I think recent firmware update from 4644 driver gimped Arc card.

[ador250]

Here is the Tomb Raider, Rise of the Tomb Raider and Show of the Tomb Raider benchmark result. Driver version 4577 vs 4676.

https://ibb.co/ZmjVzXR
https://ibb.co/MhFCXf0

https://ibb.co/Q7BQh3k
https://ibb.co/LdnbhXR

https://ibb.co/WWxmgpg
https://ibb.co/sjRL0PY

This gimp is not driver level, I tried to go back to the 4577 driver. Still getting less fps, so I think the gimp most probably happen to the firmware level. Can anyone investigate further this issue?

 

[abrfilho]

You can downgrade the firmware to check if it's the problem.
There's a tutorial on how to do that here.

 

[Solaris17]

101.4644 was an OPROM version change across the board. For OPROM specific changes you can download the oprom firmware from driver 101.4032.

Driver rollbacks like you describe in your reddit thread wont work, the firmware updates the driver does are new version only. It will not downgrade them.

All of that info is in the firmware matrix which it looks like you already found Once you get the firmware you need you can use the flash tool to do the downgrade.

 

[ador250]

101.4644 was an OPROM version change across the board. For OPROM specific changes you can download the oprom firmware from driver 101.4032.

Driver rollbacks like you describe in your reddit thread wont work, the firmware updates the driver does are new version only. It will not downgrade them.

All of that info is in the firmware matrix which it looks like you already found Once you get the firmware you need you can use the flash tool to do the downgrade.

I can flash the fwupdate, oprom c and d file but can't flash fwdata...Here is the error I'm getting...BTW , I'm using A750 LE

Flashing FW Data File:
 "D:\arc firmware\dg2_gfx_fwdata_448_IBC2_C8.bin"

Error: firmware data version is not compatible with the installed one (OEM version)
Image:  Fw Data Version: Major Version: 101, OEM Manufacturing Data Version: 6, Major VCN: 1
Device: Fw Data Version: Major Version: 101, OEM Manufacturing Data Version: 14, Major VCN: 1


Flashing complete!

Version 14 is the latest, I tried to go back to Version 6 @Solaris17

 

[Solaris17]

yes fwdata is locked down. Most important changes are in the files you listed.

FYI going that far back in firmware will likely re-introduce the power consumption at idle that was fixed in current oprom versions.

 

[ador250]

yes fwdata is locked down. Most important changes are in the files you listed.

Got it, means we r effed. The gimp happened in that file. Good Job, Intel. I have a weak CPU. So in Zen3/Zen4 + intel 12th/113th gen, the gap should be much bigger.

 

[Solaris17]

Got it, means we r effed. The gimp happened in that file.

how do you know that?

 

[ador250]

how do you know that?

It is, what else? We didn't get any cumulative windows update in last 2-3 week. My 4577 data is less than 3 week. The performance hit must be on the firmware.

Look at this two incident...

wow1
wow2 @Solaris17

 

[Solaris17]

It is, what else? We didn't get any cumulative windows update in last 2-3 week. My 4577 data is less than 3 week. The performance hit must be on the firmware.

you literally flashed the firmware. the fwdata does not generally house the changes. the main firmware blob along with the code and data files do.

It seems like you are taking the failure to flash a single version info file to mean it didnt work, but you havent re-ran your tests or showed any proof.

To date all major changes such as clock rate, and power consumption levels were held in one of the 3 files you were able to flash.

 

[ador250]

you literally flashed the firmware. the fwdata does not generally house the changes. the main firmware blob along with the code and data files do.

It seems like you are taking the failure to flash a single version info file to mean it didnt work, but you havent re-ran your tests or showed any proof.

To date all major changes such as clock rate, and power consumption levels were held in one of the 3 files you were able to flash.

There is also another file flashed in the 4644 beside this 4 file, hdmi firmware "dg2_pcon_fw.bin". This thing doing anything crazy? @Solaris17

 

[Solaris17]

There is also another file flashed in the 4644 beside this 4 file, hdmi firmware "dg2_pcon_fw.bin". This thing doing anything crazy? @Solaris17

no pcon is for the HDMI controller as indicated by the firmware matrix. It fixes the black screen issue people get with certain monitor combos.

 

[ador250]

no pcon is for the HDMI controller as indicated by the firmware matrix. It fixes the black screen issue people get with certain monitor combos.

I want to go way back in the 101.3277. Can I flash the oprom c file to both Code and Data? this doesn't have seperate d file. @Solaris17

I mean, I'm doing this

 

[Solaris17]

I'm not sure why you would. It will fail though. Not all files are needed for a procedure; like the guide says you can flash only 1 file if you want.

If I were you I would actually try and do some testing and post results.

As for files missing, some firmware files simply did not exist in certain driver versions, thats why the matrix exists.

 

[ador250]

I'm not sure why you would. It will fail though. Not all files are needed for a procedure; like the guide says you can flash only 1 file if you want.

If I were you I would actually try and do some testing and post results.

As for files missing, some firmware files simply did not exist in certain driver versions, thats why the matrix exists.

I think I found the issue. Adaptive sync is the problem here. I turned it off and on again, now the issue seems to solve itself. I think after the hdmi firmware update, windows + monitor seetings kinda jumbled up. @Solaris17

 

[Solaris17]

I think I found the issue. Adaptive sync is the problem here. I turned it off and on again, now the issue seems to solve itself. I think after the hdmi firmware update, windows + monitor seetings kinda jumbled up. @Solaris17

Nice glad the issue is resolved with the downgrade and nice job catching the firmware regression!

the HDMI firmware I include just to include. It actually doesn’t use the same firmware tool as the vbios from what I can see, the driver handles it using a different method. Which is why my tool can’t flash it.