Processors

[fesg]

Hello.
I really don't know where to locate my problem...
I have random crashes (in any situation, just on desktop or just with browser).
It's my first AMD after 20 years with Intel without a failure, anyway...

With this situation, the system would restart and wait for it to happen again.
While using the browser, suddenly I notice in the task manager that instead of "seeing" the 32GB that it has installed, it only sees about 18GB (that is, more than the memory of a module but less than the total), after a while it reboot only.
I have two Windows to rule out issues on different SSDs but the problem persists no matter the Win.

Equipment is Ryzen 9 7950x, ASUS Crooshair X670E Hero, 2x16 GB Gskill 6000Mhz, CL 36 (listed as valid by ASUS), Corsair RM 1000i font, MSI Suprim X 3800Ti graphics. The equipment is cooled by custom water and EKW water block. Win 11 new installation, clean. I tried another install of Win 11 on a different SSD. The same. SSD is Samsung 990PRO 1 TB, and WD Black SN770 1 TB, Win each one, and same problem. Idem with SSD SATA.

My problem is random crashing, but I add things:

When rebooting and entering the Bios, of the 2x16GB of RAM, it "only sees" 18432 GB (that is, more than a single module and not the total...). The way to go back to being "okay" is to go to the memory profiles and set ren to "automatic" (EXPO profiles do not appear, the bios does not see them). It reboots, and when I go back to the Bios the correct amount of memory 32768 DDR5 already appears. The EXPO I and II profiles are already reappearing. I choose any one, restart, and go back to Windows. The problem reappears randomly, hours without passing, and others does not last even 5 minutes, simply leaving the windows as it has started without running any program. In Task Manager, before it crashed and crashed, I had time to capture the screenshots. IN THIS CASE, EXPO PROFILES WERE NOT USED, it was left on Automatic and Windows was started, but as can be seen, it "loses memory" again. But with the profiles, I have tried the I and the II, it also fails; although without the profiles, even if it fails by leaking memory, at least it doesn't hang or reboot.

Somehow it is as if the bios suddenly and with Windows running, loses part of the memory and this leads to a crash or having to restart it.

I have tried another source Corsair AX 860i, with the same result.

THE PROBLEM DOESN'T HAPPEN JUST PLAYING, IT HAPPENS JUST BEING AT THE DESKTOP.

Today I will receive another new set of memories to try, although I am afraid from what I have read that THE PROBLEM IS IN THE INTERPRETATION OF THE EXPO PROFILES BY THE MOTHERBOARD BIOS

Best of all, seeing the issue of memory, "I began to follow it", and I have seen that with Windows working (and with the Expo profiles removed, everything automatic, and Resize BAR also removed), suddenly it happens to me from 32 GB to about 19 GB (NOTE: more GB than those of a single module.....). But Win is still working.

As I mentioned, in case of having EXPO activated (I or II does not matter), it went from 32 GB to about 8GB, everything became slow, and after a few seconds it restarted. It is when I restart when I see in the Bios that it only "sees" about 18 GB.

I think that little by little it will be fixed. My computer is working and, as you will understand, a working tool restarting randomly, bad business. And more after having spent a small fortune to be a PC, trusting AMD (sorry but it is so...)

PLEASE HELP...

Thanks in advance.

[fesg]

Hello again.

SOLVED : don't buy GSkill for AM5 and 7950x

Posting the same issue on Reddit, it turns out it's not an isolated occurrence; even if it were, AMD should take good note as it is a new platform like AM5.
It turns out that among the users with a similar problem, one of them has changed 4 MSI motherboards and he had the exact same thing happened to me, AND HE HAS FOUND THE PROBLEM AND THE SOLUTION (late for me, I'll explain it later).

I had only tried changing to new GSkill modules (recognized by GSkill and ASUS as compatible...). This other user had the same thing. Until he tried another brand, CORSAIR DOMINATOR, and the problem was fixed, everything stable.

My computer is working. I couldn't be like this. It randomly shuts down and you lose all work. So I have taken the step of going to INTEL, core i) 13900k and Kingston memory.
Expensive but radical solution. It cannot be that involved in teams of this level not obtain a clear answer. The memory manufacturer tells me that either the kit is wrong and I return it (it was already on the second change) or that some AM5 pin is bent (checked and everything is perfect). And ASUS kind of doesn't know...
Manufacturers already have to know very well that this is happening. Surely a BIOS update will fix it; but you can't be at the expense with a team of so much money waiting for something that you don't know where or when...

Pity. For once after 25 years with Intel I want to give AMD a chance and this situation... VERY EXPENSIVE situation to be in with no response whatsoever.

[MADZyren]

I have 7950X and do not experience this issues with Asus B650E-E with 2x32Gb Kingston 

According to information from you, problem is with G.Skill memory kits, so why blame AMD?

Also I don't experience instability except when trying to push voltage curve too low.

AM5 is a new platform and some things will likely be improved over time, but in my case I still use an early BIOS  - never flashed myself - and have not encountered issues at stock/XMP/DOCP settings.

I think people should stop installing Ryzen Master, MSI Afterburner and other such programs, like Asus crate and simply use the computer as is. Just throw in drivers and let Windows update itself. If you want to test things, use BIOS and not some program.

[fesg]

In fact, as you well indicate, your computer has nothing to do with the one that I mention that I have the problem and therefore I don't understand why something similar has to happen to it. I'm glad everything is going well for you. We are talking about the X670e chipset, Gskill Memories, and in any case with the latest EXPO profiles that are not compatible with XMP as you claim to have. It's a totally different scenario.
It has been tested with various "clean" installations of windows with nothing else. Even removing the main graphics card and leaving only the one integrated into the microprocessor. Leaving a single memory module. Moving it around. With different Nvme SSD in different ports, and with SATA SSD. Continue to fail ....

[MADZyren]

It is unfortunate that you have problems with your new computer and I understand that it is frustrating. 

My point was that:

a) G.Skill memory kits with EXPO timings for some reason have been noticed  to have issues, which is unexpected as G.Skill is a good brand, but maybe even the best make mistakes sometimes (The fact that Seasonic these days sells some PSUs which they didn't manufacture themselves gives me the chills)

b) I have the same CPU with same mainboard manufacturer and no issues, so it's not like this memory limitation is due to CPU- or memory controller design.

I hope you get your system running well soon. I would imagine the memory issue with G.Skill could be fixed either by loosening subtimings or lowering memory clocks - or changing memory voltage. The issue you have does not sound like typical memory instability though, so make sure you have seated your memory sticks properly (LevelOneTechs -channel mentioned on their YouTube channel, how important this is with DDR5) and that you don't have any bent CPU pins. I have to say memory capacity changing like that is not something I have encountered over the decades.

[TheHunter]

For me, I was having constant windows freezes in random hours... tested many different things, SSD, HDD, different CPU... and 4 different Corsair sticks, and 4 different Kingston sticks. But so far the problem was gone with the third test with another memory kit, with adata / xpg...

The other kits does not have errors in stress and memtests, but was freezing windows in random hours...

So, even the memory is in QVL, or even the tests have no problem, the problem could be the memory... stranger things...

[fesg]

To all of the above I would like to add that the memories passed OK all the tests including the one of Windows itself. All ok. But running the operating system, the system fails. It is possible that at the moment we are not many. It is an "expensive" combination so be safe. But surely it will end up being a trending thropic when the topic spreads. At the moment the manufacturers do not give an answer and surely they know the problem and they know that the solution is on the way. But let's set the scene: It comes to light that a certain high-priced product and very, very, very good competition from Intel gives problems of this type. Sales would drop to the bottom. It's better to have the beta tester users, keep selling and hiding the problem, and wait for the fix.
I can't afford the above. Having a set of more than €6,000 as a work tool that is useless.
At the moment happy owner of Intel Core i9 13900k. PERFECT
And I'm not against AMD... Can anyone think that I am against AMD, spending such an amount of money on AMD????

[fesg]

Surely there is a serious underlying problem here.

AM5 is totally new and has startup errors. Worse when so much money has been spent this is not a consolation. I already changed everything to Intel. I imagine that successive Bios updates (and everything in general) will solve the problem.

But currently it is not a valid platform as a work platform. You can't be working and suddenly the computer shuts down and you lose everything.

[wim-bart]

First check the partnumbers of the memory and if they are tested by the motherboard manufacturer for that specific motherboard. If it is not on that list as tested successfully in the configuration which is used then fix that first. If the manufacturer did test 2 modules and someone is using 4 then also problems arise.

The issue I read here is that either memory is not compatible with the motherboard, might be subtimings not correct, or the motherboard has an issue (a motherboard is a very sensitive component and ESD or scratches can hurt). And also the CPU can have an issue. Without replacing either memory, CPU or Motherboard nothing can be ruled out. Even the memory n wrong slots is a possibility.

But when memory is swapped between slots and the failure does not change slot position, but keeps appearing on same DIMM slot, the only two possibilities are: Motherboard or CPU. When Swapping CPU and issue remains the Motherboard is the issue ( and visa versa ).

[fesg]

As I already indicated, the two modules (2x16GB Skills) are authorized by ASUS, Gskill for this motherboard and should be fully compatible. They are EXPO. Tested with and without EXPO, all in maximum motherboard compatibility mode with no tweaks, clear BIOS, and all fully updated.
Gskill answers me to change the modules, I already changed them for new ones and everything is the same; or that some pin of the CPU socket could be bent (all verified by me and by ASUS, which sent me to return the board and they have behaved at their level, WONDERFUL, thanks ASUS).
There is a latent problem, some combination at some point, something, that produces this error: random loss of part of the memory, and shutdown of the computer; I needed to reboot twice to get it back and start again. And without touching anything else. After a while the problem returns. Sometimes every "nothing" (2 or 3 minutes) and others not half an hour. Sometimes leaving the PC without doing anything at all, just as the desktop remains as the PC starts up.

[Jskill]

I have the same problem with asus prime x670-7800x3d (7600x previous cpu) after i updated bios to the latest, in worst case i had 4gb shown and used and i have to shut down and reseat my ram. Oh, kingston fury 2x16gb amd expo verified 5600mhz. seems it's no single time happening either since iv'e had 4-18-24gb randomly on start. need better bios now...