GPU Identity Crisis (Is that an RX 550? No it's an RX 460. IT'S AN RX 560D?! But it's a POLARIS11)

[Yosuru]

Got this graphics card from a bargain bundle.

For $20 I got:
- AOC 16" monitor, USB Powered, VGA Only. (No Issues)

- 2.5" 500GB HDD (No Issues 100% Health)

- "RX 550 2GB" (Untested)

For the price, I couldn't really complain if I got a defective GPU.

I plugged it into my PC and BAM!, no POST. I quickly put my GTX 1060 6GB back and plugged the "RX 550 2GB" into a riser card and it booted into Windows just fine.

I opened Device Manager and there I see it "Microsoft Basic Display Adapter", then I opened GPUz and there I see it. (pardon me for not taking screenshots around this time)

GPUz sees it as an RX 560 4GB card, wow that's sus. That's very not what the seller said it was.

 

Oh before I forget, I should show what the GPU looks like. It has no label anywhere, the appearance of which does not associate it with any of the known GPU vendors except... POWERCOLOR but not quite.

*The GPU was pretty much never been opened when I got it, the sticker on the screw was intact.

Here's what a Powercolor RX 550 looks like:

It's similar but not really. Does anyone recognize the logo at the fan hub?

 

So anyways I thought it was just an incorrectly flashed VBIOS in the card, GPUz shows the card as an RX 560 4GB and the core frequency default is 1076MHz and Memory Clock default is 1500MHz.

Although the current reading both reports 0MHz.  (pardon me for still not taking screenshots around this time)

It also bugged me that the shader unit count was 896 instead of 1024. A bit of searching led me to believe what I got on hand is an RX 560D. A chinese variant of RX560 with only 896 Shaders.

I used GPUz to save its VBIOS as is when I got it and the filename was defaulted to Baffin.rom.

1 [VBIOS IT CAME WITH].rom

 

From a bit of research I discovered that Baffin is of POLARIS 11? But how can that be for an RX500 series graphics card?

So I tried getting my hands on a VBIOS but I had no way of confirming its memory type.

So I decided to open it, and there it is Micron memory (GYZY 8NA47 D9TCB). Tried searching to no avail which was odd.

From here I also took the opportunity to verify the GPU from its DIE number (216-0896088) but my search led me knowhere except for one thread.

 

https://forums.macrumors.com/threads/2011-imac-graphics-card-upgrade.1596614/page-392

on post number "9793"

KennyW stated that "216-0896088 is RX460 (likely rebranded RX560D)" and a lot more insight as to how confusing the RX 460s and RX 560s was during the time.

 

I was already very confused at this point so I just tried flashing an RX 560D VBIOS with Micron memory support.

ATIFLASH detected the GPU as an "ATI RADEON POLARIS11" which just adds to the confusion because the detected VBIOS (the VBIOS it came with) was of an RX 560 POLARIS21 4GB as seen in the pic.

*How reliable is AtiFlash in identifying GPUs? Because GPUz relies on the card's VBIOS in identifying cards right?
 

 

Pardon me for this was the only screenshot I could find.

 

I did not go through with that flash just yet though.

 

The very first FLASH I did I tried force flashing an RX 460 4GB into it.

2 [VBIOS I first flashed].rom

GPUz saw the memory value of 4GB just fine but the core and memory were still 0MHz.

I tried installing an AMD driver and it detected the "RX 460 4GB" and it installed. But the GPUz still shows 0MHz so something must be wrong.

Device Manager reports Code 43 on the card and so I immediately suspected that the VBIOS I flashed wasn't just quite it.

 

So I did try to force flash it to an RX 560D Sapphire but then it failed and bricked the card.

Sapphire.RX560D.4096.170929.rom <It was this but I deleted the one I edited with Polaris Bios Editor (just changed the vendorIDs and deviceIDs so I could flash it)

 

After almost an entire day of attempting to unbrick it (Error reading ROM & Adapter Info Not Found), there it was IT JUST FLASHED AGAIN.

 

^Trying to undo the damage, the FLASH suddenly working actually failed again but at least it remained flashable so I flashed the last known flashable ROM into it.

 

I was able to reflash it back to the RX 460 VBIOS I tried just before that and I have not touched it ever since.

3 [VBIOS It has right now].rom

 

Can anyone help identifying this card? I really want to be careful this time into the VBIOSes I try to flash onto this card.

 

GPU: AMD 216-0896088

Memory Chips: GYZY 8NA47 D9TCB

 

Forgive me if the post is a bit confusing, ask me anything if it helps.

[191x7]

That is a Yeston RX 560D 4GB

 

https://www.gpuzoo.com/GPU-Yeston/RX_560D_4G_D5_GA.html

[YoungBlade]

To perhaps alleviate some of the confusion, the RX 460 and RX 560 are the same graphics card. They use the same die and architecture. The only difference is that a true RX 560 has the entire die unlocked, whereas the RX 460 and some RX 560s have some of the compute units disabled.

 

The reason I say "true RX 560" is that AMD decided that it would be a good idea to release cut down versions of the RX 560, which have the exact same CU count as the RX 460, but are called "RX 560" because marketing. There was outrage in the tech community about this several years ago, and your situation here is yet another example of why companies should never release products with different specs under the same name.

 

Here's a HUB video about the situation if you want to know more:

[Yosuru]

4 minutes ago, 191x7 said:

That is a Yeston RX 560D 4GB

 

https://www.gpuzoo.com/GPU-Yeston/RX_560D_4G_D5_GA.html

Oh my you're right, the logo.

 

3 minutes ago, YoungBlade said:

To perhaps alleviate some of the confusion, the RX 460 and RX 560 are the same graphics card. They use the same die and architecture. The only difference is that a true RX 560 has the entire die unlocked, whereas the RX 460 and some RX 560s have some of the compute units disabled.

 

The reason I say "true RX 560" is that AMD decided that it would be a good idea to release cut down versions of the RX 560, which have the exact same CU count as the RX 460, but are called "RX 560" because marketing. There was outrage in the tech community about this several years ago, and your situation here is yet another example of why companies should never release products with different specs under the same name.

 

Here's a HUB video about the situation if you want to know more:

Thanks, I'll check this out.

[Yosuru]

Seems like there are also variants within YESTON for RX 560s

Their website has no reference to any RX 560D that remotely looks like what I got.

 

Though Amazon listings and other commerce sites have references to multiple YESTON RX 560D 4GB.

 

You mentioned the "Yeston RX560D-4G D5 GA (GAEA)"

14 minutes ago, 191x7 said:

That is a Yeston RX 560D 4GB

 

https://www.gpuzoo.com/GPU-Yeston/RX_560D_4G_D5_GA.html

Which is this

 

And after a bit of research I found this:

 

and Yeston RX560D-4G D5 LP

 

So now we got:

- Yeston RX560D-4G D5 GA

- Yeston RX560D-4G D5 TA

- Yeston RX560D-4G D5 LP

 

I will now try to look the the TA's VBIOS

[YoungBlade]

7 minutes ago, Yosuru said:

I will now try to look the the TA's VBIOS

Techpowerup potentially has it, although as mentioned at the top, it is an unverified BIOS.

 

https://www.techpowerup.com/vgabios/237026/237026

[Yosuru]

[UPDATE]

 

So I flashed this

1 hour ago, YoungBlade said:

Techpowerup potentially has it, although as mentioned at the top, it is an unverified BIOS.

 

https://www.techpowerup.com/vgabios/237026/237026

After restarting, the GPU is now recognized by GPUz as "RX550/550". Same situation in Device Manager.

AIDA64 thinks it's an RX460 though? also clicking on "lookup" in GPUz also shows RX460 (https://www.techpowerup.com/gpu-specs/radeon-rx-460-1024sp.c2918)

 

Also realized that this vBIOS from TechPowerup is similar to the vBIOS this GPU first had when it arrived. hmmmmmm.

 

Curiously, based from the marketing images of a listing from Aliexpress

 

The BIOS Version does not match the one I currently have.

 

Clockspeeds still at 0MHz, and Bus Width is 32bit. But that could be because it's still on a riser card.

 

I want to believe there is another vBIOS out there.

 

For now, I will now try putting it on the main PCI-E x16 slot and see if it at least spits out an image this time.

[Hairless Monkey Boy]

2 minutes ago, Yosuru said:

Clockspeeds still at 0MHz

I've seen this behavior with AMD GPUs multiple times. Once with an RX 480, and once with the Vega 8 in a 5600G. It seems normal for this to happen, but I don't know why it does.

[Yosuru]

Before I rebooted, I tried to do a little searching.

 

It got me vBIOS "015.050.002.001.000000"

 

 

[Yosuru]

I think I'm close to figuring out the problem.

 

All of my previous flashes I've assumed went successfully. amdvbflash always simply disappears after flashing, the report only appears for a fraction of a second.

 

This time I tried recording it, i've figured out that it actually fails.

 

And the vBIOS won't be readable depending on how much was programmed into the chip.

 

This first flash I tried with the 015.050.002.001.000000 vBIOS

 

After it failed, I couldn't get GPUz to read its vBIOS and ATIFlash GUI version would just say "Error reading from ROM"

Only the CMD one works, but only after a restart.

 

 

I kept trying, each attempt different value of programmed bytes.

 

The command I use is

 

"amdvbflash.exe -f -p 0 (vBIOSFile).rom"

 

Is the chip itself the issue or is there any other parameter I can use with amdvbflash?

 

For now I will try to download a version of amdvbflash that's a bit older.

[Yosuru]

Hello, I figured that running CMD as admin prevents the amdvbflash from opening a separate terminal window and closing immediately upon completing its task.

 

Now i've got these:

 

At first I tried flashing and got an error, found on a thread somewhere that disabling the card helps a bit in flashing.

So at 2nd time, I had the card disabled and it did get me further. But it is still a fail.

 

I don't understand why it won't complete a flash but here's an "amdvbflash.exe -i" the test still says "pass"

 

Anyone knows what's up?

[Yosuru]

Found a haven of Yeston vBIOSes

[Yosuru]

Oh we are getting close.

 

Anyone knows of a DOS compatible AMDVBFLASH / ATIFLASH / AMDFLASH that has "-f" or bypass SSID mismatch and knows RX560 exists?

[Yosuru]

Found a vBIOS from a random thread of unlocking RX560D shaders from 896 to 1024.

A random person uploaded their Yeston RX560D 4GB GDDR5 Micron memory to be unlocked.

I tried that vBIOS and it causes system hang whenever the driver is installed without needing to use ATIKMDAG patcher.

Something else must be wrong with this card but at least we now know that it is in fact a "Yeston RX560D-4G D5 TA".

 

Thanks to 191x7 for the very swift identification of the card. I don't know how long it would take me to identify this card without his help.

8 hours ago, 191x7 said:

That is a Yeston RX 560D 4GB

 

https://www.gpuzoo.com/GPU-Yeston/RX_560D_4G_D5_GA.html

 

With that said, this thread is done. Thanks ya'll.

[191x7]

4 hours ago, Yosuru said:

Found a vBIOS from a random thread of unlocking RX560D shaders from 896 to 1024.

A random person uploaded their Yeston RX560D 4GB GDDR5 Micron memory to be unlocked.

I tried that vBIOS and it causes system hang whenever the driver is installed without needing to use ATIKMDAG patcher.

Something else must be wrong with this card but at least we now know that it is in fact a "Yeston RX560D-4G D5 TA".

 

Thanks to 191x7 for the very swift identification of the card. I don't know how long it would take me to identify this card without his help.

 

With that said, this thread is done. Thanks ya'll.

You can try baking it, nothing to loose.