Sapphire RX 590 showing up as an ASUS GPU after driver install

[Tropick]

Just bought a Sapphire NITRO+ RX 590 AMD 50th Anniversary edition to add to the collection. Been testing it the last couple of days and so far it's working fine (at least it seems fine, benchmark numbers are similar if not better than other 590s, no weird bugs) however it shows up as an ASUS Radeon RX 590 everywhere. GPU-Z, task manager, computer management, even in Adrenalin it shows as an ASUS card. I've tried clean installing Windows 10 twice now and I've noticed that it only begins showing as an ASUS card after I install the latest drivers. When using the standard driver that auto downloads when windows gets a connection it doesn't have the "ASUS" prefix but as soon as I update to 24.3.1 it immediately begins saying it's by ASUS. The weirdest thing is that even though the name shows incorrectly, GPU-Z is still picking up the correct vendor ID as Sapphire.

I don't have reason to believe the BIOS has been tampered with for mining as when I submit to TPUs database through GPU-Z I get an exact match for this vBIOS. I'm at work at the moment so I can't provide screenshots but when I get home I can upload them. Strangely enough, I've been able to find a few other instances of this behavior experienced by some other people:

Reddit

AMD Forums

AMD Forums again

Could this just be something busted in the driver? Manufacturer ID mismatch? Just a weird bug? Again, I initially thought it had been tampered with, but even after confirming the matching vBIOS and reflashing with the above link it still shows up as ASUS. Any ideas?

This is what I get in most interfaces. Name of the card is "ASUS Radeon RX 590 Series" but the Vendor ID still reports correctly as Sapphire. Same thing in every GPU application/Windows

 

[Caring1]

@W1zzard ?
Could this be an issue in GPU-z?

 

[W1zzard]

@W1zzard ?
Could this be an issue in GPU-z?

No, GPU-Z shows the name of the device in Device Manager

@Tropick: did you install the official drivers from AMD?

 

[Macro Device]

Just a weird bug?

I bet it is what it is. Must be someone in Sapphire got it wrong whilst making the BIOS.

Since your GPU doesn't misbehave and gets the job done I'd recommend focusing on that rather than on an outta whack naming situation.

 

[datnoz]

could be the old owner that did a shitty flash and that somehow only flashed the name of the device

 

[Dr. Dro]

could be the old owner that did a shitty flash and that somehow only flashed the name of the device

Unlikely. Subvendor is identified correctly as Sapphire. What's happening here is likely due to AMD's driver, I have seen this problem before. It's cosmetic in nature and shouldn't affect anything but, uninstalling the driver fully and reinstalling another version may fix

I bet it is what it is. Must be someone in Sapphire got it wrong whilst making the BIOS.

1DA2 = Sapphire. BIOS is right. This is an AMD driver bug.

 

[Imperator]

1DA2 = Sapphire. BIOS is right. This is an AMD driver bug

Hello.
I confirm that.
Many users complained lately about this.
You either let it as it is now, or you can DDU old drivers and reinstall them.
As you like.

 

[Macro Device]

This is an AMD driver bug.

Doesn't hurt actual use so whatever. A minor peculiarity.

 

[Tropick]

No, GPU-Z shows the name of the device in Device Manager

@Tropick: did you install the official drivers from AMD?

Figured out the issue, it's definitely an AMD driver problem. I had a fresh install of Adrenalin 24.3.1 which is what AMD offers you when looking up the RX590 in the current driver search tool on their site. This was the version that was giving me the naming issue. I DDU'd the driver last night and rolled back to 23.9.1 (which was the last driver version to explicitly state support for the RX590) and poof, the naming issue is gone.

 

[W1zzard]

AMD probably screwed up the INF file, try to find the device name text, should be some device ID matching patterns nearby

 

[Tropick]

AMD probably screwed up the INF file, try to find the device name text, should be some device ID matching patterns nearby

I'll dig through the files when I get home today and see if I can make some edits. As everybody is saying, this does seem to be purely cosmetic. No performance difference between 23.9.1 and 24.3.1 in Time Spy. Though if I can figure this out I'll make a quick guide on how to fix this in case anyone runs into this "problem" in the future. Thanks for the assist everyone.

 

[Tropick]

Don't mean to necro the thread but figured I should update it so I don't confuse anyone looking at this in the future. AMD fixed the naming issue in the latest 24.9.1 legacy driver release. W1zz had the right idea, it was a simple INF issue. Naming is back to normal in everything. GPU-Z, Adrenalin, task manager, Trixx, etc are all fixed.

 

[eidairaman1]

Don't mean to necro the thread but figured I should update it so I don't confuse anyone looking at this in the future. AMD fixed the naming issue in the latest 24.9.1 legacy driver release. W1zz had the right idea, it was a simple INF issue. Naming is back to normal in everything. GPU-Z, Adrenalin, task manager, Trixx, etc are all fixed.

View attachment 367439

So before this everything was working correctly then?

I presume you are using overdrive/trixx/afterburner to have the Higher than Stock clocks?

 

[Tropick]

So before this everything was working correctly then?

I presume you are using overdrive/trixx/afterburner to have the Higher than Stock clocks?

That's correct, the card was working without issue even with the previous naming error. It was boosting to and stable at stock core and memory clocks (1560MHz/2100MHz) in both synthetic tests such as Fire Strike/Time Spy/Heaven as well as real games (Horizon 5, BF2042, Control). Passed OCCT 3D Variable and VRAM testing too. Been dialing in an OC using trixx and it looks like the highest I can push it while remaining fully stable for extended periods is 1650MHz core and 2225MHz memory at a +200mV offset and +50% board power target. For quick benches it can withstand 1680MHz core and 2250MHz mem but after 15 minutes or so it'll start artifacting.

I completely disassembled the card and gave it a thorough 99.9% isopropyl cleaning, epoxied the substrate capacitors and applied liquid metal TIM to the die/heatplate. I redid the VRM/memory pads with some .75mm Kritical thermal pads according to Sapphire's specifications for the card. With a relatively aggressive fan curve the card now tops out at 72c overclocked/overvolted during the OCCT 3D static load test. The gains aren't that impressive so I usually just use the thing at stock clocks but knowing it can hit a stout 1650MHz on air makes me happy.

 

[eidairaman1]

That's correct, the card was working without issue even with the previous naming error. It was boosting to and stable at stock core and memory clocks (1560MHz/2100MHz) in both synthetic tests such as Fire Strike/Time Spy/Heaven as well as real games (Horizon 5, BF2042, Control). Passed OCCT 3D Variable and VRAM testing too. Been dialing in an OC using trixx and it looks like the highest I can push it while remaining fully stable for extended periods is 1650MHz core and 2225MHz memory at a +200mV offset and +50% board power target. For quick benches it can withstand 1680MHz core and 2250MHz mem but after 15 minutes or so it'll start artifacting.

I completely disassembled the card and gave it a thorough 99.9% isopropyl cleaning, epoxied the substrate capacitors and applied liquid metal TIM to the die/heatplate. I redid the VRM/memory pads with some .75mm Kritical thermal pads according to Sapphire's specifications for the card. With a relatively aggressive fan curve the card now tops out at 72c overclocked/overvolted during the OCCT 3D static load test. The gains aren't that impressive so I usually just use the thing at stock clocks but knowing it can hit a stout 1650MHz on air makes me happy.

Considering the 580SE by Sapphire was boost at 1430, I'd say that's a healthy OC