Just upgraded from RX580 to Radeon VII (it's just $350 from a local seller, like new condition, with 2 years+ warranty. I really can't skip it 
)
Anyway, as usual, all AMD GPU are seriously overvolted from factory (Default setting in Wattman shows 1801MHz @1093mV). Properly downvolt the card can lower the power draw, temperature, and fan noise.
Since we can't mod the Radeon VII's ROM, so, the best method should be inject a soft PowerPlay table.
I ran Luxmark for three times. 1st and 3rd run with PP table, 2nd run without PP table. And the Radeon VII clearly draw less on the 1st and 3rd run (same score range for all three runs). So, I believe this is a good enough evidence to confirm that my kext mod is effective.
(N.B. the very high power draw before the 1st Luxmark run is the Furmark test. The peak from the 2nd run is really the power draw from Luxmark, not Furmark)
WARNING: Only install the CustomPowerPlay.kext if you have proper back up, and know how to recover (in worst case).
The pre-created kext for 10.14.6 (link) (depreciated)
The pre-created kext for 10.15.2 (link) (depreciated)
Pre created kext for Mojave and above
Power Radeon VII with mini 6pins ONLY (link)
(Depreciated, LudaCrisvp helped me to create the required kext) Technically, make a codeless kext to insert the PowerPlay table is a better way to do it, but I am not good at that. If anyone know how to do it properly, please join this thread, hopefully we can make a way that no need to mod the system kext.
So, back to the topic. SIP must be disabled to make it work, otherwise, kext won't load, and all you have is just black screen (If you mod the original AMD kext, but not install the extra kext that I posted in the above link)
Step 0)
Disable SIP, backup the boot drive, backup your data.
Step 1)
To find out the min voltage that work for your own card. My suggestion is to boot into Windows, and stress your GPU (e.g. by running Unigine Heaven Extreme Preset), then slowly lower the voltage in AMD Wattman until you see artifacts. In my case, I only move the orange dot down, but leave everything at default.
And I can see artifacts when the voltage reduced to around 971mV. So, I give it 10mV buffer, and set 981mV as the target voltage.
Step 2)
Once you find the voltage, you can now go back to macOS, download and open the attached excel table (credit to CMMChris, TonyMacX86)
Step 3)
The table is pretty much edited already. All you need to do is just mod the number in the AB column. e.g. my target voltage is 981mV, therefore, I enter 981 in the box AB6.
P.S. Power Limit (%) is changed to zero in that table, but test shows no effect (by running Furmark, max power draw still at the same range). If you want to limit the max power, you may change the Power Limit (W). E.g. if change that to 200, 200 x 120% = 240W max. I tested this parameter. It works in macOS. With 250W (120%), Furmark cost a 36A power draw from the PSU 12V line, With 200W (120%), that will reduce to 31A.
Since 250W x 120% = 300W
And 200W x 120% = 240W
The expected difference is 60W. And (36-31)A x 12V = 60W. Exactly as per the prediction. Therefore, we can safely confirmed the "Power Limit (W)" has effect in macOS.
Also, for info, even I limited the card to 200W (120%), Luxmark still shows the same score. So, this should not affect the card's performance for absolutely most real world task.
If your target voltage is above 1000mV, then you should modify MaxVoltageGfx and MaxVoltageSoc accordingly.
At this moment, both parameters set to 4000, which mean 1000mV x4. If your target voltage (minimum stable voltage) is 1010mV, then you should give it a little bit buffer, e.g. 1020mV, and enter 4080 in box AB11 and AB13.
(Depeciated, I tried quite a few different irregular numbers, seems won't break anything, and the downvolt still work as expected) I am not 100% sure, but there may be some limitation for these two parameters. e.g. the voltage must be at 50mV step. Which means, 4200 (1050mv x4) is a valid number, but 4120 (1030mV x4) may be not.
If you want a bit more performance, you can set the HBM clock speed to 1100MHz. Tests shows this will let the card draw another ~10W, but run ~5% faster (please test if your card can really do that in Windows before you set that in the PowerPlay table).
Step 4)
Once finish edit the table, go to box H72, there will be a hex code generated. e.g. in my case
Step 5)
go to
http://tomeko.net/online_tools/hex_to_base64.php
Step 6)
place your own hex code that generated from step 4 into the convertor to get the base64 code. e.g. in my case
Step 7) (The steps from here is depreciated, please use the new kext injection method at post #21, which no need to mod any system kext)
Now, we can start to mod the kext. Open terminal and enter
after enter your admin password, enter
inside the editor, move down to <key>aty_properties</key> and add four lines (update: 2 more lines added, total 6 lines now)
please replace the "data" content by your own base64 code. The modded kext should looks like this
Step 8)
After finish editing, press Control+X to save and exit (as per the instruction inside the editor), you should need to press "Y" and "enter" to properly save the modded kext.
Step 9)
Back to terminal, enter
After you finish this step, everything is done. You may open the info.plist inside AMD10000Controller.kext to have a look, the PowerPlay table should be there.
You may also check the System Reports -> Extensions, the AMD10000Controller.kext now should be obtained from "Unknown" but not Apple anymore.
If you are happy, you can reboot your Mac now (WARNING: if SIP is enabled, the kext won't load, and all you have is just black screen. You may setup remote control to handle this kind of situation, just in case).
After reboot, the PowerPlay table should be working now. In my case, everything works perfectly after the kext mod. Interestingly, both Unigine Heaven (~4%) and Valley (~7%) shows better score after the mod, my feeling is because less thermal limitation. And since Luxmark more affected by the VRAM performance, so, not quite affected.
Last but not least, if black screen only after reboot, perform a NVRAM will re-enable SIP and disable the PowerPlay table. So that, you can remove the CustomPowerPlay.kext.
You should also able to recover the entire OS in recovery partition if you want to (hold Command + R to boot).
)Anyway, as usual, all AMD GPU are seriously overvolted from factory (Default setting in Wattman shows 1801MHz @1093mV). Properly downvolt the card can lower the power draw, temperature, and fan noise.
Since we can't mod the Radeon VII's ROM, so, the best method should be inject a soft PowerPlay table.
I ran Luxmark for three times. 1st and 3rd run with PP table, 2nd run without PP table. And the Radeon VII clearly draw less on the 1st and 3rd run (same score range for all three runs). So, I believe this is a good enough evidence to confirm that my kext mod is effective.
(N.B. the very high power draw before the 1st Luxmark run is the Furmark test. The peak from the 2nd run is really the power draw from Luxmark, not Furmark)
WARNING: Only install the CustomPowerPlay.kext if you have proper back up, and know how to recover (in worst case).
Pre created kext for Mojave and above
Power Radeon VII with mini 6pins ONLY (link)
(Depreciated, LudaCrisvp helped me to create the required kext) Technically, make a codeless kext to insert the PowerPlay table is a better way to do it, but I am not good at that. If anyone know how to do it properly, please join this thread, hopefully we can make a way that no need to mod the system kext.
So, back to the topic. SIP must be disabled to make it work, otherwise, kext won't load, and all you have is just black screen (If you mod the original AMD kext, but not install the extra kext that I posted in the above link)
Step 0)
Disable SIP, backup the boot drive, backup your data.
Step 1)
To find out the min voltage that work for your own card. My suggestion is to boot into Windows, and stress your GPU (e.g. by running Unigine Heaven Extreme Preset), then slowly lower the voltage in AMD Wattman until you see artifacts. In my case, I only move the orange dot down, but leave everything at default.
And I can see artifacts when the voltage reduced to around 971mV. So, I give it 10mV buffer, and set 981mV as the target voltage.
Step 2)
Once you find the voltage, you can now go back to macOS, download and open the attached excel table (credit to CMMChris, TonyMacX86)
Step 3)
The table is pretty much edited already. All you need to do is just mod the number in the AB column. e.g. my target voltage is 981mV, therefore, I enter 981 in the box AB6.
P.S. Power Limit (%) is changed to zero in that table, but test shows no effect (by running Furmark, max power draw still at the same range). If you want to limit the max power, you may change the Power Limit (W). E.g. if change that to 200, 200 x 120% = 240W max. I tested this parameter. It works in macOS. With 250W (120%), Furmark cost a 36A power draw from the PSU 12V line, With 200W (120%), that will reduce to 31A.
Since 250W x 120% = 300W
And 200W x 120% = 240W
The expected difference is 60W. And (36-31)A x 12V = 60W. Exactly as per the prediction. Therefore, we can safely confirmed the "Power Limit (W)" has effect in macOS.
Also, for info, even I limited the card to 200W (120%), Luxmark still shows the same score. So, this should not affect the card's performance for absolutely most real world task.
If your target voltage is above 1000mV, then you should modify MaxVoltageGfx and MaxVoltageSoc accordingly.
At this moment, both parameters set to 4000, which mean 1000mV x4. If your target voltage (minimum stable voltage) is 1010mV, then you should give it a little bit buffer, e.g. 1020mV, and enter 4080 in box AB11 and AB13.
(Depeciated, I tried quite a few different irregular numbers, seems won't break anything, and the downvolt still work as expected) I am not 100% sure, but there may be some limitation for these two parameters. e.g. the voltage must be at 50mV step. Which means, 4200 (1050mv x4) is a valid number, but 4120 (1030mV x4) may be not.
If you want a bit more performance, you can set the HBM clock speed to 1100MHz. Tests shows this will let the card draw another ~10W, but run ~5% faster (please test if your card can really do that in Windows before you set that in the PowerPlay table).
Step 4)
Once finish edit the table, go to box H72, there will be a hex code generated. e.g. in my case
Code:
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 5)
go to
http://tomeko.net/online_tools/hex_to_base64.php
Step 6)
place your own hex code that generated from step 4 into the convertor to get the base64 code. e.g. in my case
Code:
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Step 7) (The steps from here is depreciated, please use the new kext injection method at post #21, which no need to mod any system kext)
Now, we can start to mod the kext. Open terminal and enter
Code:
sudo bashafter enter your admin password, enter
Code:
nano /System/Library/Extensions/AMD10000Controller.kext/Contents/Info.plistinside the editor, move down to <key>aty_properties</key> and add four lines (update: 2 more lines added, total 6 lines now)
Code:
<key>PP_PhmSoftPowerPlayTable</key>
<data>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</data>
<key>PP_DisablePowerContainment</key>
<integer>1</integer>
<key>PP_DisableDIDT</key>
<integer>1</integer>please replace the "data" content by your own base64 code. The modded kext should looks like this
Step 8)
After finish editing, press Control+X to save and exit (as per the instruction inside the editor), you should need to press "Y" and "enter" to properly save the modded kext.
Step 9)
Back to terminal, enter
Code:
touch /System/Library/Extensions; kextcache -update-volume /After you finish this step, everything is done. You may open the info.plist inside AMD10000Controller.kext to have a look, the PowerPlay table should be there.
You may also check the System Reports -> Extensions, the AMD10000Controller.kext now should be obtained from "Unknown" but not Apple anymore.
If you are happy, you can reboot your Mac now (WARNING: if SIP is enabled, the kext won't load, and all you have is just black screen. You may setup remote control to handle this kind of situation, just in case).
After reboot, the PowerPlay table should be working now. In my case, everything works perfectly after the kext mod. Interestingly, both Unigine Heaven (~4%) and Valley (~7%) shows better score after the mod, my feeling is because less thermal limitation. And since Luxmark more affected by the VRAM performance, so, not quite affected.
Last but not least, if black screen only after reboot, perform a NVRAM will re-enable SIP and disable the PowerPlay table. So that, you can remove the CustomPowerPlay.kext.
You should also able to recover the entire OS in recovery partition if you want to (hold Command + R to boot).
Last edited:
