FYI, Intel C2000 family of processors: System Fault may lead to dead system.

[Mazian]

After seeing the EEVblog video above, I discovered that a user in another forum found the location for the resistor fix on an A1SA7 board: pins 1 (LCLK) and 9 (+3.3V) on the JTPM1 header. The same header is in place for the A1SAi-2550F, A1SAi-2750F, A1SRi-2358F, A1SRi-2558F, and A1SRi-2758F boards, possibly more. Page 44 of the manual has the location and pinout.

I've got a 2750F board that has no problems yet, but I made and installed a 100 ohm jumper and... it still boots. Would be interesting to test it out on a dead board!

Show : Resistor jumper

100 ohm resistor, 5-pin header, and a pair of connectors to make a simple jumper. The other side had some extra plastic for a clip arrangement, like a CPU/case fan connector; I trimmed it off so it wouldn't push against the adjacent pins on the board.

Show : Position on 2750F motherboard

 

[alexkey]

Any solution for Asrock C2750D4I? Thanks!

 

[PDM]

Just a follow up with some good news here: I put in an RMA request (March 10) for my A1SAM-2550F board that died due to this issue after 4.5 years of full-time use (citing C2000 MB in the request). Super Micro responded within a day (March 11), explained their RMA process, I chose cross-shipping where I pay a deposit ($272 USD), they send me a new board, and I send mine back, once verified the deposit is refunded. I signed and they sent it out the next day (March 12) and it arrived the following day (March 13). The new board was a A1SAM-2550F rev 2.00, and worked perfectly on first boot. I then shipped back the bad board on Monday (March 16)... but by that time a lot of places were on lockdown. It was finally delivered on March 30th, processed April 2nd, and refunded via Paypal on April 7th. Overall it was a pretty good experience. I had a new board within 2 days, and it only cost me ~$30 in shipping and exchange fees, after floating the cost for just under four weeks.

Was A1SAM-2550F rev 2.00 clearly marked on the MB?

 

[BriggTrim]

Hi...the LPC is a PCI-to-ISA connect regulator and is one of the two upheld BIOS boot areas; the C2000 can either boot from SPI (default) or LPC/ISA (set by means of outer sense pins at powerup). This is fixed in a venturing, and inquisitively the "fix" comprises of disposing of the capacity of muxing the LPC transport pins with GPIO - they as of now not become programming selectable. This is essentially ALL I've had the option to discover regarding the matter. There's a workaround which comprises of adding an outer 100 ohm resistor, yet it's not satisfactory what sticks this is added to. It's additional across two cushions on a connector on some Synology NAS units, so it's anything but a yield current limiter however in all likelihood a solid pullup or pulldown. This leads me to presume it truly goes on an arrangement sense pin. Intel hasn't made their "foundation level change" public. Following it out on a board is somewhat hard since the SoC is an enormous BGA bundle that would should be desoldered.

Does anybody find out about this? Like, for instance, where the resistor is added - specifically is it added to the LPC clock yields, or to the sense inputs?

 

[Ericloewe]

There's some additional information in this other thread:

How to fix Asrock c2750d4i with C2000 bug

--- Mod note --- Please keep this thread for the specifics of the C2750D4I board, as used in the FreeNAS Mini. For general C2000 LPC clock degradation discussion, use this other thread instead. ------- Hi all, I just wanted to report back my findings after my Asrock c2750d4i board froze one...

www.truenas.com

 

[Rob235]

Some feedback from my experience:

I have a five year old A1SRi-2758F Supermicro board that has been randomly crashing for months for no clear reason. Then a few days ago it became hourly and then wouldn't boot at all. I've installed the 120ohm resistor on the JTPM header as mentioned above and it's been stable for 24hours. Thanks for the advice above and fingers crossed!

 

[vec4f]

I'm very happy I stumbled upon this thread, as my experience was similar to others (and thankfully, so too was the solution).

I built a small NAS back in 2014 using the SuperMicro A1SAi-2750F-O Mini ITX board. A little over a week ago it suddenly died (nearly 8 years of continuous use!). I could still access the IPMI interface but the system wouldn't boot. I thought it was gone forever until I went looking for a replacement, and stumbled upon some information about the Atom bug.

Thank you Mazian for giving me the information I needed to try the fix! In my case, I used a Dupont wiring/crimping kit to attach a 100ohm resistor directly to two single-wire Dupont connectors (no soldering required), which when inserted on the right pins brought my system back to life.

This system has served me well for almost 8 years and I'm glad to be able to extend its life a bit further :)

After seeing the EEVblog video above, I discovered that a user in another forum found the location for the resistor fix on an A1SA7 board: pins 1 (LCLK) and 9 (+3.3V) on the JTPM1 header. The same header is in place for the A1SAi-2550F, A1SAi-2750F, A1SRi-2358F, A1SRi-2558F, and A1SRi-2758F boards, possibly more. Page 44 of the manual has the location and pinout.

I've got a 2750F board that has no problems yet, but I made and installed a 100 ohm jumper and... it still boots. Would be interesting to test it out on a dead board!

Show : Resistor jumper

100 ohm resistor, 5-pin header, and a pair of connectors to make a simple jumper. The other side had some extra plastic for a clip arrangement, like a CPU/case fan connector; I trimmed it off so it wouldn't push against the adjacent pins on the board.

Show : Position on 2750F motherboard

 

[alugowski]

Thank you to the folks in this thread for both describing the issue and sharing your experiences in fixing it, both hacky and official.

My 7-ish year old Supermicro A1SRI-2758F started showing signs of C-2000 bug recently, and since folks had reported RMA success I thought I'd try it too. I'm happy to report that Supermicro replaced my board with no hassle, in July 2022.

I'm summarizing how my Supermicro RMA experience went for other folks. I was very impressed.

• Submit RMA form, specifying "C-2000 bug MB" in reason field.
• Next day received email that RMA is approved with instructions on where to send. I replied to ask if cross-ship is possible, which was approved a few hours later. Cross-ship required signing a cross-ship agreement (pay a deposit, send back board within a reasonable time frame). After signing, they generated a PayPal deposit charge for $475 + sales tax. They ship out the replacement FedEx Overnight on the same day that the deposit is paid.
• I used their cross-shipped packing material and box to mail back the old board with USPS.
• Received PayPal refund of the deposit about a week after they received the old board.

I chose cross-ship for three reasons. 1) Much less downtime, obviously. 2) I don't keep mobo packing materials on hand. 3) One SATA port was flaky and the ethernet ports started to be as well, maybe or maybe not due to C-2000 bug. A new board would fix those issues too.

The new board doesn't have visual differences that I can identify, but it's probably not just old stock with a resistor. Apart from an extra "-P" appended to the model number in some documentation, I identified two differences:

• The new board has a unique IPMI password, the old board used ADMIN/ADMIN. Supermicro introduced unique passwords around 2019 for new boards.
• The new board's IPMI has a Remote Control option for iKVM/HTML5, which works without Java! The old board had up to date firmware but required a Java applet or Java Web Start, neither of which works on any of my client machines. I'm excited to no longer have to drag the server out to a monitor to make BIOS changes or debug TrueNAS boot issues.

I admit that the flaky ports and reliance on Java applets long after Java applets weren't a thing anymore had soured me on the brand, but the very positive support experience flipped that opinion back to very positive.

 

[Ericloewe]

Oh wow, I didn't realize they were still RMAing those boards. I think ASRock stopped years ago. Though perhaps it's specific to the C2x58 models?

 

[Constantin]

When I was ready to graduate from my ASRock-based Mini, I chose SuperMicro for a reason. They support even single-board-owner newbies like me without question, just as iXsystems did when I had issues with my ASRock board (replaced IIRC 3+ times). So, IF I had had a choice to buy a file-server oriented motherboard from like the X10SDV-7TP4F from iXsystems, I would have, in a heartbeat. Instead, I bought it from WiredZone and got excellent support from both WiredZone as well as SuperMicro.

Naturally, YMMV, sample size n=1, the gods may have been smiling, etc.

But overall, SM seems like the go-to shop when it comes to building serious fileserver motherboards. The key issue is researching your choices carefully, for example, I would have been better off buying the X10SDV-2C-7TP4F vs. the X10SDV-7TP4F for my use case. At the time, I thought the extra cores might be helpful, but ultimately the D-1508 would have been a better match and saved me $400.

 

[alugowski]

Oh wow, I didn't realize they were still RMAing those boards. I think ASRock stopped years ago. Though perhaps it's specific to the C2x58 models?

Not just that, they're still selling them new: https://www.wiredzone.com/shop/prod...l-atom-c2758-system-on-chip-2904?category=112

Drop-ship from manufacturer. A whole portfolio of C-2000 boards can still be bought new: https://www.wiredzone.com/shop/category/components-motherboards-112?filter=8-84

 

[Constantin]

Ask it another way: if your use case is a small, low power, file server, that board is decent even if it’s ancient now. The folk at iXSystems apparently wrote some custom drivers to deal with some of the flakier components, the BIOS issue is resolved, and Intel stopped making C2000 with a built-in time Bomb. These days, I wonder if it’s still easy to get the RAM for the thing. But on a replacement basis, it’s likely decent if you need a simple HDD-based file server with a lot of SATA ports and require a mini-ITX form factor.

But, the pricing is a bit high as the x10sdv-2c-7tp4f offers even more SATA ports, has all sorts of other ports and features for TrueNAS, yet costs less once you start adding up missing features on the ASRock board (10GbE, additional SATA ports, 2 PCIe 3x8 etc). But your case has to be able to accommodate the bigger board... and it’s for a specific use case, ie dedicated SMB/AFP file server where the low number of cores is not an issue (yet fast clock speed is perfect for single thread processes like SMB).

 

[Ericloewe]

DDR3 is on its way out of volume production and turning into a niche product for legacy applications. Still, there are tons of used DIMMs out there waiting to be repurposed.

 

[Constantin]

Yup. At one point I had a whole 64GB of RAM in mine for less than $300 IIRC thanks to eBay. But two of those sticks died later. I reverted to the original OEM RAM and live with 32GB now. Works just fine as a file server without VMs and like add-ons.

 

[Ericloewe]

I'm buying a Supermicro C2758 system (Chassis + PSU + Motherboard) used off of eBay. Apparently its former life was some thing for cinemas to manage the endless stream of ads they throw at you after you've paid for a ticket and popcorn.

Am I:

1. Crazy?
2. Adventurous and eager to try my hand at repairing the thing?
3. Curious to see if Supermicro would RMA such a system without original proof of purchase or anything?
4. All of the above?

I guess I'll find out sooner or later. The package is 200 bucks, which about what the chassis+PSU combination can be bought for with some shopping around for deals, FreeBSD 13 finally has support for QAT on C2000 and C3000 CPUs and there's a decent chance I can work around a degraded SoC... So I figure I'm not completely insane? Maybe just a bit delusional?

 

[Redcoat]

I'm buying a Supermicro C2758 system (Chassis + PSU + Motherboard) used off of eBay. Apparently its former life was some thing for cinemas to manage the endless stream of ads they throw at you after you've paid for a ticket and popcorn.

Am I:

1. Crazy?
2. Adventurous and eager to try my hand at repairing the thing?
3. Curious to see if Supermicro would RMA such a system without original proof of purchase or anything?
4. All of the above?

I guess I'll find out sooner or later. The package is 200 bucks, which about what the chassis+PSU combination can be bought for with some shopping around for deals, FreeBSD 13 finally has support for QAT on C2000 and C3000 CPUs and there's a decent chance I can work around a degraded SoC... So I figure I'm not completely insane? Maybe just a bit delusional?

Good luck! Do keep us informed…

 

[Ericloewe]

So, the C2758 system got here earlier today. After playing around with it for a bit, I'm 90% confident it's been pre-bitten by the C2000 degradation issue. Zero boot progress on VGA, no beeps even with no memory installed. "Helpfully", IPMI was configured by a paranoid former owner (Look man, moving the BMC web page to port 65000 does basically nothing for your security while pissing off your admins), so ADMIN/ADMIN is out.

Since at this point I'd need to set up an oscilloscope to probe the SPI bus from which the system firmware is loaded, I'll just pack up and send it back for a replacement or something.

Nah, I'm kidding, I'm totally going to set up the scope and try the resistor "repair" to see what happens!

 

[Ericloewe]

LPC clock looks deader than Elvis. Not marginal, not "yeah, that's not in spec", completely dead. So I dropped in a 220 Ohm resistor between pins 1 and 9 of the TPM header, as widely suggested on the internet. Well, that made a difference:

I mean, sure, it looks like crap, but the "before" capture would have shown nothing - to the point that I was wondering if the scope was even working. 2.42 V sounds awfully low for a 3.3 V bus, so I think the situation is marginal at the moment rather than a good fix.

 

[Ericloewe]

As also reported, the "fix" does not fix the connection to the BMC, which is no fun.

 

[Etorix]

So I figure I'm not completely insane? Maybe just a bit delusional?

Not even. You've got a free motherboard along with your chassis+PSU combo, which is objectively a good deal.
Now would be the time to try 3. and ask for RMA: At best you get a new board from the bug-free revision; at worse, have fun with that soldering iron…