Nitro R9-380X Crashes and artifacts. Unusable?

[Aditya Rao]

After skulking for years and hunting for ways to mod my existing machines I'm finally here. Albeit with a rather odd problem. My machine is as follows:
CPU: FX6300 (NO OC)
MB: M5A97 LE R2.0 Latest BIOS 2601
OLD GPU: Sapphire Nitro R7-360 2GB
New GPU: Sapphire R9-380X 4GB This VBIOS
Memory: FuryX 1866 8GBx2
SSD: 850 EVO
PSU for R7: Deepcool DN400 80+
PSU for R9: Antec VP650 (BRAND NEW)

This R9 card came to me as a birthday gift from my father. It replaced an existing R7 360 2GB. As soon as I log on to the machine. The screen starts glitching and has artifacts. The Windows 10 buttons and controls start disappearing.

The card runs for a maximum of 5 minutes before windows itself becomes completely unusable. I followed to test it with a completely fresh install of windows. The install goes through smoothly but the moment the graphics driver (15.11) is installed AMD settings itself vanishes.

That is I get a transparent overlay where settings should display. I only managed to run HWMonitor and GPU-Z for a short period. Temperatures were stable and the fans were running only periodically. Attempting to take a screenshot fails every single time. I have tried with drivers 15.11, 16.4.1 and 16.4.2.

The same install when replaced with the R7-360 works perfectly without any glitches. I did not get the chance to save the VBios of the R9-380X either. The card was bought in Singapore as he lives there it was brought to India in handbaggage and everything including the box came in prisitine condition. I've also tried 15.12 driver on a clean install. On one particular run I was able to install Kombustor to see if it was a clock issue. Kombustor runs perfect. Until you mouse over any of the Windows controls. At that point everything goes to the dogs again. You can see that here. Anyone got any ideas? The card is less than a week old. There is a ticket raised with Sapphire support.

For those who will ask me to test this in another machine. It does EXACTLY the same things on a brand new G3258 and H81M-S ver 1.0 with 4GBx2 FuryX memory. Changing PSU does not help either.

Before I forget these happened too:

 

[Mussels]

If the old GPU works fine, then the new one is simply faulty.

Was it brand new with warranty? Return it for replacement if so.

 

[Aditya Rao]

Yes it's under a 2year warranty but was bought in Singapore as a gift for my birthday and I live in India. Therefore the difficulty

 

[95Viper]

If you have not tried it... try switching the bios switch to the other position and testing.

Spoiler

But, I gotta feeling Mussel is correct.

 

[Mussels]

Yes it's under a 2year warranty but was bought in Singapore as a gift for my birthday and I live in India. Therefore the difficulty

Sapphire may honour the warranty locally, contact them with invoice numbers and find out?

 

[Aditya Rao]

If you have not tried it... try switching the bios switch to the other position and testing.

But, I gotta feeling Mussel is correct.

I've tried this already. Additionally was asked to try with CSM enabled/disabled. Still Nope. One is UEFI and the other is Legacy mode on my card. And yes, I think Mussels is right about a faulty card.

Sapphire may honour the warranty locally, contact them with invoice numbers and find out?

That effort is on. Who the distributor is India has become the question. But Sapphire seem to take forever on their ticket system to respond to anything. I was considering a VBIOS flash but that would certainly void the warranty. Fingers crossed. Hope they come back with a solution.

Meanwhile if anyone else has bright non-warranty breaking ideas that might work. I'd be happy to try! I have the card a few more days before it is picked up to go to Singapore again. I won't see it for atleast two more months which is when the next person will make a trip back... However Sapphire's website does say that they will accept an RMA anywhere in the SEA region provided the person is willing to ship to HongKong.

 

[F-Zero]

The card is faulty just like Mussels said nothing much you can do. If the problem persists even on a completely different Pc ( like you said ) then you know with what you are dealing with. Faulty GPU. Good luck with RMA.

 

[Aditya Rao]

The card is faulty just like Mussels said nothing much you can do. If the problem persists even on a completely different Pc ( like you said ) then you know with what you are dealing with. Faulty GPU. Good luck with RMA.

Thanks I'm pushing for it as I'm writing this. Will come to know tomorrow morning what the story is...

Couple questions:
Can you boot to safe mode without the artifacts?
Before you do a clean install of drivers, does it all work ok normally booting into win 10?
Did you clean out any old drivers possibly still existing (with the driver removal tool)?
Do you feel any heat coming off the cards fans?
If no, is the heatsink loose?
if yes, is it very hot? are all 3 fans up to speed?

I'm hoping you can boot into safe mode without issue.

Safe Mode/No drivers is a non-issue. Without any driver it runs dandy. but I haven't tried that for an extended period of time. As soon as I install drivers; it fails. I've mentioned this in the OP. Fans work ok. They are doing what they are supposed to do. Temperatures are normal. 35-55. Depending on if the room air-con is on or off. Heatsink is fitted correctly and seems to be doing it's job fine.

 

[DeathtoGnomes]

Couple questions:
Can you boot to safe mode without the artifacts?
Before you do a clean install of drivers, does it all work ok normally booting into win 10?
Did you clean out any old drivers possibly still existing (with the driver removal tool)?
Do you feel any heat coming off the cards fans?
If no, is the heatsink loose?
if yes, is it very hot? are all 3 fans up to speed?

I'm hoping you can boot into safe mode without issue.

 

[Filip Georgievski]

Unstable for sure.
Maybe factory faulty.

Or drivers not properly cleaned.
I had a similar issue with my second hand 280x. A DUAL X model would do the same stuff on my pc no matter what i did (Windows 10). But on another pc it was working fine.
Then i replaced it with a VAPOR X model and now it works everything fine.
Drivers were cleaned with DDU and multiple versions were tried.

So my suggestion is try the card on another system, maybe swap out with a new PSU and see if that works.
If nothing else fails, RMA IS A GO.
Good luck.

 

[Caring1]

Google shows so many faults with the 380's, most cause by overclocked cards, not running stock.

 

[Aditya Rao]

Unstable for sure.
Maybe factory faulty.

Or drivers not properly cleaned.
I had a similar issue with my second hand 280x. A DUAL X model would do the same stuff on my pc no matter what i did (Windows 10). But on another pc it was working fine.
Then i replaced it with a VAPOR X model and now it works everything fine.
Drivers were cleaned with DDU and multiple versions were tried.

So my suggestion is try the card on another system, maybe swap out with a new PSU and see if that works.
If nothing else fails, RMA IS A GO.
Good luck.

After each install of drivers for the R9-380X and subsequent failure. The gpu was replaced with the R7-360. Which would proceed to run like nothing ever happened. I would uninstall the driver. Boot into safe mode. DDU clean drivers. Reseat the R9-380X. And try the next driver.

Google shows so many faults with the 380's, most cause by overclocked cards, not running stock.

quite possibly.

UPDATE: FOUND the distributor in India.
Supertron Electronics Pvt. Ltd.,　+91 33 40371000 　　+91 33-40371099
Supertron House, 2, Cooper Lane, Kolkata - 700 001, India. www.supertronindia.com

Apparently they have a local guy here in my city. Now tomorrow I'll go there in the hope that they will honour an international warranty.

 

[jaggerwild]

Have you tried to down clock it? Reset it to default clocks, under volt it?RMA n wash...................

 

[P4-630]

Have you tried to down clock it?

It might work, however it should not be needed.
RMA.

 

[Aquinus]

Is there any reason why no one has suggested using DDU after changing the GPUs? I would run DDU and then try re-installing the drivers then report back. It's entirely possible that not cleaning out the drivers is causing issues, even more so if you installed a driver over a half-baked uninstall/install in the first place.

 

[Mussels]

Is there any reason why no one has suggested using DDU after changing the GPUs? I would run DDU and then try re-installing the drivers then report back. It's entirely possible that not cleaning out the drivers is causing issues, even more so if you installed a driver over a half-baked uninstall/install in the first place.

read post #10

and tbh i change video cards constantly in my spare parts builds between AMD and nvidia, and never have any issues other than CCC/NVCP crashing on system startup with a no supported card error

 

[Aditya Rao]

I would run DDU and then try re-installing the drivers then report back.

several times. Maybe you missed reading that. I should have put a potato in the end of the long OP maybe.

Have you tried to down clock it? Reset it to default clocks, under volt it?RMA n wash...................

None of it works. MSI Afterburner does not see it after a driver install reboot. Whereas it sees my R7-360 perfectly.

It might work, however it should not be needed.
RMA.

Thanks that is what I'm doing. Trying to speak to them to get it RMA'd from India.

and tbh i change video cards constantly in my spare parts builds between AMD and nvidia, and never have any issues other than CCC/NVCP crashing on system startup with a no supported card error

same here. Thanks for the quick responses. I've been in the same boat several years now. But this issue is one that made my lose my marbles. Never had this happen to me. I've had cards die. But never brand new ones.

UPDATE: Sapphire has asked me to RMA it back to the dealer in Singapore. I have asked the service ticket team why I can't RMA from India; seeing as it has to reach the Hong Kong office no matter where in SEA I RMA it from... Also, now apparently there is such a thing as an international warranty; As I gathered from one of the local dealer's service guys I spoke to yesterday. Will go try my luck this afternoon. 3pm (GMT +5:30). Let's see what happens.

 

[Aditya Rao]

SUPER UPDATE!
As it turns out the graphics card was glitchy. It was within the 15 days DOA period. Sapphire Intl ticketing team agreed to contact Supertron Electronics in India. On a parallel i got in touch with Convergent Systems the regional distributors for Sapphire in Malaysia and Singapore. Their team agreed that it was possibly a DOA card and agreed to give me a replacement as soon as someone visited their space in Singapore. The also agreed to contact Sapphire to see if they would be willing to help out. As it turns out Sapphire did help out.

Supertron India turned out to be an awesome space in terms of customer service. Their head office said that as long as the device was Sapphire and was brand new and genuinely had a defect they would gladly replace it. It took another two days of back and forth visiting their Bangalore branch. On the first day itself they agreed to replace it. The question was was it to be an open-box unit or a new one. Company ruled in the favour of a new one. Happily for me I got a BRAND NEW Sapphire R9-380X Nitro edition. With a fully valid India warranty now.

Frustrating update!
Insert new card into new machine and poof. Same old set of issues. A little digging around and switching OS's showed that this time it was not a dead card but two things in conjunction.

Microsoft's latest Windows 10 branch 14332 does not like my R9-380X. The glitches were the same and persistent across both the FX6300 config and the G3258 config. Yes even with fresh install. No, DDU cleaned driver changes had no effect.
And then AMD outs a fix in their 16.5.1 update. That fixed issues for me on Windows 10 14332 to a small extent but not completely. With the new card. Windows 7/8.1 had no issues. Windows 10 stable branch is now running absolutely fine with the FX6300.

Lessons learned:
1. you have to do the leg work. But if there is something actually wrong the companies and kind and oblige with their warranties. Gotta remember to be nice and not lose your head.
2. Email all associated people. Have a complete log as soon as you start to see something go wrong with your hardware. Finish all simple troubleshooting in advance.
3. Don't trust unstable or testing software builds (this was a painful lesson for me).

All in all- Thanks to all of you for all the help support and ideas!

 

[ASOT]

@Aditya Rao Try a clean windows install,update bios version and get the latest drivers.

 

[Aditya Rao]

@Aditya Rao Try a clean windows install,update bios version and get the latest drivers.

Might want to re-read my post there...

SUPER UPDATE!
Windows 10 stable branch is now running absolutely fine with the FX6300.

Lessons learned:
3. Don't trust unstable or testing software builds (this was a painful lesson for me).

All in all- Thanks to all of you for all the help support and ideas!

It's an FYI cum thank you post. But thanks for the response.