Unusable Sapphire R9 380 Nitro

T

tchalikias

Weaksauce
Joined
Jul 12, 2016
Messages
103
I've never had so many problems with a graphics card in my 15 years of building and maintaining PCs.

Almost a year ago, I recommended this card to my cousin, whose 6850 bit the dust after a couple of years.

The R9 380 will run a few select games perfectly, while on others it will black screen a couple of minutes in the game. The monitor loses signal (tried both HDMI and DVI), the sound continues and after a while the PC hard locks.

I've tried literally everything, including dismantling the entire PC and rebuilding it from scratch, days of stability and stress testing, to no avail.

The system is an i5 2500k on a Gigabyte Z68 mobo, with a beefy Corsair 650W PSU. It's rock solid - tested for 24 hours via Prime, Memtest etc. I've even stress tested the card via FurMark - no issues.

Games like Total War: Rome 2, Endless Space, Endless Legend, the latest Civ games, Galactic Civizilizations 3 etc. are unplayable due to this issue. Even older titles, like Mass Effect 2 can't even launch (crash after a few seconds of loading a game). Other titles work fine for hours on end, like Oblivion.

The card works flawlessly on desktop usage, both in Win 7 and Win 10. The most 'stable' (if I can call it that) driver appears to be 15.11.1 beta. With it, some of the aforementioned games are playable for a few minutes, at least.

I've tried the various fixes posted online, including disabling low power mode completely via the registry.

Apparently this is a widespread issue, but AMD has yet to release a definite fix. It's been a year since my cousin got this card, he can't play his favorite games on it and I've spend weeks troubleshooting it.

Should I just tell him to sell it and go get an Nvidia card? Or is there a fix I might be missing?
 
I know Z68 boards had an issue with some cards that were PCI-E 3.0 only, or that featured EFI-compatible BIOSes. May be somewhere to start looking.
 
This particular board has two types of BIOSes, a legacy one and a (beta) UEFI one. I tried both, made no difference... I think this problem is motherboard-agnostic, from what I can find online. It's something to do with AMD's crappy driver and power saving modes.
 
My wife just recently has started having issues with her 7770 (replacing soon anyway) that are just like you describe. She only plays Wow but it will black screen with sound still going, usually in particular situations. I felt it was more of a DX issue but yesterday while upgrading the AMD drivers it would go black, normal during update except, then show a GRAPHICS DRIVER HAS RECOVERED, which has never happened during update ever with me or any of my GPUs. Not sure if it is a sign that the GPU is starting to falter (thankgod for XFX lifetime warranty) or some driver issue with her PC. Gonna do a clean install on hers soon.

Add: I used this 7770 before when I did play WoW and never had a single issue even during CF.
 
I think it might be a win10 issue, or how the drivers are on w10. I have a nvidia card, and occasionally It will flake out and give me that "graphics driver has recovered" message.
 
have you tried bumping up the vram voltage a bit? have you checked the temp.?
 
Have you tried those games without an overclock on your CPU and/or ram?

I'm asking, because sometimes overclocking can put your mobo out of spec on other buses. Your 380 may not like that.
 
Temps are all ok, everything runs at stock and the previous 6850 card worked perfectly. As I said, the pc is rock solid. I'm waiting for a driver fix, although I'm guessing even AMD don't know what the exact cause of the issue is...
 
AMD cards unfortunately sometimes are sold with incorrect voltage values. My current card for example would red screen when it was brand new anytime it was used in 3d. I had it figured out the first day by just playing around with the voltage values in afterburner. Once i figured that out, i found a different bios with the much higher clocks and raised voltage values and its worked perfect ever sense. Its tougher to diagnose for some cards since the faulty card will appear to work fine in some games. The reason all this made sense to me very quickly is i noticed so many different amd cards had different voltage values. (so i knew it was safe to at least use a voltage that was already being used stock for a different card)

In all that time you NEVER thought about tweaking the gpu or vram voltage? you can safely try it with software like trix or afterburner without bios flashing. AMD cards are better suited for more advanced users who dont mind tweaking brand new cards....people always assume its a drivers thing but most times it only appears that way. Oh and one other thing i forgot, it could be fixed by just under clocking the gpu boost. Drop it 100 mhz and see what happens...if that fixes it go up 50 and eventually you figure out what clock it should have been set at. The driver stopped responding can be from an unstable clocks on a card.
 
Last edited:
I had a great system with a delidded i7-4770k. I had odd game crashes (GTX970) which were hard to replicate, but occurred in only certain games. I got black screens, occasional "display driver has stopped responding", etc. Everything pointed to the gpu. Ends up my cpu was damaged. (It may have had something to do with my hammer-and-block delidding actions. :) ) Point being, it was the cpu, not the gpu.

Since there are not widespread reports similar to yours, the obvious fault is somewhere in your box: hardware or software. If you can drop a spare gpu into that machine and test it, that would help narrow it down...

Good luck. These can be tough to pin down.
 
primetime said:
AMD cards unfortunately sometimes are sold with incorrect voltage values. My current card for example would red screen when it was brand new anytime it was used in 3d. I had it figured out the first day by just playing around with the voltage values in afterburner. Once i figured that out, i found a different bios with the much higher clocks and raised voltage values and its worked perfect ever sense. Its tougher to diagnose for some cards since the faulty card will appear to work fine in some games. The reason all this made sense to me very quickly is i noticed so many different amd cards had different voltage values. (so i knew it was safe to at least use a voltage that was already being used stock for a different card)

In all that time you NEVER thought about tweaking the gpu or vram voltage? you can safely try it with software like trix or afterburner without bios flashing. AMD cards are better suited for more advanced users who dont mind tweaking brand new cards....people always assume its a drivers thing but most times it only appears that way. Oh and one other thing i forgot, it could be fixed by just under clocking the gpu boost. Drop it 100 mhz and see what happens...if that fixes it go up 50 and eventually you figure out what clock it should have been set at. The driver stopped responding can be from an unstable clocks on a card.
Click to expand...


I'm sorry, but no. That is an absurd, absurd point of view. The card should work out of the box. PERIOD. If it doesn't, maybe you try a tweak or two, maybe you even reformat, but I would have RMA'd that thing before doing half of what OP had done. I've owned plenty of AMD cards and not once had to tweak voltages to get it stable at stock speeds.
 
Activate: AMD said:
I'm sorry, but no. That is an absurd, absurd point of view. The card should work out of the box. PERIOD. If it doesn't, maybe you try a tweak or two, maybe you even reformat, but I would have RMA'd that thing before doing half of what OP had done. I've owned plenty of AMD cards and not once had to tweak voltages to get it stable at stock speeds.
Click to expand...

Trying a different bios is absurd? lol ok I never claimed it shouldn't work out of the box. lol AMD didnt drop to 20% market share because all their cards ship perfect lol Of course my experience and someones else will always be different but if i had your pov my current card would have been sent back;)
 
Last edited:
primetime said:
Trying a different bios is absurd? lol ok I never claimed it shouldn't work out of the box. lol AMD didnt drop to 20% market share because all their cards ship perfect lol Of course my experience and someones else will always be different but if i had your pov my current card would have been sent back;)
Click to expand...

Yea, IMO the recommendation that somebody flash their GPU bios as a basic troubleshooting step is absurd.... or rather, passing judgement on somebody because they didn't do so is completely unrealistic and unfair. The blame here lies with the vendor, not the customer, for supplying an out of spec GPU. This is basic quality assurance. Modifying voltages or bios are steps taken for people looking to maximize performance. I know BIOS updates are a regular part of life for motherboards, but I don't think the same for GPU's... and clearly the manufacturers think the same as GPU bios files are not readily available from them.
 
I had a 7970 prior to my 980TI and everything was working perfectly. Then I bought a MSI 980ti gaming 6g OC edition and installed it. Booted up and nothing. Only beeps at 30 second intervals and if I let go long enough, like 1 minut or two it would load into windows but was unable to install drivers. I was clearly dissapointed thinking my new card was toast. Before rmaing though I decided to check and see if there was a new mother board bois for my Asus P8Z68-V PRO/GEN3 . There was. Downloaded, flashed and boom, problem solved. :)
 
Activate: AMD said:
Yea, IMO the recommendation that somebody flash their GPU bios as a basic troubleshooting step is absurd.... or rather, passing judgement on somebody because they didn't do so is completely unrealistic and unfair. The blame here lies with the vendor, not the customer, for supplying an out of spec GPU. This is basic quality assurance. Modifying voltages or bios are steps taken for people looking to maximize performance. I know BIOS updates are a regular part of life for motherboards, but I don't think the same for GPU's... and clearly the manufacturers think the same as GPU bios files are not readily available from them.
Click to expand...
If I followed your advise, I'd have a non working card. How bout we just don't agree? That's all I can say about that
 
Unforunately there is no newer version vBios for this (reference design, custom cooler) card. I could perhaps try flashing another vendor's bios but I'm afraid it may make things worse (most of the newer revision BIOSes are for oc'd cards). Motherboard, as I said, has the latest bios. I will try playing with Trixx and see if upping voltages solves this, although, as I said, Furmark stress testing seems to run fine for hours on end - I can't see how higher voltage will help with less demanding loads, such as turn based strategy games.
 
tchalikias said:
Unforunately there is no newer version vBios for this (reference design, custom cooler) card. I could perhaps try flashing another vendor's bios but I'm afraid it may make things worse (most of the newer revision BIOSes are for oc'd cards). Motherboard, as I said, has the latest bios. I will try playing with Trixx and see if upping voltages solves this, although, as I said, Furmark stress testing seems to run fine for hours on end - I can't see how higher voltage will help with less demanding loads, such as turn based strategy games.
Click to expand...
what exact bios are you currently using? It is best to try making small adjustments via software like afterburner/ trix. And yea you right about flashing incomparable bios as thats a terrible idea. Probably the easiest thing is under clock its slightly and see what effect it has. (thats where i always start) You could also do this kind of testing with the amd drivers alone.

Ill give you an example of my card. My card was Brand new card from RMA from xfx (took less than a week) so i was super happy to get it so fast. Sure enough just about anything 3d would give my immediate red screen. I thought fuck, i dont want to send another card back. So i lowered the gpu speed 50mhz and all the problems went away. I compared my gpu voltage to other cards and noticed mine was only 1.115 and another xfx card (same model but was called black edition) used 1.150. I rased the voltage with afterburner and what do you know? it worked perfect...not only that i could overclock it to 1150mhz as well compared to red screening at only 950mhz. I figured fuck it if this works ill just flash the black edition bios to mine so it says that way....been perfect ever since. People are mistaken when they call every problem driver issues when in reality its a hardware issue, and in my case i just need different voltages. (still a hardware fault more or less cause drivers cant fix it)

You cant wait for a driver to fix faulty hardware is my point....rma it if your not comfortable with certain trouble shooting
 
Last edited:
just an example of what asus uses on their card. I would use it as a comparison to what your card currently uses
1451194273SgpnPYnELm_4_3.png

What voltage is your card using in 3d?
 
Doesn't Sapphire come with Trixx software, couldn't you at least test increasing the voltage to see if that cures it?
 
Yup that's the next thing I'll try. As it's not my PC but my cousin's, I don't have it here right now. Although as I said, some games run fine, every benchmark/stress test utility I've tried runs fine, it's certain games (a lot of them) that seem unable to run with this card.
 
tchalikias said:
Yup that's the next thing I'll try. As it's not my PC but my cousin's, I don't have it here right now. Although as I said, some games run fine, every benchmark/stress test utility I've tried runs fine, it's certain games (a lot of them) that seem unable to run with this card.
Click to expand...
What your describing isn't that far fetched. You notice how the newer crimson drivers have separate over drive settings per game now? Some games might be right at the limit of stability while others are well over. It could take a lot of trial and error but you can still get it sorted out. In the end if it troubles you to much just rma it. Myself, i just hate being without a card. I see people ALL the time blame the drivers when i highly suspect the hardware is faulty. (its easy to confuse the two)
 
Further things I've tried:

1) Replaced stock thermal paste with Noctua NT-H1 (no noticeable drop in temps)

2) Flashed the newest bios I could find (MSI), which fixed some small image corruption at the Gigabyte logo when booting the PC in UEFI mode

3) Another fresh Windows 10 install, with the latest hotfix driver (16.7.3)

4) Using MSI Afterburner, disabled ULPM, forced constant voltage

5) I can't set the GPU voltage manually for some reason, I tried however increasing the power envelope through AMD's control panel

The same crashing, as before.

FurMark tops at 75 degrees C after two hours. No graphical errors, no crashes.

As soon as I run a game like Endless Legend, even before I get to the main menu, my monitor turns green and then loses signal. The computer hard locks a few seconds later, with the sound looping in the background.

Then I started testing earlier drivers.

From my experience, the earlier the driver version, the more I can play a game without crashing (for example, with Catalyst 15.7 Endless Legend loads fine, I can load a save game and play for a minute or two before crashing).

This is obviously a driver issue, as well as a pretty good reason to keep away from AMD cards.

An RMA is not an option, since according to AMD forums it's not a hardware fault. One year later and AMD says "we're still investigating the issue, thank you for your patience".


So my cousin's only option right now seems to be to sell this card to some unlucky fellow and buy a gtx 960 or similar.


If anyone has any other ideas, I'm all ears.
 
The card needs to be RMA'd thru sapphire. IT IS NOT THE DRIVERS! If it were it would effect ALL or Most 380 cards which it isn't. Should have been RMA'd the first week TBH since its not yours and you cant be there all the time to work on it. By all means keep blaming the drivers. RMA is an option for goodness sake! Its a hardware issue when the FUCKING CARD KEEPS CRASHING! Just contact sapphire and send it off lol It ain't that hard. I absolutely know if it was an XFX card you would ALREADY have a brand new replacement by now. (because their rep helps here)

Are you a teenager or something? For fucks sake dont go selling someone a used faulty card...Thats down right Deceitful and stupid especially when the card still is covered under warranty.
 
Last edited:
Whoah, hold your horses. No, I'm not a teenager. No, I would never sell something knowing it is broken - however, this is not my card or PC, I'm simply trying to troubleshoot the issue. Yes, RMA is always an option, and probably the last thing I'm going to try. Does anyone know Sapphire's RMA turnaround time for Europe?

Thing is though, AMD reps themselves don't recommend RMAing the cards, insisting that this is a known issue they are working on, and recommending older Catalyst drivers to hold users over until they roll out a software fix. True enough, with Catalyst drivers a few games that were unplayable with Crimson (like Rome 2), play fine. As far as I know, not a single user with this issue who has RMA'd their card had this 'black screen/signal loss' problem solved (new card behaved like the old one).

Some combination of hardware + Windows 10 + R9 380/X seems to be problematic.

R9 380 black screen issues - AMD's persistent, unsolved bug - PC Invasion
 
Probably looking at 3 to 4 weeks under normal circumstances. As far as driver bugs we almost always get multiple people here describing the issues like we did with the power saving issue the 390/380's had a while back where the cards would drop clock speed and loose performance. When AMD screws up on the drivers and people are waiting for a fix, someone somewhere almost always provides a work around till AMD finds the problem. (clock blocker ClockBlocker (profiled AMD power-management control) - Guru3D.com Forums) If this was as wide spread as you think there would at least be one thread here. Videocards - AMD Radeon Catalyst Drivers Section - Guru3D.com Forums
 
Also amd didn't make or sell your card. They only make reference cards. They really have no idea how to support aib cards cause they don't make, sell, or support them. Saphire made and supports the card. (which is faulty) So have it replaced Before the warranty runs out. That way you can sell it new/factory refurbished if so inclined. You been asking the wrong reps for help this entire time lol!
 
The problem is convincing Sapphire that there is a hardware issue with the card. For all I know, they run Furmark and/or 3dMark and a couple of AAA titles to verify that a card is ok on a hardware level. As I said, all graphics benchmarks I've tried, as well as a lot of games work normally. And a ton of other games don't seem to work at all. We 'll see.
 
tchalikias said:
The problem is convincing Sapphire that there is a hardware issue with the card. For all I know, they run Furmark and/or 3dMark and a couple of AAA titles to verify that a card is ok on a hardware level. As I said, all graphics benchmarks I've tried, as well as a lot of games work normally. And a ton of other games don't seem to work at all. We 'll see.
Click to expand...

Have you tried a different, quality built PSU? If not, please humor me and give it a try...You can also try the ClockBusters app that PT linked you to...I was going to suggest that myself as well.
 
my powercolor does the same thing.. (290x).. I increased the voltage +25 and so far it has fixed my issue (I'm using the MSI Afterburner)
 
ccityinstaller said:
Have you tried a different, quality built PSU? If not, please humor me and give it a try...You can also try the ClockBusters app that PT linked you to...I was going to suggest that myself as well.
Click to expand...

I've tried two Corsair PSUs (650 & 600w).

I would really like to overvolt the card, however both Sapphire Strix and MSI Afterburner won't let me (the slider is locked out). I've tried unlocking voltage control, however it didn't make any difference. How do you increase voltage on this card?
 
tchalikias said:
I've tried two Corsair PSUs (650 & 600w).

I would really like to overvolt the card, however both Sapphire Strix and MSI Afterburner won't let me (the slider is locked out). I've tried unlocking voltage control, however it didn't make any difference. How do you increase voltage on this card?
Click to expand...
My question for you...Did you try slightly under clocking the gpu like maybe to normal reference speeds? Try setting it to 980 or 970mhz vs 1010mhz under global. Of course it doesn't raise your voltage but with the lower clocks it wont need the increased voltage to stabilize it. This is pretty easy stuff to try to be honest. Im betting just this simple suggestion fixes it in all the games it was black screening in with only 3-5fps loss in frame rate. If your unhappy with that then its best just to rma it before warranty runs out.

ALSO....
TechPowerUp

Flash it to one of the other bios with the lower clocks form here. This would be a good option if under clocking it solved all the issues and you wouldn't have to keep changing the clocks after changing drivers and what not.
 
Last edited:
When i play any game using a game specific profile (on radeon crimson) the computer freezes and i have to reboot, same thing if i use a general profile for all games.

Disable all profiles on Crimson -> Games, and on General Settings remove "Power efficiency" and "FrameRate Target Control".
 
It's been about a week since I opened a support ticket with Sapphire. Their responses have been timely, however they keep telling me to try this and that in the hopes that the issue will disappear (most of the things they 've suggested I've already tried). They even sent me a 'new' bios to flash (which turned out to be the exact same bios version, per GPU-Z). Sigh. I hate this card with a passion. Why couldn't it just die completely so that I could send it back ASAP.

In the 20+ years I've been building PCs, this is the second time a graphics card has been playing games with me (see what I did there?) - about 8 years ago, there was an Asus GTX295 that overheated under very specific conditions, thankfully the problem became apparent almost immediately, so back to the store it went.
 
I'd say if you've already tried it don't bother trying again, just tell them the results as if you had just tried after they suggested it (so they can check it off and move on). Hopefully you'll have an RMA or a fix by the end of it.

Did you look at the SHA hash of the two bios'? They may have modified the bios but didn't change the version number, though it'd be weird for them to do that and not tell you it was a modified bios.
 
tchalikias said:
It's been about a week since I opened a support ticket with Sapphire. Their responses have been timely, however they keep telling me to try this and that in the hopes that the issue will disappear (most of the things they 've suggested I've already tried). They even sent me a 'new' bios to flash (which turned out to be the exact same bios version, per GPU-Z). Sigh. I hate this card with a passion. Why couldn't it just die completely so that I could send it back ASAP.

In the 20+ years I've been building PCs, this is the second time a graphics card has been playing games with me (see what I did there?) - about 8 years ago, there was an Asus GTX295 that overheated under very specific conditions, thankfully the problem became apparent almost immediately, so back to the store it went.
Click to expand...
This type of shit is why a lot of us in the states buy XFX cards.....Cause we can work directly with the rep here! (And he rather send out brand new cards then risk bad press) Course it doesn't help you guys in Europe. I have heard mixed results of sapphire but the bad ones are the folks shipping cards to the china offices. Some people say there great, others say the opposite
 
The plot thickens...

"Dear Customer :
Of course , Amazon told you the card was no warranty due to they purchased the card from us without wanting manufacturer's warranty.
They got the goods as discount and then Amazon said they will directly take care of the customer.
As you can see below that this SN carried no manufacturer's warranty protect.
And our RMA center will not accpet this back to go RMA process , you need help to contact Amazon and follow their warranty policy.
If they said the card is only 30 days , then the card also cannot RMA via them.
Please deal with them again , thanks."

They also sent me the following screenshot:

gFHFDos.png


FYI, the card was purchased on amazon.fr on the 28th of August, 2015. It was sold directly via Amazon, not a third party reseller.

I've contacted them and I'm awating a reply. What a fustercluck.
 
tchalikias said:
The plot thickens...

"Dear Customer :
Of course , Amazon told you the card was no warranty due to they purchased the card from us without wanting manufacturer's warranty.
They got the goods as discount and then Amazon said they will directly take care of the customer.
As you can see below that this SN carried no manufacturer's warranty protect.
And our RMA center will not accpet this back to go RMA process , you need help to contact Amazon and follow their warranty policy.
If they said the card is only 30 days , then the card also cannot RMA via them.
Please deal with them again , thanks."

They also sent me the following screenshot:

gFHFDos.png


FYI, the card was purchased on amazon.fr on the 28th of August, 2015. It was sold directly via Amazon, not a third party reseller.

I've contacted them and I'm awating a reply. What a fustercluck.
Click to expand...
it was a normal brand new purchase right?
 
You must log in or register to reply here.
Back
Top