AMD RAM issue

I’ve got a poll going on in the AMD Batch 1 thread for RAM issues in case you guys want community data on it. [RESPONDED] AMD Batch 1 Guild - #844 by TheTRUEAsian

I really appreciate you taking the time out to address this issue, and thank you for being proactive about the situation.

If you’d like me to hold onto the RAM modules (so I could test a patch) or send them to you, please let me know. My return deadline is Thursday (10/19).

That’s news to me, but thank you for sharing! Would you happen to know which 5200s have been reported working? I tried searching a moment ago and came up short.

On a happier note…

My FW booted up in less than two minutes with the Kingston modules (KF556S40IBK2-64), and I am happy to report I am seeing 5600 in the Task Manager!

One thing I noticed that was interesting this time around was that there was an additional blip in the LED sequence. Approximately 10 seconds after the last of the blips in the main sequence, there was one red LED. After about 10 more seconds, the FW logo came on screen.

4 Likes

Tuesday is the 17th. Thursday is the 19th.

1 Like

Thank you for spotting my mistake!

Sadly I’m now in the camp of does not boot (7640U). My LED sequence is 12 green and then GGBBBBGG for the POST code, display backlight comes on for very short burst and occasionally the fan does too. My RAM is the KSM56T46BS8KM-16HA (5600MT, CL46). Not directly on the KB ram article, but KSM48T40BS8KM-16HM is, which is the 4800MT CL40 model. Only hardware difference according to Kingstons Memory part number is A-die for mine versus M-die for the ones in the knowledge base. Number of chips, PMIC, SPD Hub, Thermal sensor, everything is the same. I tried disconnecting the main battery for a few minutes to prevent an earlier DRAM training during testing interfering, but no dice :confused:

Edit1: After some tinkering the POST code changed to GRBGGBBBBB. Curiously 2 more blinks this time…

Edit2: Thanks to @Jonathan_Haas identified the sequence to actually be

indicating a RAM problem.

That would equate to “everything ok, entering Sysprep mode” (which I believe is a mode the the manufacturer flashes the base system image). This looks mostly ok, maybe it’s not detecting the SSD or the system installed on it.

That sounds weird, after the orange color there should be only 8 blinks. Are you sure you didn’t mix up red and orange? When it really was GGGGGGGGGGRG O BGGBBBBB, that would indeed indicate a RAM failure. But maybe you just made it worse by “tinkering” with it?

You are probably correct. I stepped through a video of the sequence and the second reddish one seems to be orange:


But it is very close, maybe another color would be nice :slight_smile: Tinkering means switching the modules around, trying each one in different channels etc. Sysprep POST was actually my first try, will try that again, only one module in channel 0. My SSD is empty, but I did not get the framework cog on display or any other output.

I have now seen GGGGGGGGGGGG O BGGBBBBB and GGGGGGGGGGGG O GGBBBBGG while trying all possible combinations of slots and my two sticks of RAM. Display stays black but with active backlight. Roundabout a minute between pressing the power button and POST code blinking.

I believe both of these POST codes are fine and indicate no RAM problem, however both indicate that there is no system to boot from. I’d look somewhere else, either the system is booting (or trying to boot) but the display is unconnected or off, or it’s simply missing a boot device and it is supposed to not display anything at this point. My AMD laptop is unfortunately still in the pipeline, so I have nothing to compare. I would try plugging in a known good bootable USB stick (like a windows or Linux installer medium) and maybe an external screen and see if something shows up then.

Also be sure to be patient and wait for the memory calibration.

You should probably also message support.

2 Likes

Thanks, I have contacted support. Tried installing the working system NVMe SSD out of my current notebook as well as connecting a known good bootable USB (UEFI bootloader). No changes at all. No difference to no bootable medium installed.

So even the RAM bought directly from framework has issues?

My RAM is not bought from Framework.

1 Like

I think you are correct that memory is not my issue. When I plug an external monitor AND unplug the eDP port on the mainboard I can enter the system setup, there I see the 5600MT and 32GB capacity correctly displayed. Thanks for giving me the idea. It does not work with my DP via USB-C monitor, but the HDMI card in the lower right bay works…

1 Like

@halemmerich, are you seeing any red LED(s) about 10 seconds after the primary sequence finishes?

It only blipped once for me upon successfully booting for the first time, but I’m curious if you are seeing that with your (potential) display issue. I might try my 5200s again just to see if that red LED was firing before and I just never noticed it.

No, I did not see any additional blinks outside of red blinking for chassis intrusion if I take of the input cover.

Thanks for the feedback. Maybe you accidentally damaged or disconnected the screen connector while messing with the RAM? Had the screen worked before?

No, it never did anything beyond the backlight turning on/off. Since the display is detected correctly including resolution I suspect the connection is physically ok. Backlight brightness can be controlled by hotkeys. I do not have another Framework to switch parts around, so can’t really debug this further, but the notebook behaves as if the display was working fine. If the BIOS had been mirrored on both screens I would never tried so much getting the RAM to work since it worked most of the time. Just 2 times it actually did not boot because of the RAM. It never occured to me that my ECC RAM could not be the problem. Additionally 3 boot devices that work on my XPS 13 did not boot on the FW (current Arch install iso, Refind based bootloader I use daily for booting my XPS and a working system drive) which caused the boot to stop on the no OS error that I could not see.

I’ll have to wait for support to get to my ticket.

1 Like

Good news, support sent a new display and everything is working fine now. I never had a RAM issue :sweat_smile:.

11 Likes

Are you still using the server memory with ECC support?

Yes, but without parity. Some more discussion on the ECC stuff can be found in https://community.frame.work/t/will-the-new-amd-boards-support-ecc-ram/28373.