AMD/ATI Pixel Clock Patcher

[edstewbob]

I used the patcher with 17.3.1 and it worked fine but had a sort of strange behavior I wanted to mention. I am adding up to 6 GPU's in a server and each time I added a new patched GPU some errors showed up in the Device Manager Display Adapters section for Radeon RX470's that I was using, even tough I had already previously run the patcher for the prior GPU. I would run the patcher again and reboot and all would be hunky-dory. I'm not sure why this was required but it seems like something may have been refreshed because it would allow me to install the patch again each time. I did one time get the message saying the patch was already installed so I do know that the program recognizes this condition but in the case of adding additional Radeon's and rebooting I had to re-patch and reboot and they were fine after that.

[Black Octagon]

I've just updated to AMD drivers v17.3.2 and am having problems.

Version 1.4.4 seems to patch the drivers fine and I can then add custom resolutions with CRU Version 1.3...

But when I restart, only 60Hz is visible in Windows display settings. My custom resolutions (96Hz, 120Hz) are not visible.

I'm on an HD7970 and using this software to overclock an Overlord Tempest to 120Hz.

Toasty, do I perhaps need to roll back to a previous version of the AMD drivers?

Thank you

(EDIT: Just seen your remark in the CRU thread. Have reported this to AMD as requested. Nevertheless, if you know the latest version of the AMD drivers to which I can roll back I'd still be grateful to know this)

[ToastyX]

(03-18-2017 10:20 PM)Black Octagon Wrote:  (EDIT: Just seen your remark in the CRU thread. Have reported this to AMD as requested. Nevertheless, if you know the latest version of the AMD drivers to which I can roll back I'd still be grateful to know this)

17.1.2 works, or you can toggle GPU scaling after booting.

[ToastyX]

(03-17-2017 01:23 AM)edstewbob Wrote:  I used the patcher with 17.3.1 and it worked fine but had a sort of strange behavior I wanted to mention. I am adding up to 6 GPU's in a server and each time I added a new patched GPU some errors showed up in the Device Manager Display Adapters section for Radeon RX470's that I was using, even tough I had already previously run the patcher for the prior GPU. I would run the patcher again and reboot and all would be hunky-dory. I'm not sure why this was required but it seems like something may have been refreshed because it would allow me to install the patch again each time. I did one time get the message saying the patch was already installed so I do know that the program recognizes this condition but in the case of adding additional Radeon's and rebooting I had to re-patch and reboot and they were fine after that.

Windows automatically reinstalls the driver after adding a GPU even if it's the same driver, so you have to patch it again. It does this for everything, like when plugging a mouse into a different USB port.

[edstewbob]

@ToastyX when adding the 6th GPU and patching again upon reboot I get BSOD with error 0x000000fc (ATTEMPTED EXECUTE OF NOEXECUTE MEMORY) and the only way to get around it is to run without the modded RX 470 BIOS. I reinstalled the drivers at the same level and reapplied the patch but continue to get the BSOD. Any possible way this could be caused by the patcher since it was written to support an earlier version of the driver? The patch always seems to work with no issues and I assume it validates the code offset before applying the patch.

[ToastyX]

(03-19-2017 10:04 PM)edstewbob Wrote:  @ToastyX when adding the 6th GPU and patching again upon reboot I get BSOD with error 0x000000fc (ATTEMPTED EXECUTE OF NOEXECUTE MEMORY) and the only way to get around it is to run without the modded RX 470 BIOS. I reinstalled the drivers at the same level and reapplied the patch but continue to get the BSOD. Any possible way this could be caused by the patcher since it was written to support an earlier version of the driver? The patch always seems to work with no issues and I assume it validates the code offset before applying the patch.

I've confirmed that the current patcher (1.4.4) works correctly with the latest driver (17.3.2). Try renaming the patcher to atikmdag-patcher-bios.exe and repatch so it only patches the BIOS signature check. If you're already doing that, try not renaming the patcher so it patches all the limits. If that doesn't help, you're probably running into a different issue possibly triggered by the 6th GPU.

[Jan]

Hi ToastyX,
first of all thanks for this little tool!

However I can confirm with a patched RX460 (usual bios patch to 1024 cores)
that the current patcher (1.4.4) does not work correctly with the latest driver (17.3.2).

It crashed Chrome while watching YT every 3-5 minutes.
It also doesnt matter whether hardware accerlation is disabled or not.

So I thought, hey lets just patch the signature check and not do all the other stuff I know nothing about except that it causes an unstable system...
and i found the info to rename the tool to atikmdag-patcher-bios.exe
by chance after some googling. Oo

Please rethink this renaming solution as it really is as bad as it gets.
just make some checkboxes to allow patch selection.

I cant tell right now if it would have worked renaming the tool, as im back to WHQL driver for stability reasons.

regards Jan

[edstewbob]

ToastyX,

I think my problem was that I ran the patch while the drivers were still being updated. When there are 6 GPU's it takes several minutes for all of the driver updates to occur and this happens each time I add a new GPU. I tried it again and waited longer until all of the screen blinks and flashes had finished and the problem did not occur again. Thanks a bunch for this software it was very helpful to me.

[ToastyX]

(03-20-2017 07:57 PM)Jan Wrote:  Hi ToastyX,
first of all thanks for this little tool!

However I can confirm with a patched RX460 (usual bios patch to 1024 cores)
that the current patcher (1.4.4) does not work correctly with the latest driver (17.3.2).

It does work with the latest driver. I've been using it for the past week with an RX 480, which is in the same family. Not only that, I've confirmed it's patching the correct parts and not accidentally changing anything it wasn't supposed to change.

(03-20-2017 07:57 PM)Jan Wrote:  It crashed Chrome while watching YT every 3-5 minutes.
It also doesnt matter whether hardware accerlation is disabled or not.

That doesn't sound like a patch issue. It sounds like maybe the modified BIOS isn't stable with the latest driver.

(03-20-2017 07:57 PM)Jan Wrote:  So I thought, hey lets just patch the signature check and not do all the other stuff I know nothing about except that it causes an unstable system...
and i found the info to rename the tool to atikmdag-patcher-bios.exe
by chance after some googling. Oo

Please rethink this renaming solution as it really is as bad as it gets.
just make some checkboxes to allow patch selection.

I cant tell right now if it would have worked renaming the tool, as im back to WHQL driver for stability reasons.

I doubt it would. I only suggested it as a troubleshooting step. Patching all the limits should not cause stability issues. It only patches what needs to be patched to remove certain limits. It doesn't change any functionality by itself.

Adding checkboxes is easier said than done because the UI would have to handle situations where the driver is partially patched, which would complicate the simple patch and restore system. It would also cause people to ask what limits they should patch when in most cases it won't matter if all the limits are patched. Renaming the file makes it act like a separate program that only checks and patches the BIOS signature check.

[OneGun]

When i play BF1 with new drivers patched the game won't even open..If i unpatch works fine on my Xfire 7950s