I just purchased used, a H11SSL-i Rev2 and a Epyc 7532 and I’m chasing some issues on first setup (Likely a help thread waiting to happen!)
I’m at the point of trying a BIOS update, I’m on 2.1 and there’s a 2.6a available.
Im on IPMI version 01.52.00 and it has an activated License, looking through some threads in my search for an answer it would seem this isn’t always the case and limits some features of the board when no license is in place?
I’m wondering if just updating BIOS will require me to reenter this license code (which I don’t have) or is separate and should go as smoothly as butter
Also another likely stupidly obvious question, there is an IMPI revision to 1.52.21 also available which has the caveat " Please note that IPMI firmware version v1.52.18 or newer is with the key signed. Once you upgrade to this firmware version, the device cannot rollback to the previous un-signed IPMI firmware." Could anyone explain to me key signed? and again if I would need to reenter the license key I don’t have after updating the firmware?
Thanks for the reply, and I understand that, it was cool to not have the board populated with hardware and be able to still get access to information about the board.
I suppose having not dealt with any real “enterprise” specific hardware Im just being cautious not to fall into any trap designed to get more money out you for subscriptions and licenses.
The signed firmware requirement helps ensure the firmware update you’re trying to install actually came from SuperMicro and not some bad actor who tweaked the file for nefarious reasons.
The only license in the ipmi is for “out of band bios updates”.
Without it, everything still works except updating the bios from the the ipmi web-interface.
Up to and including the H11 series, those licenses depend on the BMC management NICs MAC address and can be “easily” generated yourself should it disappear in an update.
I can’t specifically remember it ever disappearing in an update i did.
So i would not expect it to, but if it does, not a catastrophe.
No not created one yet, I was hoping that I might be able to resolve it myself.
If I power off the machine, via IPMI or the power button or I’ve got proxmox on it atm, thats the same. The machine wont post again till I remove power.
So when it wont post:
no vga
IPMI continues to function
POST Snooper shows EITHER stuck on 54 (Unspecified memory initialization error.) or its cycles around its usual numbers, hits 54 stays there for 30seconds then cycles again.
BMC Heartbeat light is normal
Board status light is normal.
Will ALWAYS boot up fine after a CMOS reset, but this involves removing power so think that was a red herring I was chasing.
Remove power, Power back on and its stable, all functions work.
I’m using 3200mhz 32GB sticks x4 [M393A4K40DB3-CWE]. This behavior doesn’t change with 1,2,4 sticks.
I’ve remounted the CPU (I’ve got a torque wrench)
BIOS reset
Bios Update
Empty board, in case or out.
I’m going to order some different lower speed ram next I think, and try it. Only 1 stick though.
Hmmm, one thing i can say with certainty is that the meanings for postcodes you are quoting, aren’t actually correct.
That is because AMD has multiple sources for POST Codes and even different types of POST Codes.
On a two digit display, you can’t see the other higher digits telling you more.
Additionally, the webinterface is a bit sluggish.
You probably won’t be able to see the actual codes happening quickly at the end before the board resets.
The only real display i trust is a 7 segment display.
Really helpful would be a history of POST Codes.
I still need to try a RPI Pico for that.
Sorry for that tangent.
I think it is unlikely to be a memory issue.
It might be a defective Bios EEPROM.
Or possibly physical damage.
Common places are around screw holes and around the BMC where the io-shield of a PCIE card could hit the board.
Since it is a revision 2, you could cut the bios in half and mirror the top 16MB to also be the lower 16MB.
If there are defective cells in the top, that is gona put the BIOS you boot from into a different place.
My suspicion has actually been Bios EEPROM. And as I’m still within my return window with the Chinese eBay seller I got him to tell me to try the BIOS update just incase writing over the BIOS killed the board.
I’m considering 1 of 2 things:
Return the board, time consuming potentially costly depending whos paying postage.
Suggest a partial refund and I live with the issue and potential degrading board. I’m on a gen1 board with a gen2 chip because A Price and B right now gen4 PCIE isn’t a necessity but I wanted the efficiency of gen2 Epyc. And the plan is to jump to a H12 when prices drop and play with some NVME storage. So it might work in my favor.
I will check the board, but with this not being my first build by a long way, im already aware of the areas you mentioned getting knocked so I’m gentle.
I’ve read about the gen1 having a 16MB and gen2 having a 32MB, and about removing / recoding them which is interesting but I doubt I would try that unless I’m a year in, the board dies and I dont have the £350 for a new one.
Are you testing this in a case or outside of the case? If inside the case, have you check if you have removed extra ATX standoff in the case prior to installing the board? If using regular PC cases there should be one standoff directly above the RAM trace that should be removed.
