Modding catalyst inf file so 290X use 390X driver

I recently read a post someone stating this can be done and also stating the cards name (ie shown in GPU-Z right at top) can be modified, I didn't bookmark post and now can't find it.

So I was wondering if someone can provide information or link the post? TIA

 

They use the same drivers now I think you mean the BIOS.

You just download the 390X bios and install it on your 290X google Radeon bios flashing guide

 

Cheers for taking time to post .

I'm not gonna flash a 390X bios as have a Vapor-X 290X with custom PCB (ie extra phase VRM). Know how to flash card and been modding my own bios for a while now.

Having added tighter RAM timings to my factory ROM clock for clock I'm beating a 390X by 3% in 3dMark FS graphics score and in Valley.

I read a post where someone flashed card with a 390X rom and did fresh install of driver to use with it, the rom didn't work for them well. SO they reflashed back to their 290X rom and state gained better benches as believe were using 300 series driver due to not changing anything but rom.

Now as I won't flash a 390X rom I thought I'd try and force use of 300 series driver.

 

I found post I had seen and couldn't find.

http://forums.guru3d.com/showpost.php?p=5109144&postcount=56

Then I cross ref'd this post by The Stilt.

And come up with this.

SO I go to C:\AMD\AMD-Catalyst-15.7.1-With-DOTNet45-Win7-64bit\Packages\Drivers\Display\W76A_INF

Then open C7187669.inf.

Go to lines:-

"%AMD67B0.4%" = ati2mtag_Hawaii, PCI\VEN_1002&DEV_67B0&REV_00
"%AMD67B0.5%" = ati2mtag_Hawaii, PCI\VEN_1002&DEV_67B0&REV_80

and change to:-

"%AMD67B0.4%" = ati2mtag_Hawaii, PCI\VEN_1002&DEV_67B0&REV_80
"%AMD67B0.5%" = ati2mtag_Hawaii, PCI\VEN_1002&DEV_67B0&REV_00

and further down I can change title as anything I want.

AMD67B0.4 = "AMD Radeon R9 200 Series"
AMD67B0.5 = "AMD Radeon (TM) R9 390 Series MODDED"

Will be testing ASAP.

When I search for ati2mtag_Hawaii or Hawaii it doesn't seem there is anything specific to 390 Series and to me it seems they use same driver.

 

@gupsterg do you think that there is a difference between those two drivers? i know that 290X driver virtually limit the card to use fonctionnalities like VSR 4K, so i'm curious if it will make the 290X perform better.

 

Don't know, at present I see no difference.

I will be comparing benches between running driver as AMD supply and with modded inf, when have time and will post back what happens .

 

this can be interesting

 

Right now all GCN 1.1 (x90(X))have the same 1800p limit in VSR.

Right now...

 

Indeed, I can't help but think some auto detection is going on. If you see the guru3d link Fox has edited a release of driver without Fury X support by editing 290 line as Fury X.

I don't know enough about driver, but would think a 290 driver path shouldn't really support Fury X. Later in that thread is a post with screenie showing successful use of mod.

 

390x is also limited to 1800p

 

290X and 390X use the same driver. No need to change the INF. If both cards have different VSR ranges, this is due to a hardware difference in the scaler.

 

They use the same drivers as they should: it's the same hardware after all been the only difference a better standard pcb tolerance in temps and a marginally better timing.

What can be easy done in "old" 290(X) with a minimal OC ( i mean MINIMAL) or buying a better a GPUs with better base from vendors who made their own modified PCBs.

In GCN 1.1 GPUs the artificial limit in drivers is 1800p right now, before it was 2160p in 290(x) using modded drivers (15.3--15.5) when series 300 where not on the market.

I guess these drivers can be inf moddified to accept "new" series 300 GCN 1.1 (390(X) GPUs and have 2160p.

It will not be a surprise at all AMD will start again to do 300 series drivers only.

The current driver VSR res limitation in GCN 1.1 after adding the res to DAL entry in registry is exactly:

3408x1917 for 16:9
3408x2130 for 16:10

(page 3)

http://forums.guru3d.com/showthread.php?t=404125

There is no such scaler difference, in fact all the scaler idea is proven BS.

Well they use the same driver now but...

...on launch they used a PR stunt 300 series only to sell more "new" 390(X) GPUs showing "old" 290(X) look bad using a less performing driver for 200 series.

When this "special" driver was unlocked to all series including "old" GCN 1.1 290(X) we see the truth: all GCN 1.1 perform exactly the same at the same timings cause they are the same hardware.

In following drivers AMD unified drivers again in the "hardware" difference dissapear...by now.

If hardware scaler fairy tale was true a simple reg value modification will be unable to add absent hardware like they did for 7000 series and increased range for GCN 1.1 (290(x)).

Artificial driver limits are very able to limit hardware capacities (lowering VSR res...) but they can't add hardware only features from thin air.

 

Hi guys,

Interesting topic. Does anyone have a backup or know where I can find an R9 370x bios ? Googled and nada so far, It was released only in China a while back.
I need the Bios for testing it on a 270X card (4GB). I'm assuming that the chips are capped from the bios and which is further enforced by a driver level lock.

Figured I should post here since it is a related topic.

 

@sammarbella, don't forget...

3408x1917 for 16:9
3408x2130 for 16:10

...is upper limit for 290/290X. For 280X and bellow we stuck at 2720 x 1530 :wanker:

 

Sorry for no update guys.

I lost 200-300 points in 3dMark FS with modded driver.

My card uses a modded ROM by me:-

- Stock GPU/RAM 1030/1325 boosted to 1100/1525.
- GPU Freq. DPMs 2 to 7 boosted per % as in OP heading Making OC bios like factory pre OC'd card/ROM (On OCN).
- 3 States VDDCI (150MHz@0.850mV 1250MHz@0.925mV 1525MHz@1.000mV).
- GPU Clock 2 matched to DPM 2 GPU Freq.
- Mem Freq. DPM 1 & 2 @ 1250MHz.
- All DPMs manually voltage fixed and tested.
- Efficiency @ idle table matched to DPM0.
- Timings improved for straps 1250MHz & 1375MHz to Stilt's AFR.
- Timings improved for straps 1500MHz & 1625MHz to Stock 1250MHz AFR.
- Standard fan mode Profile improved for clocks/voltage being run.

Above ROM would with normal driver beat a MSI Gaming 390X clocked from factory@1100/1525 by 1-3% in limited testing I did.

Basically 390/X from factory have tighter RAM timings, besides another mod stated below.

Now I plan to retest driver mod as have this added to my ROM:-

- 390/X Memory Controller Timings

This is due to The Stilt highlighting this is another difference 390/X ROM have from factory.

Link to quote.

 

There is just absolutely no point in flashing 390 bios in 290.

now that we've heard from the expert that slapping a random BIOS is bloody stupid (imagine that!), lets see how slapping a random driver things works out

hmmm....i wonder

 

I don't use 390X bios on 290X :banana: .

I have modified my stock one to have tighten RAM timings, like the expert The Stilt does for his MLU ROMS and differing ones for Mining ROMS.

The 390 MC timings I use again is the right way The Stilt has highlighted for modifying the 290/X bios to gain performance.

Perhaps you should read more the topic before stating such things :3eyes: .

Soon GPU-Z will have EDC counter included to test when we tighten RAM it effect. Again like "silicon lottery" on say how GPU clock there is that in MC on GPU .

 

did that 2.5 years ago lol

keep us posted

 

Well perhaps you should have shared it then? LOL

But I guess it's easier to pick at others sharing or testing? LOL

Perhaps it would have helped us when people like me, DDSZ & Oneb1t were marking powerplay, etc and creating mods? in the 390x leaked bios thread on Guru3D. I don't recall any post in there by you? LOL

Guess you also knew how to set manual VID per DPM 2.5 years ago? LOL

And any how FYI, stock ROM 1100/1525 vs with tightened ram vs tighten ram + 390 MC

http://www.3dmark.com/compare/fs/6485151/fs/6468814/fs/7374909

Consitently seeing improvement and will be testing OpenCL, etc.

I have results for 3 owners on OCN gaining as well, their gains differ due to varying quality MC.

Stability? well done several 7hrs plus on F@H (1100/1575).

Spoiler

And I've in the past done 8hrs RB stress mode several times.

 

sorry, we was suche busy...wow