the issue is with the U4..U5. that are Voltage regs sliced out of the wafer (bare silicon)
H12SSL-I Stuck at "bmc initiating"
- Thread starter Dick Lee
- Start date
-
- Tags
- bmc boot failure supermicro
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
Hi all,
I didn't take my own advice and picked up another H12SSL-I that is "new" (listed new, came in OE box, seems new etc)
I was able to get into the BMC on this one but I'm not getting any iKVM function at all. It shows me the loading graphic then a white window in the html5 mode and the java one shows a black one with no video. (yes it's powered on and the CPU fan is spinning). Have tried on 2 browsers and 2 computers.
I've updated BMC and BIOS firmware with the boxes to save configuration unchecked, I've tried a VGA cable to a VGA monitor and also gotten no video output. I checked the jumper on the board it's in default position so video out should be "on". I pulled the CMOS battery and shorted the pads for a few minutes as well.
None of my hardware is known good as I've just gotten it on ebay.
Epyc 7402
MEM-DR464MC-ER32 64GB DDR4 3200MHz ECC 2Rx4 RDIMM x 4 sticks.
Feel like I'm at the end of my troubleshooting knowledge but hard to believe the CPU or ram could be causing this right? In the BMC dashboard it has hardware loaded that is not mine so assuming that's a test stand in the factory.
Do i just RMA or what?
I didn't take my own advice and picked up another H12SSL-I that is "new" (listed new, came in OE box, seems new etc)
I was able to get into the BMC on this one but I'm not getting any iKVM function at all. It shows me the loading graphic then a white window in the html5 mode and the java one shows a black one with no video. (yes it's powered on and the CPU fan is spinning). Have tried on 2 browsers and 2 computers.
I've updated BMC and BIOS firmware with the boxes to save configuration unchecked, I've tried a VGA cable to a VGA monitor and also gotten no video output. I checked the jumper on the board it's in default position so video out should be "on". I pulled the CMOS battery and shorted the pads for a few minutes as well.
None of my hardware is known good as I've just gotten it on ebay.
Epyc 7402
MEM-DR464MC-ER32 64GB DDR4 3200MHz ECC 2Rx4 RDIMM x 4 sticks.
Feel like I'm at the end of my troubleshooting knowledge but hard to believe the CPU or ram could be causing this right? In the BMC dashboard it has hardware loaded that is not mine so assuming that's a test stand in the factory.
Do i just RMA or what?
sure, dead CPU and/or RAM can cause this.
the kbd numlock led any sign of light after some time ?
I'm not even getting keyboard rgb backlights let alone num lock. Have been trying each stick of ram solo with the CPU in the furthest down slot and haven't gotten a post or bios flash of life.
Hello, I am currently encountering a similar problem, so I would like to know if you finally solved it and what is the solution?
My CPU is EPYC 7K62, and the memory is Samsung DDR4 3200 32G*8
I'm facing same issue , i just ordered the voltage chip
but could we apply something like this : https://m.media-amazon.com/images/I/71AtI-a+3tL._AC_UF1000,1000_QL80_.jpg , heastink ?
The board in the picture is : ASRock Rack ROMED8-2T
but could we apply something like this : https://m.media-amazon.com/images/I/71AtI-a+3tL._AC_UF1000,1000_QL80_.jpg , heastink ?
The board in the picture is : ASRock Rack ROMED8-2T
Last edited:
the chip is bare silicon, a touch of metal and it bricks again. just apply constant air flow(like usual in server chassis)
and install/swap PCie cards only with removed AC(wait longer to discharge capacitors)
but i think this board is similar like supermicro right? even the heatsink hole is the same
try and report back.
btw: the newest rev. of H12ssl has these VRMs on the back of PCB.
It will not make a difference.
If you crack the die, you are lucky if it doesn't bust the BMC. It will have issues then in any case.
@m7mdcc
I tried replacing those already and one thing i can say is that just a "hotair" is not enough to do it*
I still wish you success and a working board!
But also please check if the BMC is still fine on that rail. Because if it is not, as in it being dead, replacing the VRChip will not revive that.
*reliably or in any way nicely.
BTW: As far as I can say, Rev 1.10 has replaced ALL Bare-Die VRMs / WCSP (Wafer Chip Scale Package) VRMs against normal packaged variants.
last owner sayd they just moved to the back in rev. 1.02
i hope they moved also from TI to another vendor.
Last edited:
Just another packaging would be fine enough.
TI is as far as i know, pretty reliable if used within spec.
To me, the location and usage of the wafer-scale package is basically a trap to the user by SM.
the issue with TI:
they do not test parts on voltage dropout conditions.
means what happens if the +5 voltage slowly sinking down after AC loss.
that is the point they brick.
Do you have any source for this claim / scenario?
What do you mean with them Bricking
TI engineer forum. you can get around IF you know the existance of the problems.
problematic X11DPi up to rev. 1.21 (VRM bug) was a TPS xxxxx TI model, replaced from rev.2.x by MPS VRM controller.
example from the past, linear voltage reg brick if output voltage is higher than input voltage(output el.capacitor)
solution: diode from output to input, later integrated.
the TI VRegs used by supermicro do not brick only ba machanical damage, switch off switch on can be enough.
if all is ok, there is no reason to replace them with otherparts in rev. 1.10