xfx rx480 8gb flashed to saphire rx580

[Raevenlord]

Just posted this on the front-page news. Good job on treading where no others had yet (this may be inaccurate, but that is besides the point =))

https://www.techpowerup.com/232498/rx-480-cards-can-successfully-be-flashed-to-rx-580

 

[TonybonJoby]

As for the drivers they were a little weird to install. No official release for windows 8.1 so I tried a windows 7 and 10 from the AMD site. None of them would install. So I manually installed the drivers through device manager and got the drivers working but, the newer Catalyst wouldn't install, so I ran Crimson 17.2.1 Feb22 and it seems to work fine.

 

[Shou Miko]

1.188 not 1.88 lol

so it was reading correctly even wrote wrong so that's good and nice job by the way

 

[TonybonJoby]

Thanks! Its been a long night, but I still got some more testing to do, but hopefully its all worth it.

 

[buildzoid]

This should work fine. Just for other people trying to copy this. TonybonJoby is lucky that his card the RX 480 RS uses the NCP81022 voltage controller that the RX 580 Nitro+ does. Most RX 480s run the IR 3567B and flashing those with an NCP81022 BIOS might not work at all.

As for what it did to the clocks and temperatures. Anything under 85C is fine. 1.22V is safe for daily use. The gains the RX 580 BIOS caused are basically only because the 580 is a 480 with a higher power limit and voltage set in the BIOS. I can't see this causing major issues in the long term beyond what a regular OC on an RX 480 could cause. Also the more refined low power states are IMO a nice benefit for daily usage.

 

[TheoneandonlyMrK]

You are aware of the fact that "stock clocks" are something the manufacturer decided to be stable and safe?
And that there could be a reason why you couldn't OC past 1350MHz...
RX500 is said to have improved manufacturing process (even if with the same design). You're simply doing OC beyond what your RX480 was designed for...

The Rx 480 ref was limited in oc terms by it's 50%max power slider and hard set tdp(150watts), this could be subverted initially by using the Dev BIOS but later drivers prevented this
So a new bios with higher tdp could help , though possibly only to brick or kill the card, i mean I'd rather know what Vrm is in use pre bios flash for example not just blindly try.

 

[fullinfusion]

Just posted this on the front-page news. Good job on treading where no others had yet (this may be inaccurate, but that is besides the point =))

https://www.techpowerup.com/232498/rx-480-cards-can-successfully-be-flashed-to-rx-580

And let the bricking start

 

[Carebear-Killer]

This should work fine. Just for other people trying to copy this. TonybonJoby is lucky that his card the RX 480 RS uses the NCP81022 voltage controller that the RX 580 Nitro+ does. Most RX 480s run the IR 3567B and flashing those with an NCP81022 BIOS might not work at all.

As for what it did to the clocks and temperatures. Anything under 85C is fine. 1.22V is safe for daily use. The gains the RX 580 BIOS caused are basically only because the 580 is a 480 with a higher power limit and voltage set in the BIOS. I can't see this causing major issues in the long term beyond what a regular OC on an RX 480 could cause. Also the more refined low power states are IMO a nice benefit for daily usage.

It doesn't surprise me that there could be hardware limitations. The HD6900 cards were a good example. Specific 6950 cards could be flashed to the 6970 bios and gain higher clocks and shaders. Later on, the lower spec cards actually had a hole punched in them to prevent the flash from working.

So, maybe some of the higher spec 480s from xfx and Asus can take the rx580 bios. This could also explain the bios/driver lock down AMD did a few months back with the rx480s to keep people from modding the bios.

 

[notb]

Why are you guys giving him a hard time? This is what enthusiasts forums are all about, thanks.

Is this a forum for computer enthusiasts? I though it's simply for computer users. You can call me a computer pragmatist.

The Rx 480 ref was limited in oc terms by it's 50%max power slider and hard set tdp(150watts), this could be subverted initially by using the Dev BIOS but later drivers prevented this
So a new bios with higher tdp could help , though possibly only to brick or kill the card, i mean I'd rather know what Vrm is in use pre bios flash for example not just blindly try.

Exactly. I guess I would be more optimistic if this was a low-end card limited artificially. But as RX480 was the most powerful card in the series (and way slower than the flagship R9 Fury), it really made no sense. It had to be important for stability.

Unless... they were already planning the RX580 rebranding with no true improvement - just removing clock limits. But I guess that's impossible, because AMD really cares about it's clients and so on (unlike other bad companies that just want to take our money).

 

[TheoneandonlyMrK]

Is this a forum for computer enthusiasts? I though it's simply for computer users. You can call me a computer pragmatic.



Exactly. I guess I would be more optimistic if this was a low-end card limited artificially. But as RX480 was the most powerful card in the series (and way slower than the flagship R9 Fury), it really made no sense. It had to be important for stability.

Unless... they were already planning the RX580 rebranding with no true improvement - just removing clock limits. But I guess that's impossible, because AMD really cares about it's clients and so on (unlike other bad companies that just want to take our money).

I'm an engineer, i know how stuff gets " improvements " and I see no wrong here , you don't know what you are talking about if you are suggesting that a chip maker would run a new chip on a new node at its theoretical max to millions of customers out the starting gate.
It's both the way it's done by all except some and it's clear and open at least , that's what is important to me, that and price /performance.

 

[cdawall]

This should work fine. Just for other people trying to copy this. TonybonJoby is lucky that his card the RX 480 RS uses the NCP81022 voltage controller that the RX 580 Nitro+ does. Most RX 480s run the IR 3567B and flashing those with an NCP81022 BIOS might not work at all.

As for what it did to the clocks and temperatures. Anything under 85C is fine. 1.22V is safe for daily use. The gains the RX 580 BIOS caused are basically only because the 580 is a 480 with a higher power limit and voltage set in the BIOS. I can't see this causing major issues in the long term beyond what a regular OC on an RX 480 could cause. Also the more refined low power states are IMO a nice benefit for daily usage.

I am impressed that the RS is having no issues with the additional power draw. I may have to try the higher clocked BIOS on my Nitro's after I get the rig reassembled.

 

[mtrai]

Give the guy a break...we do not know things can be done until someone tries it. I did the same and was the first back with the 290X to 390X. He deserves kudos. My Rx 480 is out for RMA so could not try it.

Anyhow I sent him a suggestion on fixing the Subvendor not showing on the 480 flashed to 580. I have my suspicions on why and what needs to be done to fix it but unfortunately I can't check until my card returns. I will leave the instruction that I believe will fix that.

Anyhow bios editing and using different bios with both 290/390 and at least with the 480 yielded nice performance gains such as the memory timings.

Anyhow I would do this but like I say my 480 is out right now.

Anyhow...GRATS first of all...if your willing I think I know how to make the GPU-Z show the subvendor correctly if you are willing to try.

You will need Polaris Bios Editor 1.4.0 or 1.4.1 ( I prefer 1.4.1)

You need to load the bios RX 580 you plan to flash your 480 with into the editor.

Now you need to edit the sub ID and Sub vendor id to XFX.

It is possible then after flashing you will need to run ati pixel patcher-bios to patch the driver bios signing verification.

If all goes right then your Sub Vendor will now appear in your "new 580"

 

[buildzoid]

I am impressed that the RS is having no issues with the additional power draw. I may have to try the higher clocked BIOS on my Nitro's after I get the rig reassembled.

I think the VRM might end up wearing out a lot quicker with the 250W power spikes. I do all my VRM ratings at 125C which is really high. If the RS VRM hit that temperature with the RX 580 BIOS running it would probs be toast. But at 80-100C it should be fine.

 

[cdawall]

I think the VRM might end up wearing out a lot quicker with the 250W power spikes. I do all my VRM ratings at 125C which is really high. If the RS VRM hit that temperature with the RX 580 BIOS running it would probs be toast. But at 80-100C it should be fine.

My youngest brother has an RS in his build, I may take a look into what the VRM's are hitting. It would be a nice free performance boost for the card.

Out of curiosity what VR was on the old Nitro's I don't have Ethernet worth a hoot right now and can't pop into your youtube.

 

[notb]

I'm an engineer, i know how stuff gets " improvements " and I see no wrong here , you don't know what you are talking about if you are suggesting that a chip maker would run a new chip on a new node at its theoretical max to millions of customers out the starting gate.
It's both the way it's done by all except some and it's clear and open at least , that's what is important to me, that and price /performance.

There was no sensible reason for AMD to downclock RX480 on purpose, if it could handle more. The gap to R9 Fury was huge, they were behind NVIDIA products. They could have used RX580 clocking back then.
So IMO this is a simple choice. They either planned this rebranding or RX480 is generally unable to sustain such load (though some managed better OC).

And I also have nothing against such rebrands, but I was told that AMD is a good company that treats clients better than competition. Sadly, I'm not an engineer so maybe I didn't understand something correctly...

 

[TonybonJoby]

Here is the performance that I am getting with Benchmark. Which is surprising given the fact that compared to true rx580 it is out-performing them as well.






I increased the Memory speed to 2100 this time for the test. At 2000 I was getting 78.8%

Here are my msi afterburner settings.

 

[mtrai]

I would suspect the RX 580 would be the same as the RX 480 in the fact that the 480 was very much memory bandwidth starved 2150 based on extensive testing was the sweet spot vs power over at overclock.net Also using the 1750 memory strap in place of the 2000 memory strap yielded better results generally...some mixed straps also work rather well.

Here y'all go for people very concerned about the VRM temp difference and TDP, TDC and Max Power Limit

The top one is Powercolor Red Devil 580 (non golden sample), The bottom one is the released PowerColor Red Devil Unlocked RX 480 Bios. The one they adjusted power up and released to owners and fully supported any issues if used...but also recommended not to run furmark or occt burn in with the bios.

Note the slight differences in these values:

[/url][/IMG]

 

[TonybonJoby]

Alright so ran Furmark on the 2100 setup, started to see some artifacting and then display driver crashed. Toned down to 2050, same results but just a little longer to get there. So most stable settings for memory is 2000 Mhz.

 

[mtrai]

I would not run Furmark or the OCCT from what Powercolor advised when they released the Unlock 480 bios to Red Devil owners.

 

[TheoneandonlyMrK]

There was no sensible reason for AMD to downclock RX480 on purpose, if it could handle more. The gap to R9 Fury was huge, they were behind NVIDIA products. They could have used RX580 clocking back then.
So IMO this is a simple choice. They either planned this rebranding or RX480 is generally unable to sustain such load (though some managed better OC).

And I also have nothing against such rebrands, but I was told that AMD is a good company that treats clients better than competition. Sadly, I'm not an engineer so maybe I didn't understand something correctly...

Finally , that's clear, if you're pushing the boundaries , failure rates go up and warranty payouts too, this can kick a company in the ass and it's happened a few times i know of ie big warranty claims based on for example laptop GPUs going faulty.
Chip maker's don't like this so they play it safe and business whise of course It makes sense as well as technically sensible , i bought two ref 480s surely i should be butt hurt , no because it is what it is ,mine do 1400 easily but no more ,well 24/7 ,I'm fine with that.

 

[gupsterg]

Thanks for share OP .

 

[Trender]

I can get 1425 MHz and 2200+ on my RX 480, but im not flashing as there isn't any 4 gb 580 bios version. Do you get crashes underclocking it as well?

 

[AntDeek]

This is really cool. I have an RX 480 RS that I'll flash up to an RX 580 just for testing.

Can anyone confirm that overclocking a base 480 would reap the same benefit as an 580 flash?

 

[TonybonJoby]

It doesn't not crash when it is underclocked.

 

[warhell200]

This is really cool. I have an RX 480 RS that I'll flash up to an RX 580 just for testing.

Can anyone confirm that overclocking a base 480 would reap the same benefit as an 580 flash?

I have a RX 480 rs best buy version hardswap, and i tried to flash same bios and it gives me mismatch subsystem id, and wont flash the 580 bios, is there any 580 bios out there that will work for my card?