Self-built Supermicro H8DCL-6F system troubleshooting

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Paianni

New Member
Feb 13, 2017
12
0
1
26
First, the parts:

Operating System/ Storage Platform: OpenBSD/custom Slackware build (Cinelive)
CPU: AMD Opteron 4332 HE (x2)
Motherboard: Supermicro H8DCL-6F
Chassis: orig. Lian-Li PC-7H, now Supermicro CSE732i-5o0b
Drives: IDE DVD-RW
RAM: Kingston KVR16R11D8/8 (x4) [DDR3,PC3-12800,1600MHz,Reg ECC)
Add-in Cards: Intel Dual Band Wireless-AC 7260, WBTUO LPE-8CH, AMD FirePro V5900
Other Bits: Noctua NH-U9DO (x2), DeLOCK eSATAp bracket


The symptom is that the system turns on at the PSU power switch with a continuous high-pitch beep (no display or post), which from what I have read suggested most likely a thermal or memory issue. I reassembled a cleaned (isopropyl alcohol) heatsink and CPU with fresh thermal paste supplied with the heatsink. I investigated whether the use of a 4+4 pin connector in one of the 8 pin sockets (both of which are serviced) could be throwing anything off, I removed all peripheral devices connected to the motherboard and played with different RAM configurations to see if one stick was faulty, including switching between hybrid and normal modes on the PSU, to no avail. There are only Power LED, HDD LED, Power and reset switch connections from the case, and having double-checked them I'm convinced I got the polarity right. I'm using only one CPU, it is in the correct socket for this config, as is the RAM in the correct bank of slots. I changed the jumper for SAS to the disabled position.

Any suggestions as to anything else I could try will be met with gratitude. My last resort is to contact SuperMicro support, in case it is something very simple. Hopefully nothing is damaged.
 
Last edited:

Cool Runnings

New Member
Jun 2, 2016
5
2
3
44
Sweden
Have you found the problem?

The polarity for the power/reset doesn't mater, all they do is short the pins so +/- doesn't mater. If you get the leds wrong, the just won't lit up, but doesn't effect the function of the motherboard.

You have the CPU-fan in the correct connection?

Have you tried to reset the BIOS with the jumper on the motherboard?

Are you sure you have the correct revision of the BIOS for your CPU? Maybe needs to be updated (you need another processor for that, that works with the BIOS-version to update the BIOS)

Are you sure the memory are compatible with your motherboard? Often most memory works, but sometimes the boards are picky...
 

Paianni

New Member
Feb 13, 2017
12
0
1
26
Following this post, I realised I had been sent the wrong RAM, so I had that exchanged. With the correct RAM, the system does not power on at the PSU switch, however I get the same symptom when powering on from the case switch.

Suspiciously, I get a continuous beep even when there is no CPU inserted, so there may be truth in the BIOS issue. However, I would prefer not to fork out on an Opteron 4100 in case my MB turns out to me defective, and BIOS upgrades seem to be advised against by the manufacturer, so my next port of call will be to contact Supermicro directly (well, their UK office). ServeTheHome did a review of the MB back in 2013 with a 4300 series CPU, and yet BIOS upgrades weren't mentioned at all.
 
Last edited:

herby

Active Member
Aug 18, 2013
187
53
28
I have the same board and the single socket H8SCM-F. As I recall I'd get a continuous beep without ram inserted or if the H8DCL-6F didn't have all "+12V 8-pin CPU/Memory Power Connectors" connected, the H8SCM-F was a little more forgiving and would be satisfied with just a 4-pin power for CPU.

All of the mobos were bought used, and in the case of the H8SCM-F I flashed the old BIOS with a cheap 4171 HE before using a Opteron 4300 Series CPU.

I did have trouble with dirty pads on one of my Opteron CPUs but it would boot although it was unstable.
The boards didn't seem that picky about ram. I tried several different DDR-3 DIMMS.
 

Paianni

New Member
Feb 13, 2017
12
0
1
26
I had the board sent to the reseller and back again, and tried again, this time testing it outside the case with the Noctua cooler and Seasonic PSU. It booted up with VGA output and I nosed around in the setup for a bit. Putting it all in the case however, things started to go wrong - the front panel LEDs work now but, no more video, either from VGA or discrete graphics, with the jumper set accordingly. Sometimes the long beep will come in after a minute of idling, sometimes immediately, sometimes it won't boot at all.

I figure I might have better luck with a Supermicro case, but what do you think?
 

herby

Active Member
Aug 18, 2013
187
53
28
I don't think the case should matter really, unless something is shorting to ground under the board, or something isn't wired right. I'd pull the board and try it again on the bench, and then repeat the steps of your install checking at each step to see if it POSTs till you find your problem.
 

Paianni

New Member
Feb 13, 2017
12
0
1
26
The latest: the most common bootup sequence atm is for the fans to spin up briefly, then off again, then back on with no beeps, no post, and just a combo of solid/flashing green lights on the mb. That is when the graphics card is inserted. The GPU fan spins up intermittently. When graphics is not inserted, I usually get a similar sequence, but the old continuous beep comes in about 30 secs to a minute later. Sometimes I get the same old continuous beep immediately after power on, no GPU. It's completely unpredictable. What I may not have mentioned is that a red light flashes every time the beep starts.

I also tried resetting the CMOS and taking the cell out for a length of time. I know it isn't dead because it worked when I was breadboarding it. I think there could be unused standoffs on the case that may be conflicting with something, but I have no way to remove them. I can't see how I could fry a m/b is it provides power to fans, lights, speakers and such like.
 
Last edited:

pricklypunter

Well-Known Member
Nov 10, 2015
1,708
515
113
Canada
Remove the board from the case and re-test it. Let it run for an hour doing something more demanding than displaying a bios screen, maybe running memtest or something. If all is well you'll know it's definitely not the board. Also, count the number of mounting holes on your board, and stand-offs in the case, also, take note of where they are located. Any that are in the wrong place need to either be moved, or removed all together. If they are the rivet, or spot welded type, usually a good pair of vice grips and some twisting action will remove them, alternatively you'll need to break out the drill/ dremel etc and get serious with it :)
 

Terry Kennedy

Well-Known Member
Jun 25, 2015
1,140
594
113
New York City
www.glaver.org
The most common bootup sequence atm is for the fans to spin up briefly, then off again, then back on with no beeps, no post, and just a combo of solid/flashing green lights on the mb.
The part of the quote I've bolded is normal behavior for some systems with a separate BMC (baseboard management controller) when first plugged into an outlet. Fans come on at full speed by default as a safety feature when the BMC isn't controlling them. Once the BMC starts after power is applied, it goes "Ok, I'm awake, system power isn't on - turn off the fans". This is the same reason that fans go wild while flashing the IPMI firmware - while the IPMI / BMC isn't running, the fans go to full-speed as a safety measure. This can be exciting if you're next to the system when it happens, particularly if it is one of the systems with 12 fans that have a default speed of 32,000 RPM.

The rest of the behavior you describe definitely isn't normal. I just wanted to let you know that the first part does not necessarily indicate any sort of problem.
 

Paianni

New Member
Feb 13, 2017
12
0
1
26
I have some bad news. I discovered two stand-offs (they are all built-in, btw) in the wrong place, I covered them with insulation tape and re-inserted the mb. My suspicion is that the action of booting the thing up with the standoffs exposed as fried both of my boards, because I'm still getting the same result. It'll be yet another returns procedure.

Perhaps this is how Supermicro discourages third-party chassis, by implementing non-standard stand-off placements. :(
 
Last edited:

Paianni

New Member
Feb 13, 2017
12
0
1
26
I removed both standoffs shortly after the last post, so there are no more opportunities for shorts. Supermicro refurbed the board I was using, as I can spot a chip in a CPU connector from the previous attempt. For the latest attempt I also had a second CPU of the same model ready to install, which I did, and the RAM are in the correct slots for dual-CPU operation.

I am getting the same result as last time, though I'm not getting a startup with a continuous beep anymore, regardless of config. The GPU does the same intermittent fan spinning as before when inserted, and I still have no onboard video, with the usual VGA jumper set appropriately. In all other respects, everything seems fine; all the fans spin up and the front LEDs and switches work as the should.
 

herby

Active Member
Aug 18, 2013
187
53
28
Can you get into the BIOS with the discrete GPU?

edit: If you can't even get past POST to the BIOS, Opteron 4122's are $5 on ebay, grab one put it in socket one and see if you need to flash the BIOS.
 

Paianni

New Member
Feb 13, 2017
12
0
1
26
No video with the discrete GPU either. During the breadboard attempts from a month ago, both my 4180 and 4332 HE worked fine in this exact board, the BIOS revision is from last year.

Just now I tried removing the CMOS battery, shorting the CMOS (which, if I'm interpreting the manual correctly, is placing/brushing a metal object/screwdriver across the contact pads). I also powered the thing on without a battery inserted, same result.

These boards wouldn't be picky about a couple of standoff holes not being serviced? Is this the purpose behind the rings of metal contact points around each hole?
 
Last edited:

Paianni

New Member
Feb 13, 2017
12
0
1
26
I am clutching at straws now. I thought that maybe one of the plugs on the discrete soundcard was shorting the USB 3.0 card below it, removed it, no luck. Tightened the standoff screws a bit more, obviously not too much as that can cause the board to bend over time. To quote the title of another recent thread, this is a build from hell. :( I've fired off an email to Supermicro support, now that I am convinced that there is nothing I have done with the refurbed board that has caused any damage.

It's been an extra damning experience because I still live with my parents and I only have so much time and opportunities to be tinkering with this. A month ago I effectively retired a ThinkPad T510 as my main system and resorted my work area in preparation for what I thought was now a plain sailing build, as it had just worked outside the chassis. The layout of the room makes it impractical to revert the changes, and it means a bunch of peripherals and externals hard drives have just had to sit unused since then. Perhaps I'm getting a little too personal but I need to make myself feel better somehow.
 

Paianni

New Member
Feb 13, 2017
12
0
1
26
Another update, I have come to the conclusion that as a consequence of being a dual-processor ATX board with sub-standard standoff mount points, it won't play nice with my case, it's almost definitely being shorted somewhere. In a couple weeks I will be ordering an SC732i-500B chassis, which is mentioned in the compatibility list. The old chassis and PSU could find use in a gaming/consumer-oriented system if I can find anyone nearby who wants to build one.
 

Paianni

New Member
Feb 13, 2017
12
0
1
26
I now have a Supermicro chassis, the CSE732i-500b, and I've transplanted the m/b into it. I'm using the stock PSU. On power on the Power LED flashes once, then the familiar sequence of no beep, no video, and spinning fans occurs. All the other front panel LEDs stay off for the duration.

This chassis has the PSU mounted above the board and it has a wider selection of front I/O connections, covering everything except NMI. The Power LED female was a 3-pin type, I converted it to use with the 2-pin male so that the 5V and ground were connected. There were also enough standoff holes inside the case to cover all the holes on the m/b.
 

Paianni

New Member
Feb 13, 2017
12
0
1
26
Update: I eventually dumped the SM m/b back at the reseller after they sent me a board damaged in post. But that saga is still in progress. I decided to move to an Asus KCMA-D8 board, with the same SM chassis. It works, with some caveats - the front power LED won't work for whatever reason, and my 4332 HEs refuse to sing in this board despite an up to date BIOS. I've settled with a single 4180 I had for spares, which is still okay as it's got six 'real' cores. It's probably some obscure RAM limitation, I don't have the energy to find out now, all I'm concerned with is that the build is 'finished' to a satisfactory degree.