TR Forums

Note made to indicate that this problem appears to have been fixed with a new BIOS update. Details on the 2nd page, and wholehearted thanks to all who helped/tried to help/said prayers/exclaimed, "poor sap, I'm glad it's not me!"



- BIF

-----------------------------------------------

It never fails. A holiday is coming, you have to catch a flight, and then some glossy, glittery tech in your life just stops working. Grrr!
The system affected is my main graphics, music, and F@H machine. Although right now it's completely down, in all honesty I won't be severely impacted even if it takes me a week or two to get it fixed, so at least there's that.

My system specs are in my signature below. Here's the timeline leading up to the warp reactor failure...

It started one day about 3-4 weeks ago when booting, I got 6 beeps (I think, but not certain) at boot time. The "main boot monitor" was working and it issued some sort of messages complaining about some problem with the NIC. My motherboard has two NICs and I was in a hurry that day, so I swapped the ethernet cable to the second NIC, did some other stuff, and rebooted this time with success. Well enough anyway. I never swapped to the original NIC to see if the problem came back, and even now I'm not sure that was the root cause.

Now we come to the weekend before Thanksgiving here in the US. It was a few hours before I was due to catch a flight. When preparing to travel, I usually like to install any pending updates and reboot my workstation as well as my home network to make sure that everything, including F@H, is running nice and clean before I leave town. When applying updates, I usually do a "restart" because in Windows 8.1 that really is a "cold start" for Windows. A restart is also a "warm start" for system and peripheral hardware devices, because electrical power is not removed while the machine boots. But that's (usually) minor; it's Windows that I'm really trying to give a fresh start.

Well, this time I did a "shut down" and "start". And that's when my day took a bad turn. The machine issues 6 beeps through the little speaker inside. A few seconds later, it issues the 6 beeps again. Then it takes about 1-2 minutes and repeats. There is no activity on any of the monitors this time.

I've done a few Google searches, and here's a starting list of things to check/try:

Basic Connection Checking:
1. Shut off power supply, wait, and retry. I did this with no success.
2. Double-check connections for keyboard and mouse. Try them in different USB ports. Both are USB and I tried this with no success. The keyboard is known to function correctly; I used it all day today with a different computer in my day job.

3. Replace motherboard battery (TBD)
4. Reseat GPU cards (TBD)

Isolation:
5. Try removing one GPU card, if no success, try changing power cables, then try removing the other GPU. (TBD)
6. Reseat and/or diagnose memory cards (TBD)
7. Try a new power supply (TBD)

In addition to the hardware in my signature, I also have a Focusrite Sapphire Liquid 56 audio/midi interface (Firewire) and a UAD-2 Duo DSP (PCIe with no external connections). The Focusrite has been powered down and unplugged and its absence still won't allow the system to boot. The UAD2 is still plugged into its PCIe slot, but it's been around for a few years now and there's no indication that it is causing the malfunction. Eventually, I'll remove it as part of my isolation efforts.

Can anybody think of anything else I should consider, especially with 6 beeps and nothing appearing on the displays?

I've personally seen failing CMOS/motherboard batteries do stuff like this - issues POSTing (i.e. black screens like this) that get progressively worse, until the system won't start at all (but otherwise seems fine).

I'd wager it's the RAM first, then the NIC, then one of the GPUs.

RAM first because that's the only reason aside from a mis-seated CPU that I've heard the dreaded beeps for.
I've seen NICs go bad and cause BSODs and complete system hangs without a BSOD. A faulty PCIe device wreaks havoc on a PC and can seem like all sorts of issues with other PCIe devices (such as the GPU).
GPUs because I have no idea what else it'd be.

Doesn't that board have an on-board LED BIOS code display? What does it say?

Can't say a lot till you get your TBD list done except you could try in reverse-
Break down to minimum config-boot SSD one stick of ram etc and go from there....................

Update:

1. Replaced CMOS battery. No problem. Saving the old one until I find my mulitmeter and test it.
2. Replacing battery required that I remove the Asus GTX 980 from slot 1. So this card got cleaned and reseated.
3. Replacing battery did not require that I remove the Zotac GTX 980 from slot 5 (or whatever; it hangs off the end of the motherboard), but I removed, cleaned, and reseated it anyway.

4. Reassembling was uneventful, but then I got what sounds like 5 slow beeps overlapped by 3 fast beeps and nothing on any screens. Initial reboots were done with only one USB hub plugged in, and through process of elimination and eventually booting with only a DVI monitor, mouse, and keyboard connected, I was able to get into the BIOS and reset the clock which was reading 2005-01-01 from the battery change.

The Firewire audio interface (a heavy rack-mounted unit outside of the computer) is still unplugged as I write this post (on that same machine). After more process of elimination and about a half-dozen more boots, I have determined that the Asus GTX 980 won't accept any DisplayPort monitors. As a temporary measure, I am running the system with all three of my DP monitors connected to the Zotac and I've moved the single DVI monitor up to the Asus. The system boots normally now and without any beep codes. F@H is running (seemingly normally), and I have confirmed that I can make an internet connection with either NIC on the motherboard.

I'm going to let this run for an hour or so, then I'll shut down, pull the Asus card and double-check data and power connections. I'll try swapping the power connections with the Zotac and see if the Zotac develops a new aversion to DisplayPort monitors. If no success with DisplayPort, I'll go ahead and make all of the remaining connections, restart, and get back to business. The Asus card is still good for CUDA processing while I figure out if I can get it swapped under warranty.

TLDR: Everything seems to be running okay, with the exception of DisplayPort monitors connected to the ASUS GTX 980 GPU, but even that seems to be handling its current F@H Work Unit just fine.

Possible Cause(s):
- Was the battery bad? I don't know yet and I don't have the confidence to say it was the cause.
- Why did the first ethernet NIC star working again? I don't know yet, and may never know for sure.
- Is the Asus GPU bad? I don't know yet. I will check all connections again and I'll swap power cables.
- Is it possible that the PSU is bad? Maybe. Further troubleshooting may reveal a yes or no answer.
- Is it possible that a USB hub or device had something to do with this? Maybe. I have as many as 15-20 devices plugged in at times (music, art), but not usually all at once. I'm holding this out as a possible root cause and may learn more as I dig in deeper and/or begin to plug more things in.

To answer Ned's question: Yes the motherboard has a number of status LEDs on it. They don't give a code or anything; they're just supposed to glow when there's a fault. But they're different colors (some blue, some green; none assigned to status, such as red for error), and they glow at boot too...it's supposed to be better this way but feels more confusing, so I mostly ignore the Christmas tree lights and do my troubleshooting old-school.

The system is up right now, seems to be running as well as ever, and is even folding; wrapping up the old work units that were in there when it failed. I'm going to let it run and reheat the home office while I go make some supper.

"The system is up right now, seems to be running as well as ever, and is even folding; wrapping up the old work units that were in there when it failed. I'm going to let it run and reheat the home office while I go make some supper. "

Wonderful-And supper will taste so much better now................................................

A dead CMOS battery will simply cause your BIOS to lose it's settings and time. Unless having your BIOS reset would create an unstable condition (erasing custom RAM timings/voltages and such), I'd say it's an unlikely cause.

USB devices rarely cause a failure to POST, but 15-20 of them is not an insignificant amount. How many of those are drawing power from the motherboard, and how many are connected to self-powered hubs? If your PSU is already struggling to power both cards, a large number of USB devices could push it over the edge.

If you still get grief, #1 culprit is still RAM.

Flaky RAM exhibits issues in multiple ways but when you start to get weird noises from the PC speaker and beep codes that aren't documented (or consistent) then it's likely duff RAM.

Put one stick of different RAM in the board if you can, see if POST is fast and reliable like that. Then try each of your other RAM sticks one at a time. Test each stick 3 times since flaky RAM can give inconsistent results and just booting the first time doesn't mean it'll boot the next time or not.

PSUs can also cause inconsistent grief like this and being Thermaltake I'd chalk that up as suspect #2

Update:

Well well. It's not (necessarily) the GPU. I did some process of elimination, and was able to make it misbehave with the Zotac and work correctly with the Asus, enough to say that I don't think it's the GPU. I swapped the PSU cables to each GPU too, as well as the port each PSU cable was connected to on the PSU itself. Failures do not follow the GPU, nor the power cables, nor the connections on the PSU.

In fact, the failures happen regularly with...

Observation #1. A cold boot (booting after a Windows "shutdown") can make it fail; more than 50% of the time, and now more than 100% of the time in the last 2 days. Power does not need to be removed from the PSU (via switch or by unplugging PSU) for it to fail on the next boot attempt. Once it fails, I can swap the monitor cables (moving DP cables from one GPU to the other and the DVI cable from the other GPU to the one). Then it reboots without any beeps or codes.

Warm boots (a "restart" from Windows), have not caused failure as yet. This makes me wonder when this problem began. First clue was at the install of the Liquid 56 a couple months ago. My system is on a UPS and only gets "restarted" to bring in MS patches. Even a power outage in the summer just resulted in the termination of the F@H client, after 6 hours, power was restored and the system was still running. It never even had to be put to "sleep" mode, and it just automatically resumed F@H slots from the last sync points. So even then it didn't get restarted from an "off" state back then! But Firewire wants you to power everything down before plugging/unplugging (per warnings from Focusrite), so that's what I did when I installed the Liquid 56...maybe the first time it got a cold boot since early 2015.

So this problem could actually be many months old and I wouldn't have known it.

At this time, I don't (yet) suspect memory, because this seems to be mostly repeatable and it has to do with "shutdown" and reboot.

Observation #2. Sometimes the power button (connected to the motherboard front panel connectors) acts odd. Press button, power comes on but just as soon as you release the power button, the system powers off. May happen as much as twice before the system will stay on and try to post. Touch button (without pressing it), power goes into "standby". But only sometimes. Power button shenanigans "may" be happening when I get the failures to post. I'll pay better attention going forward.

Analysis: I am starting to suspect power supply issues. It's very possible that I have not done a shutdown/restart since January/February. The fact that I never power this system off and now when I do the problem happens...that's pretty regular. It "could" be a motherboard issue and I'm still holding out the possibility that it could be memory related, but odds of either of those right now are very low I think.

At this time, I do not suspect the GPUs or USB devices (I can produce the problem with only a mouse and keyboard plugged in, so it's not a problem of too many devices right now). I do not suspect any SSD, hard drive, or CPU issues, and I don't suspect Windows in any way (since this happens at post, long before Windows is loaded).

Going forward: At this time I'm thinking of replacing the power supply. Worst case, it doesn't fix the problem and I end up with a second unit ready for a new build. It's a good start and the part would be reusable. I'll also run a Memtest this week and see what comes of it. I have suspended all F@H activities until I decide what to do. I definitely don't look forward to the prospect of trying to diagnose 8 sticks of 3 year old Corsair memory that was matched by the manufacturer, with half of them obstructed by the CPU radiator, requiring its removal and possibly the removal of the water pump from the CPU too. Memory diagnosis will be time-consuming and messy, messy, messy, and I am concerned about getting a meaningful return on my time for this system.

But first things first. PSU is an easy try. Sort of. It will still take a few hours to reroute all new cables and I don't have that much time these days. But still.

EDIT: After I made this post, it occurred to me that maybe PSU issues could be getting amplified by having 2 GPUs. I may try retesting the system with one GPU removed, alternating them to see if the absence of one GPU (and its attendant power requirements/cables/PCI connections) might result in a return of reliable cold-booting.

Question 1: Is there some sort of "motherboard reset" that I should have performed after changing the battery; maybe something that is occasionally glitching the motherboard and GPUs after a "shutdown" is performed?

Question 2: Is there such a thing as a PSU tester for the 12 volt cables? I think I have one for the motherboard connection, but if the root cause here is with the PSU, I'm thinking that I could be getting fluctuations on the other cables or their connections.

The power of elimination.(pun intended)

Observation #2. Sometimes the power button (connected to the motherboard front panel connectors) acts odd. Press button, power comes on but just as soon as you release the power button, the system powers off. May happen as much as twice before the system will stay on and try to post. Touch button (without pressing it), power goes into "standby". But only sometimes. Power button shenanigans "may" be happening when I get the failures to post. I'll pay better attention going forward.

Assuming (not a good idea on times) there's not multiple problems,you might have found the problem.
Through PSU and MB components could be the cause,I'm going 90% with dodgy power button/cable...........
Best case if you could salvage one from a old case and use to test...........................
If not you could do the screwdriver job across the pins-through that's a real pain in the a$$......

Question 2: Is there such a thing as a PSU tester for the 12 volt cables? I think I have one for the motherboard connection, but if the root cause here is with the PSU, I'm thinking that I could be getting fluctuations on the other cables or their connections.
Most testers just check if there's a voltage and within tolerance -Nothing a $10 multi-meter can't do..........................

Funny pun; thanks for the laugh!

...another reason for me to find that damned multimeter. It's in the garage someplace, probably guarded by an army of roaches.

Power button...I don't have any old cases...wait, I do have one; different maker. I'll dig it up and see if it's compatible.

The more I hear, the more I'd suspect the PSU. Give it a shot with only one card and see how it goes.

The case power button is just a momentary contact switch that talks to circuitry on the motherboard. The motherboard then sends a signal back to the power supply saying "send me power". This is all powered off the 5V standby line that is always powered when the PSU is plugged in. You don't need a power button, you just need to short the two pins on the motherboard header to get the system to turn on and off. When I'm working with a bare motherboard for an embedded project I just use my house key to power it on by shorting the pins briefly.

From your description, your problem is in either the PSU or the circuitry on the motherboard that deals with the button and sending the signal to the PSU. Odds are the PSU, try a different one first before trying a different motherboard.

Update:

I discovered that my motherboard really DOES have an LED code display. It was hidden all along by the Zotac GPU. Wacky stuff! And I got some codes to share with you too. More on that in a minute.

I pulled out the Zotac. With the Asus GPU, mouse, keyboard, network, and 3 DP monitors plugged in, I am still getting the failures to post AFTER a shutdown. "Restart" from Windows will sometimes restart (which is successful), but will also sometimes result in a shutdown, after which I'll have the failures. Messing around with moving 12 Volt cables will sometimes fix it and allow a boot.

The aforementioned LED codes show this:

...about 6-10 sets of codes show up as the system begins startup...
...then...
d3 - First set of beeps "some of the architectural protocols are not available"
97 - second set of beeps, almost immediately after d3. "console output devices connect"
98 - no beeps "console input devices connect"
0d - here it's in the ditch "reserved for future AMI Sec error codes"

I think it's the 97 that's getting me, and I do still suspect the PSU, given the goofy power button behavior and the fact that swapping the 12 volt cables (or DP cables when both GPUs are installed) seems to allow a cold boot.

So I'm looking for a 1000 to 1200 Watt PSU for quick delivery. In the meantime, only cold booting is impacted right now, so I think I can put the Zotac GPU back in, get a clean boot and let the machine continue with F@H wile I wait for the new PSU.

Any PSU suggestions? Budget is under $400, and the new PSU needs to be able to work with my UPS with that sine thing that was implemented a few years back. Yeah, that's a real scientific way to mention a requirement, but I think somebody here will know what I mean...

Also: Yes, I do think that the motherboard can't be declared "not guilty". But getting a good quality 3-year old model will require more time and effort, and the PSU is an easy first try anyway.

Wait... is your UPS a true sine-wave, square-wave (crappy), or somewhere in between?

CyberPower. This one. So the answer is yes.

More Info:

Cold boot fails with either the Asus or the Zotac plugged in, or both.

Cold boot will always fail when power button cutout happens. Every time it does this:

Press button; power on for <5 seconds...then power off by itself.
Stays off for about 5 - 7 seconds, then powers on by itself and off again in quick succession.
Stays off for another few seconds, then powers on, stays on, and beeps + post failure occur shortly thereafter.

^ happens every time. This is consistent.

Also consistent:

If I can get to it and hold the power button down and force it "off", then the next attempt to boot will be successful. Usually, I think.

This is very consistent with 1 graphic card installed, and if memory serves, it did the same thing with 2 graphic cards installed.

I am wondering a little bit if maybe there's some sort of bios setting that's tripping it up here...but then why would it have begun this year and not last year or before? Checking my UEFI bios settings to see if there's some sort of "safe mode" I can try. Also wondering if maybe there's some setting in Windows 8.1, like maybe a "get ready for Windows 10" item that could have arrived in a patch during 2015.

Have you tried disabling "Fast Startup"?

Control Panel -> Power Options -> "Choose What The Power Buttons Do" (on left) -> "Change settings that are currently unavailable" (link) -> Scroll down -> uncheck "Turn On Fast Startup"

AC power is a sine wave, and better UPSes like yours try hard to emulate it - which is good. Your UPS ought to work with most devices, though it's always good to watch the wattage - particularly if you're talking 1kW (output - it's drawing more from the wall/UPS) power supplies. Know what your load will realistically be. Cheap UPSes, on the other hand, use a square save that sorta-kinda-crappily fakes a sine-wave. They're cheaper, but some devices (fans/motors, some power supplies) won't run properly on them. A lot of switching power supplies (I think that's the one?) don't give a damn, so it doesn't matter. I have a cheap APC (BE550 I think) - my monitors, SFF HP, laptops run fine off it, and phone power bricks run fine off it, but a 120V fan slows way down (likely not good for it) when the UPS is on battery.

Fast/hybrid boot is worth looking at - I haven't tried this, but I'd imagine Windows gets confused if it's resuming the kernel, only to find the video card changed. You can also temporarily bypass it and do a true shutdown with "shutdown -s -t 0" from the command line. Might be other ways as well.

Update:

I tried a number of new things. Here are the results. Numbered, but really just for my brain and not necessarily in the order I did them.

1. I put both GPUs back in. Similar issues, except now every once in awhile wacky power button behavior will still happen but system will post anyway without beeps. A sign maybe that post and power button issues are 2 different problems? More on that below...

2. I pulled the Zotac again and reseated all of the front panel connections, which the 3-slot-taking Zotac covers better than a solar eclipse. It took 3 or 4 cold boots okay, then it failed to post with beeps again.

3. I put the Zotac back in. No better, no worse; still cold boots okay a few times, but eventually fails to post.

4. I pulled out the front panel switches and shot the power and reset switches with some contact cleaner, then worked them a few dozen times. This didn't cause any trouble, but also didn't seem to fix the kooky power button problems; plus, the system still cold boots a few times okay, then will fail.

5. Fast Boot - I have that disabled Fast Boot at your suggestions. Results same as above. The issue still happens, but more rarely than before. Maybe now only once every 3-4 cold boots. But I've also done all that stuff above, so can we say that helped? I'm leaning toward "doubtful but maybe".

6. During failure, I swapped the power cable to an alternate (non-sine) UPS. It booted a few times, then failed just like above.

7. I swapped power back to the original UPS. Again, cold boots okay a few times, then fails.

8. I swapped power cable over to my Furman power conditioner, which is hooked up to the older UPS (non-sine). The power conditioner is supposed to help further maintain steady power, and that's what I plug my audio recording interface into, for microphones and music instruments. I know that the UPS's and the PSU are all supposed to do flatten out power issues by themselves, but this is just a test.

9. USB devices and other peripherals are not yet plugged in; just mouse, keyboard, and of course internal/motherboard USB devices (CPU water pump, Zotac's OC+ port, etc).

I think that odds still favor this being a PSU issue, with the motherboard being the next possible issue.

I have not yet re-seated the main PSU plug to the motherboard. Does anybody think this will help?
Any thoughts on good PSUs in the 1000 to 1200 watt range to support the two GTX 980's and the old i7?

Right now I'm planning to break it down, remove an old fan controller and its probes (just for general cleanup, not because I suspect these), then transfer two case fans to the motherboard plugs. Then I'll do a quick job on the cable management and see if I can get a good boot on the new UPS and leave it up until the new PSU arrives. If the new PSU doesn't fix it, I'll reassess at that time.

Here's a comparison page of the PSUs I'm looking at right now...

Edit: I just noticed that these new PSUs all seem to have 7 year warranties. I wonder if my old Thermaltake has a long warrranty too...

I think you've tested everything you can possibly test at this point. The #1 suspect is still the PSU, and for something of that caliber I would only get a Seasonic. They have a handful of different 1050-1200w units in the $200-230 range.

Thank you, yes I have tested just about everything. I'm ready for a nap!

I've heard Seasonic is good, so I'll keep the option open when I go shopping...soon!

BIF wrote:

Here's a comparison page of the PSUs I'm looking at right now...

Edit: I just noticed that these new PSUs all seem to have 7 year warranties. I wonder if my old Thermaltake has a long warrranty too...

The EVGA has a 10 year warranty...........
My recommendation is Seasonic first then EVGA...........................
Through don't know if they play nice with your UPS................

edit.
"I have not yet re-seated the main PSU plug to the motherboard. Does anybody think this will help?"
Well it won't hurt-and you could do the paperclip test a few times-green wire to earth.
http://support.antec.com/support/soluti ... rclip-test
Through you should unplug everything connected to PSU................

Your UPS is line interactive so it will be just passing the mains power straight through most of the time. If the mains power is a little over or under voltage then it brings in a circuit to cut or boost the voltage to get it back to normal. It's only if the mains power is way out of spec - including cut off entirely, that the battery powered inverter starts up and you get to worry about whether it is full sine wave or not.

Yes, go for a new PSU. Sorry I don't know which ones are good.

That's correct - I tested this with a fan on a cheap square-wave UPS. Fan runs fine when the UPS is plugged in, but slows down and goes to hell when on battery. But his UPS is sine-wave anyway, so he should be fine.

I just bought a couple items on Amazon.

EVGA 1200 W PSU
Thermaltake PSU Tester

Should arrive Monday; can't wait to test the old PSU and see if the new one stops this messy boot/noboot behavior.

Yeah, the tester may have been more expensive than the multimeter I currently have (an inexpensive model from Harbor Freight; a discount tool chain in the US). But this tester will be a nice convenience since I can keep it in my home office and because it will help me easily test any PC power supply and/or cables.

Today also marks a big change for me. I compared Newegg and Amazon for this order and for the first time, I picked Amazon over Newegg to order PC equipment that was available on both sites. I have noticed for a few years now that Newegg's prices just haven't come close, but I haven't had to place a significant order for a couple years. In fact, I don't order computer hardware often enough to justify a Newegg premium account no matter what their "subscription" price is. I hate subscriptions because they make us poorer. So the fewer I have, the better. And I get a lot of bang for my buck from my Amazon Prime account. I can order books, music, watch TV, get camera accessories, replacement kitchen blender blades, and vacuum cleaner filters, and even weird off-kilter stuff like little stands for making tacos! And whatever I order arrives next-day (usually). Amazing.

I'll let you all know how it works out; will probably be a couple days.

I am installing the new PSU this evening. I'm also removing an NZXT fan controller and changing the wiring so that all fans run off the motherboard.

Question: I'm short by one fan header on my motherboard. I have two identical fans located on the bottom of the case, blowing up into the case. I'm going to use a Y-splitter to run both of these fans off of the nearby motherboard "cha_fan_2" header. Header and fans are configured for 3-pin operation, as is the Y-splitter. My Y-splitter has black, yellow, and red leads going to both fans.

What do I need to do in the BIOS to ensure that these two fans are properly reporting their RPM to the motherboard?

Edit: one forum post (not TR) I found said to disconnect the yellow lead from one of the fans if you want both to have their speeds controlled by the motherboard. And I do. So is this true?

BIF wrote:

What do I need to do in the BIOS to ensure that these two fans are properly reporting their RPM to the motherboard?

Edit: one forum post (not TR) I found said to disconnect the yellow lead from one of the fans if you want both to have their speeds controlled by the motherboard. And I do. So is this true?

I never messed with the splitter wring myself when I used one like you are attempting. I did check what the bios was reporting for RPM with both fans connected, and it was alternating between the two fans RPM reported. Fans spun up on demand last time I did use the splitter in my older AMD Phenom II build.

Thanks. I may test it with only one fan connected, and then again with both connected, and just see if they report differently and/or if they spin up differently. I'll be doing a lot of reboots anyway this evening.