Framework Desktop reboot fail with PCIe gen3 Optane SSD

Hi,

I got my Framework Desktop just set up as a computing powerhorse. The Fedora Server 42 system was installed on a Intel Optane P1600X SSD (118G, PCIe Gen3) for its high durability.

However, I noticed that I have a ~50% chance of boot failure when I reboot the system. The bios shows “bood device not found”. At such circumstances, clicking the power button can successfully reboot the system and get me into Fedora. Also, startup from cold nearly always work.

Could this be due to a BIOS incompatibility with the Optane drive, or something else?

I’m planning to install the system drive on the PCIe slot with an adapter, and lock the slot speed to gen3. Will update the result when the adapter arrives.

Thanks!

OK. I found several PCIe-M.2 adapters. I was never able to boot when the system SSD is connected to the PCIe slot. The video output is just pure blackscreen.

I tried to use my 58 gig optane drive with the system but it never once recognized it once I booted into linux. I ended up just taking it out and putting it back in my old system, and pulling the 250gig drive out of another system to use for my swap drive.

Some updates: I reinstalled the Fedora42 OS onto a PCIe Gen4 P5801X SSD, and it works perfectly stable. So I assume it was a problem with the bios handshaking with PCIe Gen3 SSDs

1 Like

Interestingly, I’ve had similar issues when running Fedora 42 from an external USB-C SSD. A reboot (or two) usually helped. No issues since I installed it on my internal drive. Wasn’t on Framework Desktop though, still waiting for mine.

Updated to latest firmware (3.03), still doesn’t work.

Plugging P1600X on the PCIe slot (set to PCIe Gen3) doesn’t pass post, and at the M.2 slot was the same as before - reboot cannot find boot device.

Can confirm same behavior from my p1600x, if I may ask how did you install a p5801x? That is edssf form factor is it not?

Yes, my P5801X has E1.s form factor.

I used an E1.s to U.2 adapter from NFHK, an sff-8654 to U.2 cable from linkreal and a PCIe to 8654 card from Cablecc. Works fine at Gen4 speed.

I just saw this and posted this yesterday in the other PCIe thread. It took three different adapters, but I found one that worked with the right settings: What are you all doing with the PCIE slot? - #58 by Ernie_Hobbs

Thanks for the info! I have tried setting the slot speed to Gen3 with my PCIe to M.2 card, and got no luck.

I have also tried a funny PCIe-8654 card/8654-U.2 cable/M.2 to U.2 adapter combination, in which the power is drawn directly from the PSU, but still didn’t work.

Same here. I’ve tried several adapters and several NVMe drives, with no luck at either Gen3 or Gen4 speeds. The system simply hangs.

@DennisU When you say ‘hangs,’ does it hang at the Framework logo with the spinning dots, or before that (black screen)?

I spent two weeks troubleshooting a ‘hang’ that turned out to be software, not hardware. Because I moved a drive from another PC, Windows 11 saw the new PCIe adapter as a hardware change and got stuck in a loop trying to validate the boot path/BitLocker, but it looked exactly like a hardware freeze.

Two things fixed it for me: 1. Simplifying the Adapter: I switched from a fancy active adapter (Glotrends PA19) to a simple passive one (PA09-HS). The active chipsets seem to conflict with this board. 2. Wiping/Swapping: The ‘hang’ only stopped after I completely wiped the drive partitions (using Diskpart from a recovery USB) to remove the old boot/encryption signatures. Windows was trying to repair a path that didn’t exist.

Hi,

@Ernie_Hobbs thank you for your quick answer. With hang, I mean it stops even before the framework logo.

After reading through quite a few posts in this community, it seems to be common problem for people that upgraded their bios to 3.04.

for the adapter I tried the first one is a very basic one with no logic at all, only rerouted traces. the second one only features an extra voltage regulator.

both are working fine in other machines.

Are you on bios 3.04 or an earlier version?

Greetings

bios version 3.03

i would love to know if the 3.04 would render the slot unusable. but as far as i understand there is no way back to 3.03….

I had exactly same symptom with yours when I was using BIOS version 3.02. I tried using a PCIe-M.2 adapter (I forgot the brand) and a cable/M.2 to U.2 adapter solution with Optane P1600X, and both were not able to boot, showing total blackscreen without Framework logo.

I’m experiencing the same issue. My NVMe P1600X can boot normally, but if I restart the computer on either Windows or Linux, the P1600X simply won’t boot unless I shut the system down completely and power it back on.

I have done some further testing recently, and I found that P4801X 100 G can also boot/reboot smoothly with a simple GLOTRENDS PA22110 adapter and the slot speed locked to PCIe 3.0. Only tested Fedora 43 so far.

1 Like

I found that my P4801X (on the PCIe adapter) gets an entry in the SMART error log every reboot… It has been behaving well on other MBs without any error logs.

Setting the slot speed to Gen4 does not prevent normal rebooting, but also gives error log.

Error Information (NVMe Log 0x01, 16 of 64 entries)
Num ErrCount SQId CmdId Status PELoc LBA NSID VS Message
0 12 1 0x0099 0xc008 0x000 1755240 1 - Data Transfer Error
1 11 1 0x0099 0xc008 0x000 1755240 1 - Data Transfer Error
2 10 1 0x0099 0xc008 0x000 1755248 1 - Data Transfer Error
3 9 1 0x0099 0xc008 0x000 1755248 1 - Data Transfer Error
4 8 1 0x0098 0xc008 0x000 1755256 1 - Data Transfer Error
5 7 1 0x009a 0xc008 0x000 1755256 1 - Data Transfer Error
6 6 1 0x009a 0xc008 0x000 1755256 1 - Data Transfer Error
7 5 1 0x009a 0xc008 0x000 1755256 1 - Data Transfer Error
8 4 1 0x009a 0xc008 0x000 1755256 1 - Data Transfer Error
9 3 1 0x009a 0xc008 0x000 1755256 1 - Data Transfer Error
10 2 1 0x009a 0xc008 0x000 1755256 1 - Data Transfer Error
11 1 1 0x0096 0xc008 0x000 1755256 1 - Data Transfer Error

I don’t think it really damages the SSD, but it’s just annoying…