtest86.com/tech_freezing-lockups.html
https://www.reddit.com/r/buildapc/comme ... ont_panic/

rabbit wrote: ↑ I can't remember if I used Memtest86 or Memtest86+, but forcing it to use a single CPU core fixed the freezing.

https://www.memtest86.com/tech_freezing-lockups.html
https://www.reddit.com/r/buildapc/comme ... ont_panic/

How is that a solution? Obviously a single core isn't going to push as much bandwidth (stress) across. That's not a fix. That's avoiding the problem.

It would be a solution if it were a software bug that we're dealing with.

Dude already stated that 2x32GB runs fine. My issue was when I ran it with 4x8GB. 2x8GB was fine.

Looks like XMP is the issue, I did some other testing, swapping RAM from the other PC, but ultimately the 4x32GB RAM tests OK with XMP off.

If I were to just accept that and run the RAM at 2400Mhz, how much of a performance hit am I taking?

evilYoda wrote: ↑ I would rma this if possible. If it isnt then I would increase the memory timing or very slightly increase the voltage to the ram.

Is this something that I need to research, like are there other settings that need to also be adjusted alongside? Or do I just go into the BIOS and try increasing those values a bit?

This RAM was an especially good deal with the promo that was on, the next best deal for a better brand would cost around 50% more. If anyone has more opinions to share I'd appreciate it, regardless thanks to everyone.

hystavito wrote: ↑ Looks like XMP is the issue, I did some other testing, swapping RAM from the other PC, but ultimately the 4x32GB RAM tests OK with XMP off.

If I were to just accept that and run the RAM at 2400Mhz, how much of a performance hit am I taking?

Is this something that I need to research, like are there other settings that need to also be adjusted alongside? Or do I just go into the BIOS and try increasing those values a bit?

This RAM was an especially good deal with the promo that was on, the next best deal for a better brand would cost around 50% more. If anyone has more opinions to share I'd appreciate it, regardless thanks to everyone.

u could try stepping the speed up manually on the ram and find out were it stops working stable then just roll the speed back 1 step.

if u can run 2 32 gigs sticks of RAM stable at 3600 though why not stick with that 64 gigs of ram is already pretty good what are u doing that needs 128 gb of ram?

Just bump up the voltage. I'm sure 0.01 increase will take care of it. XMP is just factory pre-tested overclocking. Just like CPU they only make one kind of ram. It's tested at
various speeds to see which it's capable of. Team Group probably goes for the minimum they can get away without any margin. Your motherboard is probably also v-droopping
under load. So the ram isn't getting the voltage it needs to run at that spec.

shutterbug wrote: ↑ Just bump up the voltage. I'm sure 0.01 increase will take care of it. XMP is just factory pre-tested overclocking. Just like CPU they only make one kind of ram. It's tested at
various speeds to see which it's capable of. Team Group probably goes for the minimum they can get away without any margin. Your motherboard is probably also v-droopping
under load. So the ram isn't getting the voltage it needs to run at that spec.

I have no experience messing with voltages so I just turned on XMP, looked at the settings in the BIOS, one was called "DRAM Voltage" and the other "Eventual DRAM Voltage", both were set to Auto and had the value chosen displayed as 1.356V. I tried to add exactly 0.01 but when setting manually it would only take 2 past the decimal, so I went with 1.36V.

Memtest threw a bunch of errors very quickly and froze. When I rebooted, even though I had set the values manually the displayed values next to them were higher, as if the board took my 1.36V and added to it or something? So it said 1.372V, and I tried inputting 1.38V. Memtest didn't give errors but very quickly froze.

There's also a value under Advanced that says DRAM Boost (for training), it has various set values to choose from like +100mV, -100mV etc. I didn't touch it.

Sorry I have very little experience with this stuff. Should I disable XMP and set speed, timing, everything manually? Thanks