Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Hey everybody,

Cool Kext @CMMChris
Just upgraded my ASUS Maximus X Code 8700k from a RX 580 Nitro+ SE to Sapphire Pulse 5700XT. Was an easy transition. Have a well working 0.5.8 Opencore Config for 10.15.4. Just added the pikera bootflag and disabled IGpu in bios. Using MacPro1,1 smbios. Have not implemented a powerplay table as of yet.

Question in regards to this on my hack
1) Does anyone know the position of the bios switch for the 5700XT pulse? I imagine one is performance and one is quiet, didn't change from what it shipped with, and there is no physical or online manual describing further
2) What, if any, reason is there for me to keep Whatevergreen now?
3) Would I see much improvement implenting a powerplay table with my scores below?
4) Is anyone using a different smbios (18,3, 19,1) with better results than imacpro1,1?
5) Benchmarks
Geekbench 5
i can see an improvement from my 580 to 5700 XT with your kext being used (haven't trialed without the kext as of yet)
CL Score RX580 45846 -> 5700XT 70639
Metal Score RX580 52583 -> 5700XT 78163

So happy about that. But interstingly my novabench scores have actually decreased at 558 GPU. Is novabench not a reliable benchmark?

Novabench
RX580: Graphics Score 558
Metal 3D: 60 FPS
OpenCL: 6433 GFLOPS

5700 XT:Graphics Score Same??? 558
Metal 3D: 60 FPS (shouldn't this be significantly higher?)
OpenCL: 5226 GFLOPS (why would this possibly have gone down?)

I can at least confrim with the last version of Furmark for OSX the 5700xt is registering 120-133fps

Currently have 3 displays connected, 1. 3360x1890 UHD via DP, 2. 1080p via HDMI, 3. 1440x900 via DVI to DP
not sure if that has anything to do with it.


And lastly, I'm using this primarily for FCP X work. Haven't done any real word testing/rendering/comparision yet as only got the card yesterday. But Bruce X scores are certainly faster by about 40% (this is writing to a x4 NVME export drive)
422 Pro res export: 8.6s (5700xt) vs 14.33 (RX580)
h264 export: 11.4s (5700xt) vs 18.2s (RX580)

Appreciate any words of wisdom, or confirmation of my thinking that not all OSX benchmark software is created equal, and that i really shouldn'y be bothered...

Attached my efi folder no serial below:

p.s. Love this hack, highly reccomend. Unrelated to the GPU, the only thing I've never been able to figure out with this setup is the ASMEIDA USB gen 2 controller which will imporeperly eject any drives connected after during/wake from sleep. Only software based workaround is Jettison. Just wish there was a hardware solution via an SSDT that did the trick!
 

Attachments

  • 0.5.8_5700XT_noserial.zip
    11 MB · Views: 120
Last edited:
Ok... So It looks like I need some help. I've tried step by step several times to no avail. Mac Pro 6,1 with Radeon VII in eGPU. I do not get any errors and everything seems right as rain. I reboot and nothing is different. RadeonBoost is not listed when running kextstat. SIP is disabled and I know all is well with that because I just did a fresh install and loaded PureWrangler right before this. I know there was some question as to whether this would work with an eGPU or not and perhaps that is the issue. Any help is greatly appreciated.


Hi did you get anyhelp on this? I had to open the Kext and add 6,1 because it wasn't listed. Where ever in the file it said 4,1 I put 6,1.
 
1) Does anyone know the position of the bios switch for the 5700XT pulse? I imagine one is performance and one is quiet, didn't change from what it shipped with, and there is no physical or online manual describing further
If can't find anything in the manual, you better open a ticket in Sapphire support website to ask them.

2) What, if any, reason is there for me to keep Whatevergreen now?
You need WEG to use HWAccel and DRM streaming.

3) Would I see much improvement implenting a powerplay table with my scores below?
Usually noticeable, and the score improvement is clearly outside normal error margin

4) Is anyone using a different smbios (18,3, 19,1) with better results than imacpro1,1?
I suggest you stick to iMac Pro 1,1 SMBIOS if you disabled the iGPU. iMac Pro can use AMD HWAccel, other iMac use Intel QuickSync.

5) Benchmarks
Is novabench not a reliable benchmark?

Novabench
RX580: Graphics Score 558
Metal 3D: 60 FPS
OpenCL: 6433 GFLOPS

5700 XT:Graphics Score Same??? 558
Metal 3D: 60 FPS (shouldn't this be significantly higher?)
OpenCL: 5226 GFLOPS (why would this possibly have gone down?)
No, that 60FPS is your monitor's refresh rate, V-Sync locked the test frame rate, make it unreliable.
 
If can't find anything in the manual, you better open a ticket in Sapphire support website to ask them.


You need WEG to use HWAccel and DRM streaming.


Usually noticeable, and the score improvement is clearly outside normal error margin


I suggest you stick to iMac Pro 1,1 SMBIOS if you disabled the iGPU. iMac Pro can use AMD HWAccel, other iMac use Intel QuickSync.


No, that 60FPS is your monitor's refresh rate, V-Sync locked the test frame rate, make it unreliable.

Thanks @h9826790 appreciate all the answers. Good to know about WEG, makes sense. I'll have to have a play with the power table, SSDT's are a bit new learning curve for me but I'll get there. Good to know about imacpro1,1. And had a feeling it was due to the monitor. Thanks again for the info mate 🙏
 
And had a feeling it was due to the monitor.
Not exactly the same benchmark you are talking about (OceanWave can't measure modern GPU's performance properly anyway), and my monitor is 144Hz. But you can see how v-Sync affect the result in this video

And v-Sync actually affect quite a few benchmarks. As you can see from this capture. All three "on screen" tests capped at 144FPS.
RX580 HiDPI2.png
 
Not exactly the same benchmark you are talking about (OceanWave can't measure modern GPU's performance properly anyway), and my monitor is 144Hz. But you can see how v-Sync affect the result in this video

And v-Sync actually affect quite a few benchmarks. As you can see from this capture. All three "on screen" tests capped at 144FPS.
View attachment 915262
Interesting, good to know. as I don't game don't think it's much a concern for me.

In regards to bios settings, I placed a ticket with sapphire, but also logged into Windows and pulled the ROM's off each bios.
The Right Position (furthers from case) is the Performance(1670,1815,1925)/ Left is Quiet (1605/1755/1905mhz).

Decided to implement a powerplay table with opencore's device properties for my Saphire 5700 XT Pulse going off the performance default pulled from Morepowertool

Undervolted with the following:
GPU Voltage: 1000 (from 1250)
MaxVoltageSoc: 1000 (from 1050)
Power Limit GPU (W): 150 (from 195)
Fan Target Temp: 80
Min Fan: 600 (from 1100)

I can see improvement of scores in Geekbench 5 now with the the powerplaytable
OpenCL 72461 (from 70639)
Metal 81881 (from 78163)

@h9826790 or @CMMChris do you know is there a way that I can monitor these fan/power settings in mac osx? istats doesn't see any fans or temps for the gpu (using latest virtualsmc)
 

Attachments

  • Radeon_RX_5700_XT_PowerPlayTable_Generator_PulsePerformance_Undervolted.xlsx.zip
    32 KB · Views: 130
Interesting, good to know. as I don't game don't think it's much a concern for me.

In regards to bios settings, I placed a ticket with sapphire, but also logged into Windows and pulled the ROM's off each bios.
The Right Position (furthers from case) is the Performance(1670,1815,1925)/ Left is Quiet (1605/1755/1905mhz).

Decided to implement a powerplay table with opencore's device properties for my Saphire 5700 XT Pulse going off the performance default pulled from Morepowertool

Undervolted with the following:
GPU Voltage: 1000 (from 1250)
MaxVoltageSoc: 1000 (from 1050)
Power Limit GPU (W): 150 (from 195)
Fan Target Temp: 80
Min Fan: 600 (from 1100)

I can see improvement of scores in Geekbench 5 now with the the powerplaytable
OpenCL 72461 (from 70639)
Metal 81881 (from 78163)

@h9826790 or @CMMChris do you know is there a way that I can monitor these fan/power settings in mac osx? istats doesn't see any fans or temps for the gpu (using latest virtualsmc)
No idea, I don't have any Navi card. But if the following command in terminal can't capture anything, then I don't know any software can monitor the GPU paramaters
Code:
ioreg -l |grep \"PerformanceStatistics\" | cut -d '{' -f 2 | tr '|' ',' | tr -d '}' | tr ',' '\n'|grep 'Temp\|Fan\|%\|\|Hz'
 
Last edited:
Yah that code doesnt work unfortuantely, i think i've read somewhere in these forums that apple removed navi metrics
 
April 19 2020: V1.2 Removed support for Vega 56, Vega 64 and Vega Frontier (REASON)
Okay, I have a Vega Frontier Edition GPU. I took the latest version of your kext and brought the Vega FE support back. Now I have about +25% Geekbench score and no fan issue (BTW, I've never had a fan issue). Tried both default and compute workloads, and the second one works best for my card.

P.S. Thank you for making it possible!
 
No idea, I don't have any Navi card. But if the following command in terminal can't capture anything, then I don't know any software can monitor the GPU paramaters
Code:
ioreg -l |grep \"PerformanceStatistics\" | cut -d '{' -f 2 | tr '|' ',' | tr -d '}' | tr ',' '\n'|grep 'Temp\|Fan\|%\|\|Hz'
He can probably go find the driver for the Navi card in IORegistryExplorer or IOJones and look under:
1589642385970.png

My driver is the x4000
 
Okay, I have a Vega Frontier Edition GPU. I took the latest version of your kext and brought the Vega FE support back. Now I have about +25% Geekbench score and no fan issue (BTW, I've never had a fan issue). Tried both default and compute workloads, and the second one works best for my card.

P.S. Thank you for making it possible!
Great to hear. Would you mind sharing it?
 
Of course you can go ahead and edit the info.plist of RadeonBoost.kext to your likings. For example you can add your own properties. This comes handy if you want to additionally overclock or undervolt your GPU by using a PowerPlay Table, which of course can be added to the Kext as well

Can you please tell how to do these tweaks? Or share some references about it?
 
Hi
With last ktext April 24 2020: V1.4 my cMP with RX580 Pulse 8G and original ROM is not working.

I have several freeze of Mac OS Mojave.

If I disable radeaonboost it works again without freeze.
 
Please try version 1.5 from here: https://www.hackintosh-forum.de/for...res-wie-am-echten-mac-unter-windows/?pageNo=1
Haven't added it in this topic yet. Also please note that you need to use it along Whatevergreen for Polaris cards.
@CMMChris
I installed on OC ktext folder radeonboost 1.5 and also activated in config
And I have just after login in Mac OS direct freeze :(
I tried 3 times restarts and same issue
then I removed from EFI folder OC the radeonboost ketext and it is now good

I hope it will be help you
1589886149169.png

1589886176845.png

1589886195565.png

1589886285057.png
 

Thanks for this

I have a couple of questions though: first do I copy inside lilufriend.kext plist the key and version for whatevergreen as well or only lilu (as whatevergreen may be inside lilu's plugins folder). Finaly do I install lilu in both /System/Library/Extensions and /Library/Extensions. That part is not absolutely clear in lilufriend readme.

Thanks for your time and my apologies if my questions are "naive" but when it comes to kext modifications and such alterations, I am below novice.
 
Thanks for this

I have a couple of questions though: first do I copy inside lilufriend.kext plist the key and version for whatevergreen as well or only lilu (as whatevergreen may be inside lilu's plugins folder). Finaly do I install lilu in both /System/Library/Extensions and /Library/Extensions. That part is not absolutely clear in lilufriend readme.

Thanks for your time and my apologies if my questions are "naive" but when it comes to kext modifications and such alterations, I am below novice.
You can leave it outside lilu.
Just check
kextstat | grep vit9696
 
You can leave it outside lilu.
Just check
kextstat | grep vit9696
Thanks a lot

I did that and this time I get slightly better results in Geekbench (by 700) than without radeonboost. No adverse effects in FCPX either, although there is no discernible difference in BruceX test before and after (maybe half a second)

Thanks for your help!
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.