GPU Fans ramped to 100%, no post/video signal.

[Sufi Mindtricks]

So I found my system off this afternoon.  It was on and running folding@home before I went to sleep, no issues, temps were normal - 65C on the CPU and ~75 on the GPU.

I went to turn the PC on, got no signal on screen, even though the mobo RGB was on, then I heard the 1080 just start howling, ramping up the fans to max 100%, non stop, still no signal.

 

Specs:

 

Ryzen 2700x

ASUS ROG Crosshair VII Hero (Wi-Fi) AM4 AMD X470

G.SKILL TridentZ RGB Series 16GB (2 x 8GB) 288-Pin DDR4 SDRAM DDR4 3200

EVGA GeForce GTX 1080 FTW2 GAMING, 08G-P4-6686-KR

2TB nvme SSD boot drive + 2 Sata SSD storages drives.

Corsair H115i

Seasonic FOCUS GX-850, 850W 80+ Gold

 

I turned off the PC, killed the power, left it for a few hours while I was working, came back in the evening to find no change.

I cleared CMOS and tried to boot it, no change.

Reseated the PCI power cables, no change

Reseated the Memory, no change.

Reseated the GPU, no change.

My friend had a spare Zotac 1070 mini, swapped it out, and it booted w/signal.

 

So it looks like my GPU may be dead...?  Anyone encounter this before?

Here's a video of the behavior: https://streamable.com/bufglv

 

 

Thanks.

 

[Moonzy]

6 minutes ago, Sufi Mindtricks said:

So it looks like my GPU may be dead...?

it looked like it, given the symptoms and that another GPU works

 

thing about folding is that it's heavy on VRAM according to @leadeater, but the GPU only monitors core temps (other than EVGA ICX and/or GDDR6X GPUs)

so your VRAM may be running at super high temps and you'd be none the wiser

[Sufi Mindtricks]

Hmm I think this model has ICX.. I would have thought it was monitoring the VRAM etc.

 

According to the product page: https://www.evga.com/products/specs/gpu.aspx?pn=122adebd-ab9f-42b5-a31a-589dd8498652

[leadeater]

Try putting it in the second x16 slot, sounds dumb but known to sometimes work. As to why it broke  Hard to know without knowing what's actually broken about it.

 

9 minutes ago, Moonzy said:

thing about folding is that it's heavy on VRAM according

The cooler on that looks decent enough, I can't really remember the teardowns of the GTX 10 series anymore though, could be one of those cards with non optimal vram cooling design.

[Moonzy]

6 minutes ago, Sufi Mindtricks said:

Hmm I think this model has ICX.. I would have thought it was monitoring the VRAM etc.

 

According to the product page: https://www.evga.com/products/specs/gpu.aspx?pn=122adebd-ab9f-42b5-a31a-589dd8498652

it is indeed capable of monitoring VRAM temps, to a certain degree

ICX uses external sensors outside of the VRAM module, thus the reading is much lower than what's on the inside, at least from my understanding

 

G6X have Tjunction, which measures temp near to the hottest point of the chip.

 

but as leadeater said, it's hard to say what died

is it, by any chance, still under warranty?

 

2 minutes ago, leadeater said:

could be one of those cards with non optimal vram cooling design.

i remember ICX was to address the VRAM overheating issue, so his GPU should have adequate VRAM cooling

[leadeater]

17 minutes ago, Sufi Mindtricks said:

My friend had a spare Zotac 1070 mini, swapped it out, and it booted w/signal.

Could you put this GPU in the first slot and your 1080 in the second slot and boot in to the OS? Might be useful to see what the OS is reporting and the EVGA OC tools/HWMonitor. Should also then been able to run benchmarks against just that GPU if it actually shows up working.

[Sufi Mindtricks]

Just now, leadeater said:

Could you put this GPU in the first slot and your 1080 in the second slot and boot in to the OS? Might be useful to see what the OS is reporting and the EVGA OC tools/HWMonitor. Should also then been able to run benchmarks against just that GPU if it actually shows up working.

Getting late now, need to sleep, but I'm going to test the card in my room mates rig to see if it behaves the same way.  When I get a free second.  Good idea.

 

1 minute ago, Moonzy said:

it is indeed capable of monitoring VRAM temps, to a certain degree

ICX uses external sensors outside of the VRAM module, thus the reading is much lower than what's on the inside, at least from my understanding

 

G6X have Tjunction, which measures temp near to the hottest point of the chip.

 

but as leadeater said, it's hard to say what died

is it, by any chance, still under warranty?

I lucked out it did this now, my warranty looks to be expiring in 75 days.  I've contacted Evga, but I'll see if I can do some more testing in another rig to see if it reproduces the same behavior.

 

Thanks

[leadeater]

8 minutes ago, Sufi Mindtricks said:

I lucked out it did this now, my warranty looks to be expiring in 75 days. 

Niceeee!! At least this is very good to hear.

[Sufi Mindtricks]

Hey, so an update.

 

EVGA sent me a "EVGA GeForce RTX 2070 XC GAMING, 08G-P4-2172-KR" as a replacement for my dead 1080.

 

So I installed it yesterday.  System was working fine...was running stress test (folding@home) overnight - temps were around 70c.

 

Next evening I came down after work to do watch some stuff, was in the middle of watching a Twitch stream, when my screen went black - no signal - and PC rebooted.  Signal did come back up, but now there were artifacts on the Windows logo, which cleared, but now my resolution was like 720p, and if I tried to bump it to 1080, it would show immediate artifacts.

 

I thought maybe I had corrupted drivers.  So I went into safe mode, ran DDU, installed the latest nvidia drivers, rebooted, same problem.  It would not install the graphics card and in device manager the GPU said "Windows has stopped this device because it has reported problems. (Code 43)"

 

So I did a fresh Windows Install.  Initially everything looked good good, resolution was good, no artifacts.  Drivers installed perfectly.  Started watching a video, and the video output went green - audio was still playing, no idea why.

 

Loaded up hwmonitor and Furmark.  Started a gpu stress test and the screen went blank and reboot - before the reset the temps was fine.

 

Now I'm back to the same issue - same artifacts on reboot/windows loading, same artifacts if I try to change resolution above 720p.  Tried HDMI cable, no go.  I'm going to try the old 1070 mini I was borrowing.

 

In the meantime, here are some pics and a video of the behavior.

Any advice?

 

https://photos.app.goo.gl/2wKDAgjJpvT4SeVz7

[Sufi Mindtricks]

So I swapped the 2070 for the 1070.

 

The only difference with power delivery being that the 1070 uses 8 pin vs the 2070 using 6+8 pin.  Running furmark with no issue, the 2070 would crash within 20-30 seconds.  No crash, no artifacts, temps around 70-73c.

 

I don't get it.  Did I receive a broken card ... Or something else is afoot?  Could it be my psu?  I may ask my room mate to test the gpu in his tower at this point (even though he's running Linux).  One other thing I noticed was the the back plate of the 2070 was very hot, dunno if that's normal, but it's not like it was running benchmarks or stress tests prior to being pulled out.

 

 

 

 

[leadeater]

5 minutes ago, Sufi Mindtricks said:

Did I receive a broken card

Potentially. Do you have access to another PSU, or even another system to try the card in. Another system would actually be preferable.

[Sufi Mindtricks]

Only other system would be my roommates. Will ask him in the morning. 

 

Ugh.  This is such a pain.  Anyone in the Edmonton area willing to test this 2070 lol

 

[Sufi Mindtricks]

So another update.  Turns out my former mining friend who lent me the 1070, happened to have a 750w, 1200w Corsair, and a 1300w EVGA G2 80+ GOLD in boxes just collecting dust.

 

So I grabbed the 1300w EVGA and swapped it in, using the same EVGA cables that came with the PSU as the Seasonic cables are not compatible with the G2 model.

 

System booted up, windows resized the resolution, installed latest nvidia drivers cleanly.  No errors, no failed install, no artifacts or forced resolution issues.

 

Ran Furmark for a few minutes, previously it wouldn't even run for 20 seconds before the system would go no signal and reboot.

Also ran 3D Mark Port Royal and Time Spy a few times.

 

Max draw on the GPU seemed to be 180w.  So not sure, but it seems like either a Seasonic PSU issue or possibly a PCI cable issue - could also be the custom Cable Mod cables I was using, not sure, I'll have to find the original Seasonic cables and re-test.

 

I'm going to leave Furmark or the Folding@home I was running on overnight and see how it looks tomorrow.  Both seem to pull about 170w at load on the gpu.

[Sufi Mindtricks]

Welp woke up with much optimism to have that crushed. PC had rebooted and found the device manager showing the same error 31 and garbage resolution.

 

Cant be the psu or cables.

I'll need to find a way to test it in another system or just call it a day and open another RMA with evga.

[Sufi Mindtricks]

Ordered a cheapish B550M mobo and Ram.

Have a spare 2700x, nvme ssd and several other boxed psu's.

 

Going to setup a test rig and see if the behavior reproduces itself.

[Sufi Mindtricks]

Back with another update.  So turns out the B550 board would not boot with my 2700x, turns out it's not compatible. ugh.  So I got a Gigabyte B450 board in it's place.

 

Personal Rig:

 

Ryzen 5800x

ROG CROSSHAIR VII HERO (WI-FI)

G.SKILL TridentZ RGB Series 16GB (2 x 8GB) 3200 F4-3200C14D-16GTZR

Hydro Series H115i 280mm

XPG GAMMIX 2TB S11 Pro 3D NAND PCIe NVMe Gen3x4

EVGA SuperNOVA 1300 G2, 80+ GOLD 1300W,

 

The Card failed to work properly in my actual rig.

I swapped the PSU thinking it was that, no change, same problems.

Ended up getting a new mobo/ram to test it on another set of hardware.

 

Test rig specs:

 

GIGABYTE B450 AORUS M (new)

Patriot Viper Steel Series DDR4 8GB (2 x 4GB) 3200MHz Memory Kit (new)

Ryzen 2700x w/Wraith Fan.  (used)

Samsung EVO nvme 512gb (used)

Corsair AX1200i 80+ Platinum (used)

 

Did a fresh Windows install.

 

Updates images/video on the test rig: 

https://photos.app.goo.gl/2wKDAgjJpvT4SeVz7

 

It's doing the EXACT same thing as it was doing in my person rig.

I have to believe the card is bad at this point.

Going to contact EVGA.

 

[leadeater]

8 hours ago, Sufi Mindtricks said:

It's doing the EXACT same thing as it was doing in my person rig.

I have to believe the card is bad at this point.

Going to contact EVGA.

Sounding like you've had some real bad luck recently.

[Sufi Mindtricks]

12 hours ago, leadeater said:

Sounding like you've had some real bad luck recently.

Best believe it lol.

All I want is a working card. 

Now I'm going to have to send this one back, likely at my own cost, wait another week or two and hope I get a working replacement.

[Sufi Mindtricks]

So, final update, I hope.

 

EVGA finally sent me another card.  Got a massive triple slot card in the mail.

 

Part Number: 08G-P4-2173-KR

Description: EVGA GeForce RTX 2070 XC ULTRA, OVERCLOCKED, 2.75 Slot Extreme Cool Dual, 65C Gaming, RGB, Metal

 

So far so good.  Been running it at max for a day now (folding@home). No crash, restarts, knock on wood.