Hardware RIP Thread

[crashtech]

My Supermicro C7X99 died today, melted two pins on its ATX connector, and all four power pins off the 8-pin connector. I didn't have it that long, so I can't say I am impressed with Supero's consumer-level boards. Their server grade stuff has been impeccable. RIP.

[OrangeKid]

[crashtech]

So many hardware problems! I have a Gigabyte AB350-Gaming 3 that suddenly decided it does not like to run a 3700X at full load anymore. Runs a 1600 just fine, though. Also I have a 2060 Super, my newest GPU, just deciding to go blank screen, or sometimes a muted color, at which time the machine becomes unresponsive even by remote control. It was bought second-hand, so I don't think I can warranty it. I'm wondering if it can continue on as a second GPU.... hmmm....

[biodoc]

Bummer on both accounts.

I think IEC's RTX2080Ti was ok as a second GPU when it went "space invaders" on him.

Power supplies on both rigs ok?

[crashtech]

Bummer on both accounts.

I think IEC's RTX2080Ti was ok as a second GPU when it went "space invaders" on him.

Power supplies on both rigs ok?

Yeah, pretty sure. I swap PSUs to diagnose this stuff, also moved the 2060 Super to 3 different machines before deciding it had a problem.

[crashtech]

...I have a Gigabyte AB350-Gaming 3 that suddenly decided it does not like to run a 3700X at full load anymore. Runs a 1600 just fine, though...

So, today I was playing around with this board, and realized that somehow the BIOS had reverted itself back to F31! I'm not sure if this was the problem or not, but I've re-flashed it back to F50a. Maybe I'll buy another 3700X and give it another go!

[crashtech]

So the 2060 Super is pretty much done now. It worked for while as a secondary GPU but started making the machine unresponsive on an ever more frequent basis.

Also, during a failed attempt to get an ebay P102-100 to work, one of my Asus Z10PCD8 dual-socket Xeon boards stopped working. Luckily (sort of) it will boot with one CPU installed now. Possibly I harmed it by failing to switch off the PSU prior to swapping GPUs, though I have done so many parts swaps it's all a blur.

I'm going to have to enter FB Sprint season with less power than planned, since the budget has shrunk and priorities have been shifted. Dual socket Haswell/Broadwell boards seem to be getting less common and more expensive, it might just be time to move on to single socket AM4 systems.

[biodoc]

Sorry about your string of bad luck crashtech.

[crashtech]

Sorry about your string of bad luck crashtech.

Thanks, biodoc. Part of the problem is that I buy a lot of used stuff, so failure is not unexpected. But the 2060 Super in particular was a pretty disappointing one, $350 for 4 months of use. I don't think I am going to buy GPUs that new and expensive on ebay anymore.

[crashtech]

One of my 1070's bit the dust. All computation errors no matter the project. Looking for a replacement, anyone have something same or similar to sell?

[crashtech]

So now the pump on my fairly new water cooling setup is going bad, making lots of ugly noises. Not sure if I will be able to fix it in time for the Sprint. Also not sure what to replace it with. It didn't last very long, purchased in Feb 2019.

[StefanR5R]

My first pump was an SPC pump as part of an EKWB EK-XLC Predator 360 modular all-in-one cooling unit. I used this unit for half a year in its original configuration with just the CPU waterblock, but then expanded it with a GPU block, and a few months after that with an extra reservoir and an extra radiator. Some time soon after this the pump failed suddenly. Perhaps because the loop contained debris. There was foam inserted into the little integrated reservoir of the Predator, with the purpose to catch bubbles. This foam eventually disintegrated and released particles into the loop.

Since then, I used EKWB Revo D5 PWM pumps in all of my waterloops. Some with a reservoir top, some with a simple delrin pump top (and detached reservoir, placed in front of the pump intake). All of these D5 pumps are working perfectly. I purchased them between April 2017 and December 2017. I've got 3 loops with CPU block + 2 GPU blocks + 2 radiators, in which I run the D5 pump at reduced speed. And I've got 1 loop with CPU block + 3 GPU blocks + internal radiator and external MORA with quick release connectors --- i.e. a loop with some more resistance --- in which I run a single D5 pump too but dialed to top speed.

I'm not in the know whether EKWB uses real Laing D5 pumps, or cheap knock-offs like some vendors do.

[crashtech]

Thanks for recounting your experiences. I have a feeling I will end up with a D5 pump eventually, but this time around I took the easy way out and ordered another of the same, so that I don't have to worry about modifying the loop. Right now, as busy as things are, it will be tough enough just to be able to change the pump at all. Hopefully this was just a premature failure and the next one will last longer, but I won't be counting on it. Maybe I will take the defective one apart and see if it can be repaired, so that I can have a spare ready to go. I did notice that they used to supply a replacement pump for the combo unit, but it has been discontinued.

[crashtech]

I actually ended up with this instead:
https://www.ekwb.com/shop/ek-classic-pu ... -pwm-d-rgb
I'll update when it gets installed.

[StefanR5R]

How many devices and how many Watts does this loop have to cool? My understanding is that this type of pump is for small loops. Perhaps with either 1 high-power device, or 2 low to medium power devices to cool.

[crashtech]

Just the CPU, though the radiator is fairly large. It's the same setup that I did a little build thread on back on AT, I think you saw it. Seems to be humming along nicely, and I like that it is PWM. Maybe that will help the pump last longer, since it can run closer to 60% most of the time instead of 100% like the old one.

[OrangeKid]

Laid to rest:
AMD Phenom(tm) II X6 1045T Processor [Family 16 Model 10 Stepping 0] (6 processors)

Replacement: 3950X

[crashtech]

Well, I don't think that one owed you anything!

[StefanR5R]

About my 4 AM incident:

I was just woken up by a noise, like something fell down, or a floor board expanded...
It smelled like burnt plastic in the kitchen, and I saw at my main computer that the triple-GPU computer, which is the only one in the kitchen, had just rebooted.

Its logs and sensors didn't show anything suspicious; pump and fans were running. When I started the boinc client with Amicable, the noise happened again and I actually saw some smoke rising from the computer. I stopped Amicable, uploaded the ~100 results which it had, aborted and reported the rest, and am shutting it down now. Need to look at it at the weekend.

Only just now I pulled the computer out of its corner. When I unplugged it, I got this:

The cable was 10 A rated. Now on to check whether it self-destructed, or if it was the PSU which melted it.

[StefanR5R]

I have been running ODLK1 and Amicable since then, with all three GPUs active, and so far so good.
Looks like it was really just the power cable which burnt up.

[biodoc]

That reminds me of petrusbroder having an issue with one of his computer years back where he had a power cable failure. There was no visible damage so it took him awhile to track the problem down. I guess that's one more item to add to my troubleshooting checklist.