Contribute
Register

[Guide] HP Elite 8300 & 6300 Pro (all form factors) using Clover UEFI hotpatch

What's your prediction for how long the GT710 will be supported? I'm currently on Mojave, so I know I've got at least a few years of upgrades waiting.
It's good through Big Sur betas so far. After the next macOS 11.1.0 in (Fall 2021) version it may be dropped. Hard to guess when or what Apple will do.
 
Last edited:
Successfully upgraded to OpenCore 0.6.3 Today!
Screenshot 2020-11-04 at 18.30.09.png
Screenshot 2020-11-04 at 18.31.22.png
Screenshot 2020-11-04 at 18.20.20.png
 
I was doing some tests today, and installed the new kexts and OC 0.6.3 on
my mainstream Catalina 10.15.7 HP 8300 with i5-3570K and RX560. It is
configured with iMacPro1,1 SMBIOS, and has shikigva=144 for hardware
DRM decoding and proper operation of Quicktime. The DRM chart says that
I only need shikigva=128, but I find that Quicktime doesn't work that way.

I didn't have any of the startup problems for the kexts or OC 0.6.3 like
I had on Big Sur 11.0.1. Everything worked as usual from the start. My
VideoProc reports 4K H264 and HEVC, and I verify the HEVC using the Sony
Swordsmith file.

I tested the performance of the Radeon dGPUs using Pavo's AGPMInjector.kext
versus Radeonboost.kext for RX560 and RX570, and for both Catalina and Big Sur.
Here's what my scores looked like.

Big Sur RX570-RB RX570-AGPM Catalina RX560-RB RX560-AGPM
GeekBench 4 Metal 106373 110058 72512 72412
OpenCL 104946 115152 76148 76318
GeekBench5 Metal 33328 33268 24859 23478
OpenCL 32915 32709 23292 21644

So in summary, I didn't see much of a difference between the RadeonBoost.kext and
the AGPMInjector in terms of Geekbench performance. I noticed that some of the
properties injected by RadeonBoost were only enabled on the RX570.

By comparison, I believe that Geekbench 4 scores for the Nvidia GT730 are about
18000-21000 depending on card and brand. The HD4000 scores are around 8000.

IMHO, the GT730 is the best value if you're going with an Nvidia card, and the HD4000
doesn't give great video performance. My understanding is that the HD4000 is not going
to support DRM decoding in Big Sur, and the Nvidias don't support HEVC.

I spent some time before looking at using a MacPro6,1 SMBIOS as recoommended for Big
Sur. I'm a bit confused about how to configure dGPUs for MP6,1. The native discrete
Power Management recommendations of Toleda called for renaming the MP6,1 dGPU as GFX2,
but the later kexts and guides seem to recommend against that renaming. It also unclear
how to set up IGPU power management and whether DRM settings work in Catalina. Does
anyone have any recommendations on that?
 
The native discrete
Power Management recommendations of Toleda called for renaming the MP6,1 dGPU as GFX2,
but the later kexts and guides seem to recommend against that renaming. It also unclear
how to set up IGPU power management and whether DRM settings work in Catalina. Does
anyone have any recommendations on that?
When using WEG with HP8300/6300, define GFX0, IGPU and dGPU.
FakeSMC must be used if VirtualSMC uses a dGPU that does not support the dGPU's temperature sensor input.

It's not an "Installation" issue, so could you create a thread with "Desktop Compatibility" or "Graphics"?
I would like to explain in detail in the thread you created.
 
Last edited:
I just wanted to report that I installed the new 10.15.7 security update
problem-free, and tested AppleTV DRM, VideoProc, and Geekbench on
my iMacPro1,1 SMBIOS on an HP8300 with an RX560. New Build is 19H15.

Also, I was reading the Dortania guide today, and noticed it calls for
installing SSDT-USBX for SMBIOS iMacPro1,1, so I installed it. It seems
that the kUSB power levels that I can see in IORegistry match those in
the SSDT. Still testing and working on the Big Sur update...

6:46PM update
Big Sur update went smoothly. New build is 11.0.1 Beta (20B5022a).
 
Last edited:
I just wanted to report that I installed the new 10.15.7 security update
problem-free, and tested AppleTV DRM, VideoProc, and Geekbench on
my iMacPro1,1 SMBIOS on an HP8300 with an RX560. New Build is 19H15.
The mattystonnie guide is the best choice when using the RX570 with Catalina 10.15.7.
I have posted the ACPI patch method for HP8300/6300 here, so please add it at the end of ACPI-> Patch.

1.Download "SSDT-RX580-Version 1.0.aml" from here and add it to the end of ACPI-> Add.
2.Download "DAGPM.kext.zip" and get "DAGPM.kext".
2-1.Change to a board-id that uses Info.plist->IOKitPersonalities->AGPM->Machines->Mac-7BA5B2D9E42DDD94.
(Mac-7BA5B2D9E42DDD94 = iMacPro1,1)
(Mac-F60DEB81FF30ACF6 = MacPro6.1)
Info.plist->IOKitPersonalities->AGPM->Machines->[board-id]->GFX0 is required by WEG and should not be modified.
2-2.Set the id of the dGPU to use for Vendor1002DeviceXXX (device-id = XXX) under GFX0.
( Vendor1002Device67DF = RX570/580 )
( Vendor1002Device67ff = RX560 )
3.Save the modified Info.plist and add the resulting DAGPM.kext on top of the WEG in Kernel-> Add.
(I don't know where to add it, but I do this)
4.Reboot and do NVRAM reset.

Note:When using MacPro6.1, TV + cannot be used unless iGPU = disable is set in the BIOS.

Enjoy Geekbench's scores and beautiful images.
 
The mattystonnie guide is the best choice when using the RX570 with Catalina 10.15.7.
I have posted the ACPI patch method for HP8300/6300 here, so please add it at the end of ACPI-> Patch.

1.Download "SSDT-RX580-Version 1.0.aml" from here and add it to the end of ACPI-> Add.
2.Download "DAGPM.kext.zip" and get "DAGPM.kext".
2-1.Change to a board-id that uses Info.plist->IOKitPersonalities->AGPM->Machines->Mac-7BA5B2D9E42DDD94.
(Mac-7BA5B2D9E42DDD94 = iMacPro1,1)
(Mac-F60DEB81FF30ACF6 = MacPro6.1)
Info.plist->IOKitPersonalities->AGPM->Machines->[board-id]->GFX0 is required by WEG and should not be modified.
2-2.Set the id of the dGPU to use for Vendor1002DeviceXXX (device-id = XXX) under GFX0.
( Vendor1002Device67DF = RX570/580 )
( Vendor1002Device67ff = RX560 )
3.Save the modified Info.plist and add the resulting DAGPM.kext on top of the WEG in Kernel-> Add.
(I don't know where to add it, but I do this)
4.Reboot and do NVRAM reset.

Note:When using MacPro6.1, TV + cannot be used unless iGPU = disable is set in the BIOS.

Enjoy Geekbench's scores and beautiful images.

Asural,

While I applaud your enthusiasm, I wish you could temper your ardor and take the time
to thoroughly test and compare the solutions you propose to other solutions, because the
information you're providing is not complete or accurate.

You recommend downloading, modifying, and installing the "DAGPM.kext.zip" from
mattystonie's guide for power management of the dGPU. It turns out that it is more difficult
for the average person here to follow your directions, and if they did so, they would likely
have an incomplete solution with questionable or no-power-management performance.

If you right click on the DAGPM.kext that you recommend, choose "Show Package Contents,"
and look in the "Contents" folder, you will see that it is a code-less kext that consists solely
of an "Info.plist." If you click on the Info.plist and have Xcode installed, it will open the plist
file. I opened it, clicked on IOKitPersonalities, AGPM, Machines, and Mac-7BA5B2D9E42DDD94
to display them all, and the file looks like the first screen shot below.

As you can see, this kext was originally written by Toleda as part of his guide,
"macOS (nearly) Native Discrete GPU Power Management" at post #1 in

The complete version of the original dAGPM.kext can be downloaded from there.
It is shown as the second screen shot below.

Unfortunately, mattystonie's version has been corrupted, in that the device sections
like "Vendor1002Device67DF" were originally dictionaries in Toleda's version with multiple
entries, but in mattystomie's version they have been replaced with strings and the dictionary
entries are missing. The missing entries are AGDCEnabled, control-id, max-power-state,
min-power-state, and a Heuristic dictionary. It is not surprising that several users in the long
92-page mattystonie "AMD Radeon Performance Enhanced SSDT" thread complain that they
are not getting proper power management following mattystonie's recipe, and end up with
graphics cards whose fans spin loudly and constantly. I recommend that if you want to use
the dAGPM.kext, you should use Toleda's version with the editing and the verifications that he
recommends and ensure that you end up with the correct heuristic under AGPM in your GFX0
framebuffer under the display. Toleda is also the person who recommends renaming the
MacPro6,1 AMD graphics card as GFX2, so its not at all clear that his 2018 solution is the
correct solution with a current WEG and a MacPro6,1 SMBIOS.

A simpler dGPU power management solution that is more current is provided by Pavo's
AGPMInjector.kext. I won't put a link to his website because it violates our rules here, but
one can easily find the github for "Pavo-IM/AGPMInjector" online. I downloaded the
latest release version and its easy to use. It does all the editing for you -- you just download
and run the application, choose the SMBIOS and the Graphics card that you're using, and it
does the rest of the kext composition. The instructions are clear and easy to follow. I've
checked the produced kext's Info.plist, and it contains all of the property keys and values as
recommended by Toleda. It works without excessive fan noise.

As far as mattystonie's "SSDT-RX580-Version 1.0.aml" goes, you will find that most users
who report on it in his thread report modest gains at best. There is no official RX560 or RX570
version. I have tested it, and I don't see any improvement. At one time, his SSDT and the
Radeonboost.kext seemed to be in competition, and its not clear that either was a clear winner,
as I reported in my recent testing of Radeonboost. Some of the properties that they try to inject
do not apply to the RX560 or RX570, and you can see that in IORegistryExplorer when they are
actually in use. I could not see any improvement in performance using mattystonie's SSDT.
One of the few benefits of the Radeonboost.kext is that it contains an AGPM section that works
properly.

@pastrychef has Radeonboost in some of his EFI's. His discussion and a link is in post #2633 of

Asural, if you can see a mattystonie improvement, could you please post before and after scores
from Geekbench so user's can see what kind of improvements are possible?

Thanks.
 

Attachments

  • Screen Shot 2020-11-06 at 9.32.42 AM.png
    Screen Shot 2020-11-06 at 9.32.42 AM.png
    130.6 KB · Views: 51
  • Screen Shot 2020-11-06 at 9.39.46 AM.png
    Screen Shot 2020-11-06 at 9.39.46 AM.png
    130.4 KB · Views: 42
Last edited:
Asural,

While I applaud your enthusiasm, I wish you could temper your ardor and take the time
to thoroughly test and compare the solutions you propose to other solutions, because the
information you're providing is not complete or accurate.

You recommend downloading, modifying, and installing the "DAGPM.kext.zip" from
mattystonie's guide for power management of the dGPU. It turns out that it is more difficult
for the average person here to follow your directions, and if they did so, they would likely
have an incomplete solution with questionable or no-power-management performance.

If you right click on the DAGPM.kext that you recommend, choose "Show Package Contents,"
and look in the "Contents" folder, you will see that it is a code-less kext that consists solely
of an "Info.plist." If you click on the Info.plist and have Xcode installed, it will open the plist
file. I opened it, clicked on IOKitPersonalities, AGPM, Machines, and Mac-7BA5B2D9E42DDD94
to display them all, and the file looks like the first screen shot below.

As you can see, this kext was originally written by Toleda as part of his guide,
"macOS (nearly) Native Discrete GPU Power Management" at post #1 in

The complete version of the original dAGPM.kext can be downloaded from there.
It is shown as the second screen shot below.

Unfortunately, mattystonie's version has been corrupted, in that the device sections
like "Vendor1002Device67DF" were originally dictionaries in Toleda's version with multiple
entries, but in mattystomie's version they have been replaced with strings and the dictionary
entries are missing. The missing entries are AGDCEnabled, control-id, max-power-state,
min-power-state, and a Heuristic dictionary. It is not surprising that several users in the long
92-page mattystonie "AMD Radeon Performance Enhanced SSDT" thread complain that they
are not getting proper power management following mattystonie's recipe, and end up with
graphics cards whose fans spin loudly and constantly. I recommend that if you want to use
the dAGPM.kext, you should use Toleda's version with the editing and the verifications that he
recommends and ensure that you end up with the correct heuristic under AGPM in your GFX0
framebuffer under the display. Toleda is also the person who recommends renaming the
MacPro6,1 AMD graphics card as GFX2, so its not at all clear that his 2018 solution is the
correct solution with a current WEG and a MacPro6,1 SMBIOS.

A simpler dGPU power management solution that is more current is provided by Pavo's
AGPMInjector.kext. I won't put a link to his website because it violates our rules here, but
one can easily find the github for "Pavo-IM/AGPMInjector" online. I downloaded the
latest release version and its easy to use. It does all the editing for you -- you just download
and run the application, choose the SMBIOS and the Graphics card that you're using, and it
does the rest of the kext composition. The instructions are clear and easy to follow. I've
checked the produced kext's Info.plist, and it contains all of the property keys and values as
recommended by Toleda. It works without excessive fan noise.

As far as mattystonie's "SSDT-RX580-Version 1.0.aml" goes, you will find that most users
who report on it in his thread report modest gains at best. There is no official RX560 or RX570
version. I have tested it, and I don't see any improvement. At one time, his SSDT and the
Radeonboost.kext seemed to be in competition, and its not clear that either was a clear winner,
as I reported in my recent testing of Radeonboost. Some of the properties that they try to inject
do not apply to the RX560 or RX570, and you can see that in IORegistryExplorer when they are
actually in use. I could not see any improvement in performance using mattystonie's SSDT.
One of the few benefits of the Radeonboost.kext is that it contains an AGPM section that works
properly.

@pastrychef has Radeonboost in some of his EFI's. His discussion and a link is in post #2633 of

Asural, if you can see a mattystonie improvement, could you please post before and after scores
from Geekbench so user's can see what kind of improvements are possible?

Thanks.

LisaAlterEgo

I think you can explain it well to others.
I will attach a reference file.


I found that the information has no meaning for reference, so I deleted it.
 
Last edited:

LisaAlterEgo

I think you can explain it well to others.
I will attach a reference file.
Those seem like outstanding GB5 scores.

My RX570 is an Asus ROG Strix 4 GB model. I downloaded GPU Tweak II from Asus and put the
card in Overclock-Mode but did not change any other default parameters. I don't think these settings
have any effect on macOS performance.

In Windows10 with OpenCore 0.6.3, I scored 28900 on Vulcan and 32216 on OpenCL.
I noticed that OpenCL test kept the GPU clock and memory clock at the maximum, but the Vulcan
test was mostly throttled down.

In Big Sur with OpenCore 0.6.3 and iMacPro SMBIOS, I scored 33053 on Metal and 32534 on OpenCL.
I'm currently just using Pavo's AGPMInjector and am otherwise not injecting any special RX570 parameters.

Can you run Furmark? I downloaded Furmark ROG Edition 0.8.5.0. I put it in 1920 x 1080 mode and ran the
1080 Benchmark Preset. It got 83-87 FPS at 80 degrees C for a score of 5061 points (84 FPS) and the fans
started going full blast.
 
Back
Top