WX9100 Power mods - help w/ powerplay tables

[conron]

How do you guys deal with coil wine?

 

[LabRat 891]

How do you guys deal with coil wine?

Mine doesn't have any that I can hear
(Note: I just had an ultrasound done, and could hear that)

Lexluthermeister has a guide on qwelling coil whine.

How to quickly & easily fix coil-whine(coil choke noise)

A forum user recently asked about Coil Whine, which is the noise a choke coil sometimes makes as it operates. This is almost always a video card problem, but can happen on any other computer or electronics components employing a choke in it's circuitry, including motherboards & power supplies...

www.techpowerup.com

IIRC, Mussels has used Hot Glue as well for same. Hot glue has some 'dampening' properties; super glue is a more rigid bond.

Even, with the cleanest power, many cards still 'scream'.

 

[conron]

Mine doesn't have any that I can hear
(Note: I just had an ultrasound done, and could hear that)

Lexluthermeister has a guide on qwelling coil whine.

How to quickly & easily fix coil-whine(coil choke noise)

A forum user recently asked about Coil Whine, which is the noise a choke coil sometimes makes as it operates. This is almost always a video card problem, but can happen on any other computer or electronics components employing a choke in it's circuitry, including motherboards & power supplies...

www.techpowerup.com

IIRC, Mussels has used Hot Glue as well for same. Hot glue has some 'dampening' properties; super glue is a more rigid bond.

Even, with the cleanest power, many cards still 'scream'

Isnt what decides how "clean" the power is, is the efficiency of the PSU? Or is that just PCMR snake oil?
Reason I bring that up is cause I have an EVGA 650W but it's only 80+ bronze

 

[LabRat 891]

Isnt what decides how "clean" the power is, is the efficiency of the PSU? Or is that just PCMR snake oil?
Reason I bring that up is cause I have an EVGA 650W but it's only 80+ bronze

Yes/No.
Logically a well-designed and highly-effcient PSU should have ample 'filtering' and would be expected to produce clean power.

It's not totally snake oil, but it is 100% Marketing Wank

I'm not an EE, but I do know it's actually more complicated than that.
Other consumer SMPSs make good example of that. Efficiency and 'EMC' are not as implicitly interlinked as one (with basic electronics knowledge) would assume.

There's a LOT that actually factors into 'coil whine', DC that's not so clean, is only a part of it.
While I've not seen anyone knowledgeable write-up anything solid on it, I've noticed 'inducted, audible noise' from SMPS devices is seemingly common. However, most people cannot hear it or the noise is covered-up by ambient sounds.
Just dampening / solidifying the physical source of the audible noise, seems to help immensely.

 

[conron]

Oh boy this gpu is fun. Finally got the coil whine sorted out and now Im having a new fun issue. Im stuck with a perpetual Code 43 in windows. I've DDU multiple times, taken out my other gpu, reseated the GPU, REFLASHED THE BIOS, tried both official and unofficial drivers. Im lost as to what to do guys

 

[DR4G00N]

Oh boy this gpu is fun. Finally got the coil whine sorted out and now Im having a new fun issue. Im stuck with a perpetual Code 43 in windows. I've DDU multiple times, taken out my other gpu, reseated the GPU, REFLASHED THE BIOS, tried both official and unofficial drivers. Im lost as to what to do guys

That doesn't sound good, that's what my first one ended up doing when it died. I thought I damaged the interposer but maybe it was just bad luck instead.

 

[conron]

That doesn't sound good, that's what my first one ended up doing when it died. I thought I damaged the interposer but maybe it was just bad luck instead.

I reseated it and it decided to unkill itself for now. Happy days
Also new funky video from Iceberg Tech about HBCC. The TLDR: Dont forget to turn off HBCC guys

 

[DR4G00N]

The TLDR: Dont forget to turn off HBCC guys

I mean considering it barely makes a difference for 8GB Vega these 16GB ones certainly don't need it for games. Still useful if you're doing actual work on it though.

It'd be interesting to see some good testing of rebar, I have it enabled on mine but does it really help? IDK. Seems to make it run a little hotter though.

 

[conron]

That doesn't sound good, that's what my first one ended up doing when it died. I thought I damaged the interposer but maybe it was just bad luck instead.

It rekilled itself. It's not completely dead as it still outputs just fine. Im not sure what its doing

 

[conron]

If anyone ever has an idea as to what could be causing this Code 43 lmk. Again it outputs fine but no driver functionality whatsoever

 

[eidairaman1]

If anyone ever has an idea as to what could be causing this Code 43 lmk. Again it outputs fine but no driver functionality whatsoever

Tried driver removal and install? 43 is a hardware fault and that right there if drivers wont install or if the os freezes after driver install indicats you might need the board repaired by people at northridgefox or krisfix.de just to name 2

 

[conron]

Tried driver removal and install? 43 is a hardware fault and that right there if drivers wont install or if the os freezes after driver install indicats you might need the board repaired by people at northridgefox or krisfix.de just to name 2

Im not sure I wanna bother them with that lol. Not really sure what to do. Without drivers installed, I get that "POSTed driver is not the same as driver trying to load" error. With them installed, I get error 43. Full driver wipe, full vbios wipe. Oh man

 

[DR4G00N]

Im not sure I wanna bother them with that lol. Not really sure what to do. Without drivers installed, I get that "POSTed driver is not the same as driver trying to load" error. With them installed, I get error 43. Full driver wipe, full vbios wipe. Oh man

Yeah it's probably done for, good thing they're cheap at least (IMO I'd just get a 1080 instead).

 

[conron]

Yeah it's probably done for, good thing they're cheap at least (IMO I'd just get a 1080 instead).

I need something with 16gb of VRAM so that won't work. Of course it breaks RIGHT AFTER I got the VR set up working (literally what I bought it for)

Which VBIOS are yall using?

 

[DR4G00N]

Which VBIOS are yall using?

I'm using the 11124 one, always worked fine so I never tried any others.

 

[LabRat 891]

I need something with 16gb of VRAM so that won't work. Of course it breaks RIGHT AFTER I got the VR set up working (literally what I bought it for)

Which VBIOS are yall using? View attachment 317495

I am on the vBIOS that DR4G00N recommended. I've attached the vBios and exact ATItool version I've used for 2x successful MI25 conversions (both the 6+8pin and 8+8pin variants).

I used eidairman1's guide (for Polaris, oddly) for the conversions.
Not sure if MI25's 'lock' like Polaris but, I elected to do the extra step.

Get amdvbflash 2.93

https://www.techpowerup.com/download/ati-atiflash/
Then look for Version 4.68, underneath 4.68 you will see a link to show older version, click that and you should see version 2.93

Follow these steps carefully:

1. Open your "C:" drive in Windows, create a new folder, call it atiflash.

2. Extract the contents of atiflash_293.zip to the "C:\atiflash" folder

3. Find the bios file you downloaded and rename the bios file to 5 digits with .rom on end Example RX 580, change file to rx580.rom (r****.rom)

4. Move r****.rom to the "C:\atiflash" folder

5. Open the start menu, type cmd, when cmd.exe appears right click it and hit "Run as administrator"

6. At the command prompt type in cd\ and press enter until only C:\> is on the screen.

7. Type in cd/atiflash and press enter, you should see C:\atiflash>

8. Type in amdvbflash.exe -i to identify the gpu you want to flash with a 0, 1, or 2)

9. At the C:\atiflash> prompt type in
amdvbflash.exe -unlockrom 0
(there is a space between the e and -unlockrom and 0)(0 is the number zero)
press enter and you will be prompted that the "ROM Unlocked" (or 1, 2)

10. At the C:\atiflash> prompt type in
amdvbflash.exe -f -p 0 r****.rom
(there is a space between the .exe and -f, and a space between -f and -p and 0).
(Or 1, 2)

Press Enter

Follow directions on screen and if flash is successful exit the command prompt and restart, windows should show up on screen

(Also) as eidairman1 mentioned,
The Code 43 usually indicates a hardware problem.

Neither MI25 I've played with has had even the slightest-detectable coil whine.
So, it's not impossible there already was something wrong with your example or,
'fixing it' caused the Code 43-related damage.

IIRC I *think* I've gotten a Code 43 when using a secondary card for video but had 'hard crashed' the MI25-WX9100 during tweaking and stability testing.
I believe this 'throws' the MI25-WX9100 into some non-UEFI compatible 'recovery mode' or something.
'talking out my arse but, I'm not the only one that's seen Vega(s) do this kinda thing.
In fact, one of the early-researcher/experimenters with the MI25 over @ Level1, strongly believes these cards are only RELIABLY flashable, using a clip-on programmer.

I've also had a Legitimate Vega 64 do very similar, appearing 'dead', and 'coming back' after a HARD power cycle:
The only way I've found to 'fix' was to cut all power to the system (flip PSU switch) while POSTing, wait a moment, then re-power (flip-on PSU) and re-boot the machine.

WARNING: This is against every 'good practice' I've ever learned
generally HARD power losses (of any cause) are unhealthy for components.

(Just pulling and re-inserting the card, didn't seem to work.
I'm guessing that a hard power loss immediately 'drains' any and all caps on the card, and triggers some kind of firmware-level 'reset' of sorts.)


Note: I've also had complications arise from 'hidden devices' in device manager.
DDU's supposed to clean them up but, IME it doesn't.

On the driver-side of things:
You're probably already trying these drivers:
select the (22.40 kernel?) MPT workaround during setup, and use the matching v5/22.40 Radeon Settings
use the SDI (Snappy Driver Installer) instead of the other options

I know it is even more money down the drain but,
you could pick up another MI25 for ~$50, and resell your Code 43 as For Parts for $20+ probably.

 

[conron]

I am on the vBIOS that DR4G00N recommended. I've attached the vBios and exact ATItool version I've used for 2x successful MI25 conversions (both the 6+8pin and 8+8pin variants).

I used eidairman1's guide (for Polaris, oddly) for the conversions.
Not sure if MI25's 'lock' like Polaris, but I elected to do the extra step.

(Also) as eidairman1 mentioned,
The Code 43 usually indicates a hardware problem.

Neither MI25 I've played with has had even the slightest-detectable coil whine.
So, it's not impossible there *was* already something wrong with your example, or 'fixing it' caused the Code 43-related damage.

IIRC I *think* I've gotten a Code 43 when using a secondary card for video but had 'hard crashed' the MI25-WX9100 during tweaking and stability testing.
I believe this 'throws' the MI25-WX9100 into some non-UEFI compatible 'recovery mode' or something.
'talking out my arse but, I'm not the only one that's seen Vega(s) do this kinda thing.
In fact, one of the early-researcher/experimenters with the MI25 over @ Level1, strongly believes these cards are only RELIABLY flashable, using a clip-on programmer.

I've also had a Legitimate Vega 64 do very similar, appearing 'dead', and 'coming back' after a HARD power cycle:
The only way I've found to 'fix' was to cut all power to the system (flip PSU switch) while POSTing, wait a moment, then re-power (flip-on PSU) and re-boot the machine.

(Just pulling and re-inserting the card, didn't seem to work.
I'm guessing that a hard power loss immediately 'drains' any and all caps on the card, and triggers some kind of firmware-level 'reset' of sorts.)


Note: I've also had complications arise from 'hidden devices' in device manager.
DDU's supposed to clean them up but, IME it doesn't.

On the driver-side of things:
You're probably already trying these drivers:
select the 22.40 kernel MPT workaround during setup,
use the SDI (Snappy Driver Installer) instead of the other options

I know it is even more money down the drain but,
you could pick up another MI25 for ~$50, and resell your Code 43 as For Parts for $20+ probably.

Thank you so much for putting this all together. I'll begin testing these different ideas tomorrow and let you all know thank you again

 

[conron]

I've tried everything to no avail. If anyone wants the GPU for parts or to play around with to see if they can get it working, let me know. I have no use for it now. I may be upgrading to a 6800 for black Friday. Thank you everybody for your help!

 

[amomynus]

Hi
I would like to ask if someone can help me make a wx9100 undervoltable and overclockable?
I can see it works with 23.4.2 amernime driver but a could not find the modded driver. Their latest driver listed on the website was 23.5.2.(Joined the DC server but did not find the needed driver but asked.)
My wx9100 is a flashed mi25.

I managed to make it work with softpowerplaytable editor as mentioned before but my problem is when the gpu crashes i have to run ddu and reinstall drivers. (It is a hard crash as described before.)
Would be nice to do tweaking inside adrenalin.

Thanks in advance.

 

[oneb1t]

it is working again on R-ID drivers 24.1.1 - WHQL-R-ID-Software-Hybrid-24.1.1-Win10-Win11-PolarisVegaNavi-Nebula.exe
uninstall everything with DDU and then install it again

 

[oneb1t]

i found one more problem to be able to have wattman working it is required to apply PPtable from other card which have it working then wattman interface is visible and working

 

[toshiaki_12]

In my case, I couldn't revive wattman using oneb1t's method.
I revived Whatman using the following method.
However, even if the voltage is lowered, the heat will hardly change.
I flashed mi25 to wx9100 with atiflash_293, not a ROM writer.
Maybe the situation will change if I use a ROM writer.
1. Clean using DDU.
2. Install whql-amd-software-adrenalin-edition-22.11.2-win10-win11-dec8.
3. Add "DisableWorkstation 1" to the registry.
4. Apply (save) pptable.

sorry for poor english, my native language is not english.

I've picked up various tips here and elsewhere. thank you.

 

[DR4G00N]

However, even if the voltage is lowered, the heat will hardly change.

That's because with lower voltage it has more power budget available and so it will boost higher until it reaches the power limit again.
It will still run at 170W, only with a higher frequency.

 

[toshiaki_12]

That's because with lower voltage it has more power budget available and so it will boost higher until it reaches the power limit again.
It will still run at 170W, only with a higher frequency.

I see. thankyou for the explanation.
Is there a way to disable or suppress that "boost"?

My aim is to make this MI25/WX9100 have a performance/heat ratio like my vega56 reference model.
Is it not possible?

Currently, in my opinion, this MI25/WX9100 is too hot.
I loaded my vega56's pptable and tried running mi2/wx9100 with the power limit set to -10%.
It's still hot.

If "boost" is controlled by hardware (e.g. FPGA M2S010 near the 25P80), there's nothing I can do about it.

 

[DR4G00N]

I see. thankyou for the explanation.
Is there a way to disable or suppress that "boost"?

My aim is to make this MI25/WX9100 have a performance/heat ratio like my vega56 reference model.
Is it not possible?

Currently, in my opinion, this MI25/WX9100 is too hot.
I loaded my vega56's pptable and tried running mi2/wx9100 with the power limit set to -10%.
It's still hot.

If "boost" is controlled by hardware (e.g. FPGA M2S010 near the 25P80), there's nothing I can do about it.

You need to rethink your fan shroud, it's simply not getting enough airflow directed through the heatsink. If you were to extend that plexi shroud along the full length of the card and do away with the tape it would probably work better. Basically, for fans right angles and restrictions/reductions are the enemy to airflow so the less of those you have the better.

This is how I have mine set up currently; Just a hole cut into the shroud and a piece of cardboard as an air dam to help guide the air.



Here's how it performs @ 220W without an undervolt (thanks to AMD's janky software I accidentally removed the undervolt while setting the fan speed higher).
Certainly hot, but that's totally normal for Vega. I normally run at 170W undervolted which makes it settle around 75-80C with the fan only at 2000 RPM.



I also suggest using the WX 9100 or Vega FE PPTables so the HBM runs at 945MHz, the Vega 56/64 tables make it stay at or strongly favor 800MHz.