• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Flashing my GPU

Lynxx83

New Member
Joined
Aug 9, 2023
Messages
6 (0.01/day)

tyvm for this.. is it coincidence that you published this guide around the same time,

shortly before/after intel released driver package 4644 which included the 'forced' fw update that's causing lots of problems for many users it seems ? :)

why it was forced here, this time? not sure. maybe because of this:


see, my card was also bought around that time, in Oct 2022...


but since driver 4644 , this was the first time ever after which I noticed a change in the BIOS version GPU-Z is reporting,


for example since 4644 believe it's 20.0.1064 in GPU-Z now, which is again a total different number which is confusing to me ;)

before *.4644 GPU-Z always reported 20.0.1053 since Oct 22 and all the driver updates inbetween I believe, but not sure:

bios.JPG


I wanted to go way back, from current DG02_1.3253 as Arc Firmare tool is reading it out now after 4644 & recently 4824 driver install,

to somewhere like the last DG02_1.3206 from 101.4335 Apr 26th, 2023 ...

but I see, there's only the "SOC" file available to flash,
will I miss out on anything or will it cause problems,
if I dont have corresponding oprom & fw / config files as well?

I just want go back entirely the way it was before in lets say May earlier this year including any bios/driver/config files there is, to see if it solves my numerous problems.

It might have been just 'partially faulty' right from the start and I never noticed ? (except for AV1 encoding on its HW media engine, which still doesnt work properly here but I didnt bother until now)
the reason I'm starting to suspect it might just be faulty after all, bad silicon so to speak, regardless of driver/bios,

Is because of a recent case on intel support community, whose card worked almost perfectly everywhere, but 3dmark reported it as "incompatible" for its's XeSS feature test^^ turned out it really was broken, because that test worked without problems for other Arc users & same drivers & I guess bios .

maybe mine is too. but still I wanted to try bios downgrade first.

if it wont help.. no idea what then.

€: just flashed the DG02_1.3206 file , PC still booting & no black monitor phew ;)
the flash tool confirms it's now 1.3206 , but GPU-Z still says 20.0.1064 ?

€2: and also, how can I install future driver / arc control upgrades or on occassion downgrade drivers again after DDU, but each time "preserving" the BIOS file version I got installed right now?

if there's no easy trick for doing so, does it mean I now have always do a manual FW flash as well, after each driver package install, no matter which one, if it's newer than Apr 26 ?
 
Last edited by a moderator:
tyvm for this.. is it coincidence that you published this guide around the same time,

shortly before/after intel released driver package 4644 which included the 'forced' fw update that's causing lots of problems for many users it seems ? :)

why it was forced here, this time? not sure. maybe because of this:


see, my card was also bought around that time, in Oct 2022...


but since driver 4644 , this was the first time ever after which I noticed a change in the BIOS version GPU-Z is reporting,


for example since 4644 believe it's 20.0.1064 in GPU-Z now, which is again a total different number which is confusing to me ;)

before *.4644 GPU-Z always reported 20.0.1053 since Oct 22 and all the driver updates inbetween I believe, but not sure:

View attachment 314066

I wanted to go way back, from current DG02_1.3253 as Arc Firmare tool is reading it out now after 4644 & recently 4824 driver install,

to somewhere like the last DG02_1.3206 from 101.4335 Apr 26th, 2023 ...

but I see, there's only the "SOC" file available to flash,
will I miss out on anything or will it cause problems,
if I dont have corresponding oprom & fw / config files as well?

I just want go back entirely the way it was before in lets say May earlier this year including any bios/driver/config files there is, to see if it solves my numerous problems.

It might have been just 'partially faulty' right from the start and I never noticed ? (except for AV1 encoding on its HW media engine, which still doesnt work properly here but I didnt bother until now)
the reason I'm starting to suspect it might just be faulty after all, bad silicon so to speak, regardless of driver/bios,

Is because of a recent case on intel support community, whose card worked almost perfectly everywhere, but 3dmark reported it as "incompatible" for its's XeSS feature test^^ turned out it really was broken, because that test worked without problems for other Arc users & same drivers & I guess bios .

maybe mine is too. but still I wanted to try bios downgrade first.

if it wont help.. no idea what then.

€: just flashed the DG02_1.3206 file , PC still booting & no black monitor phew ;)
the flash tool confirms it's now 1.3206 , but GPU-Z still says 20.0.1064 ?

€2: and also, how can I install future driver / arc control upgrades or on occassion downgrade drivers again after DDU, but each time "preserving" the BIOS file version I got installed right now?

if there's no easy trick for doing so, does it mean I now have always do a manual FW flash as well, after each driver package install, no matter which one, if it's newer than Apr 26 ?

To prevent the installation of newer firmware, you can overwrite the files present in the fw, fwdata and oprom folders before the driver installation, the easiest way to do that is to extract the *.exe driver files and make the changes.
Did you notice something different with your BIOS downgrade? I want to try here too, saw some report on Reddit that the newer firmware have more microstutter than the older ones (specifficaly 3251 vs 3257 for SOC1), but the real reason that I want to try is regarding Acer Predator Bifrost fan behavior, but I don't know if just changing the firmware will make difference because checking the firmware matrix, the only change for this card from driver 4644 (fw update that broke the fan behavior for Acer cards) and 4824 (the driver that supposedly fixed the fan behavior) was the OPROM, and for Acer card there's no OPROM before 4644...
 
but the real reason that I want to try is regarding Acer Predator Bifrost fan behavior, but I don't know if just changing the firmware will make difference because checking the firmware matrix, the only change for this card from driver 4644 (fw update that broke the fan behavior for Acer cards) and 4824 (the driver that supposedly fixed the fan behavior) was the OPROM
Both the main firmware and oprom are the majority of the time what makes changes to the card. It’s reasonable that the oprom will fix an issue but the firmware can stay the same.

remember though they are making modifications to the DRIVER and CONTROL PANEL during these updates. So things like fan control curves etc can be defined and regulated in software. Which means even if there is a firmware change you may still exhibit an issue.

it’s important to test changes and see what is actually causing issues don’t just blindly trust people on Reddit.
 
Both the main firmware and oprom are the majority of the time what makes changes to the card. It’s reasonable that the oprom will fix an issue but the firmware can stay the same.

remember though they are making modifications to the DRIVER and CONTROL PANEL during these updates. So things like fan control curves etc can be defined and regulated in software. Which means even if there is a firmware change you may still exhibit an issue.

it’s important to test changes and see what is actually causing issues don’t just blindly trust people on Reddit.
Thanks for the reply.

Regarding the fan problem, only the OPROM was changed, since driver earlier than 4644 don't have OPROM file for Acer Card, I found an OPROM with a different version in Acer's official support page.
Regarding the Reddit comment, I can't attest for the microstutter problem and don't want to verify, I just wanted my fan behaving like before
I opened an issue at IGCIT, Intel took some time to fix the problem, but it was only mitigated, the issue is still open there.
 
Back
Top