GPU RIP

[kriesel]

The errant gtx480 has been removed. The caps all look fine. Dust and lint was not bad under the cover, which I've had off before.

[kladner]

Originally Posted by kriesel

With EVGA Precision XOC, I reduced the core clock and memory clock as far as it allowed; core went from 725.6 to 423.4, memory from 950 to 699 Mhz as confirmed by GPU-Z. For test speed since the previous test had plenty of error count, I reduced it from 3 passes to 1. GPU temperature dropped from high 90s to 79C. Average bit error rate went up, from 2.1E-10 to 3.3E-10. The low and high portions of the memory range had zero error, while the middle worsened. Blocks 0-22 and 44-55 had zero error, plus a couple near the middle (31, 32).
GTX480s seem to run their fans full out regardless of load. This is not its first Wisconsin summer, and my air conditioning works.

480s are little furnaces, and 580s aren't much better. A 460, on the other hand, is my first compute GPU, and still running quite fast compared to its specs. Still turning out 209+ g-d/d at 67 C, hitting 70 C on hotter days. Yes, it uses about the same power as a 1060, for 1/3 the output. Still, not bad for a $100 clearance item at Microcenter.

[kriesel]

Happily ensconced in an HP Z600 since a few months after my entry into gpu computing in GIMPS, this morning the system tanked and could not be revived without some hands on hardware intervention. (Not even BIOS startup or display monitor activity.) Diagnostics for the Z600 say four red on-button flashes and beeps is a power supply problem or component problem; remove all removable components and then add back one at a time until the problem is recreated to localize the issue.

Subtracted the DVD and 2 other gpus, still a problem with the GTX1070 remaining; swap out the 1070 for the 1050Ti, ok; add back the DVD and other gpu, ok.

So the GTX1070 will get a trial in another system to confirm it's bad. That's to rule out the possibility of the system power supply having degraded to where it no longer can carry the GTX1070, and that there's likely an issue with the gpu itself.

It is, conveniently for the manufacturer, a few months past warranty expiration. Nuts.
Is there any point to trying to repair an out of warranty 2+ year old gpu these days? FYI it was a Zotac GEFORCE GTX1070 MINI 8GB 288-1N445-030Z8, apparently almost 27 months start to finish.

[kriesel]

Two tiny chiplets (larger is ~4mm across) appeared charred and were very easily dislodged from the circuit board along with some smaller debris. Part of the conformal coating was burned off and the copper layer beneath appears oxidized from heat. It's toast. Next stop recycling. It's worth noting Zotac did not even respond at all to an inquiry about out-of-warranty repair, made months ago.

[PhilF]

If you're just going to recycle it, I'll pay for the postage if you'll send it to me. Sometimes I can fix them. :) PM me if interested.

[kriesel]

The GTX1070 went to recycling, after it was determined not worth trying to fix.

My remaining GTX480 appears to have given its last, shortly after completing tuning and benchmarking for max TF output with increased GpuSieveSize beyond 128.

One of the Quadro 2000s fails to run code, despite being hooked to the same extender and system as another runs fine upon.

The failed Quadro 2000 prevented an old cpu system from powering up. That system is now running ok with a Quadro 4000. But the fan is failing on that gpu, which probably led to a bad factor recently. (The usual 38814612911305349835664385407.) It's probably not worth fixing since a GTX1650 is around $130, takes less power (at 75W needing only the PCIe slot, no additional cable), and is 7 times the TF speed, twice the primality or P-1 speed, and has more memory. Power savings pay for the upgrade in 2.5 years here.

There are some indications that bad gpus may damage the extender cards they slot into. Caps at the opposite end from the USB connector heat up along with the 3-lead device there. Or maybe they're just limited-life capacitors. They LOOK fine.