Jump to content
jnolla

Vega Frontier Edition on High Sierra

754 posts in this topic

Recommended Posts

Okay, so this is an update to VegaTab.  It is published by the author, lihaoyun6, and he has a thread over on the Tonymac site, so you can communicate with him there.

Folks, this is free software, and that means if you have any issues, contact him, not me.  Don't expect a completely flaw free utility, and absolutely, you should use it with caution! 

He doesn't speak English very well, but he is trying....so if you can communicate in Chinese, all the better!

Edited by meaganmargaret

Share this post


Link to post
Share on other sites
Advertisement

@nefilim You can't use the extracted PowerPlayTable directly due to the separators in between the hex values. If you did, your card has entered fail-safe mode which might make it look like fan control is working because it is running at a fixed speed. However, performance does suffer a lot in fail-safe mode because the card runs in a fixed power state (boot state). Please run a benchmark or check your clocks to see if you are in fail-safe mode (no clocks displayed in that case).

 

And just to make that clear: If your extracted PPT would result in working fan control, it would be working without the PPT as well because macOS does read the PPT inside your cards BIOS to get the proper settings.

Edited by CMMChris

Share this post


Link to post
Share on other sites

hi everyone.

 

i got some problem with WIFI after upgrade to vega 56, pci wifi card Broadcom BCM94360 not working 5G, with gtx 770 - i have no problem

 

i try Broadcom BCM94331CSAX with vega - everything ok, and try second Broadcom BCM94360 - same issue

 

 

vega 56 - on PCI Express x16, PCI-wifi on x4

 

based on diagram, PCI x16 and x4 not connected

 

What is the problem I dont understand, I think  the problem is in the chipset

 

someone else has same chipset and similar components?

 

 

 

ga-z77-ds3h/i7-3770/VEGA 56

diagram.png

Edited by mjslap

Share this post


Link to post
Share on other sites
On 12/12/2018 at 9:33 PM, CMMChris said:

Try this


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

I am playing Rise of the Tomb Raider with this and have no overheating or crashes.

 

Bildschirmfoto 2018-12-12 um 21.33.27.jpg

 

Hi CMMChris! I have been trying your excel tool and these settings seemed like the best I've tested so far. I have Sapphire Nitro+ RX Vega 64 and still getting some coil whine noise on macOS (on win10 it's silent). Would you suggest any specific setting for Nitro+ RX Vega 64? I suppose yours is a different brand RX Vega 64.

 

Thank you!

Share this post


Link to post
Share on other sites
5 hours ago, CMMChris said:

Try these, near to no coil whine, great performance:

 

Bildschirmfoto 2019-01-29 um 16.45.22.jpg

Thanks Chris,

 

I will try these settings tomorrow when I'm on my Hackintosh. However, lowering the P6 and P7 Core rates won't lower the performance? 

Share this post


Link to post
Share on other sites

Getting 207k in Geekbench and 32k in Luxmark with these.

Also add PP_DisableDIDT=1 and PP_DisablePowerContainment=1

Edited by CMMChris

Share this post


Link to post
Share on other sites

Just tried the settings suggested with PP_DisableDIDT=1 and PP_DisablePowerContainment=1.

 

Geekbench: 200302

LuxMark: 31101

 

The results are pretty much the same maybe a slightly lower than before. Coil whine noise sound level is the same (not too bad). I will monitor these settings for a while.

 

I guess it is impossible to get rid of the coil whine 100% with Sapphire Nitro+ Rx Vega 64.

 

And should I keep PP_DisableDIDT=1 and PP_DisablePowerContainment=1 settings there even if I switched back to previous numbers?

 

Thank you so much Chris!

Edited by yigpig

Share this post


Link to post
Share on other sites

Yes, disabled PowerContainment entirely removes power limits and prevents the aggressive throttling macOS is doing. DisableDIDT results in a general performance enhancement. You can benchmark your card with and without those settings.

 

Regarding coil whine, try to play with the voltage. Go down a bit or up and check if the coil whine changes. Every card behaves a little different. Just make sure you set the same voltage on all power levels, this plays a big role in the coil whine issues on macOS.

 

You can also try changing the P7 clock up or down. P7 frequency also plays a big role in coil whine. On my card I find 1562MHz to produce the least amount of coil whine. If I go down a bit or up a bit, coil whine increases. Just play around and try finding the sweet spot for your card.

 

For more performance you can also play with the memory clock which gives a bigger boost than increasing GPU clock. Try which maximum frequency you can reach until it gets unstable during a full benchmark run in Valley. Most cards should reach 1030MHz just fine. This way you can likely boost your scores up to 210k in Geekbench. Another thing you can try is increasing clocks for P4 to P6. P6 can also be the same as P7. This also gives a slight score boost.

 

While playing with clocks and voltages please also have a look at your UI smoothness. I find that some frequency and voltage levels result in a smoother subjective UI performance than others, totally independent from benchmark results.

Edited by CMMChris

Share this post


Link to post
Share on other sites
2 hours ago, CMMChris said:

Yes, disabled PowerContainment entirely removes power limits and prevents the aggressive throttling macOS is doing. DisableDIDT results in a general performance enhancement. You can benchmark your card with and without those settings.

 

Regarding coil whine, try to play with the voltage. Go down a bit or up and check if the coil whine changes. Every card behaves a little different. Just make sure you set the same voltage on all power levels, this plays a big role in the coil whine issues on macOS.

 

You can also try changing the P7 clock up or down. P7 frequency also plays a big role in coil whine. On my card I find 1562MHz to produce the least amount of coil whine. If I go down a bit or up a bit, coil whine increases. Just play around and try finding the sweet spot for your card.

 

For more performance you can also play with the memory clock which gives a bigger boost than increasing GPU clock. Try which maximum frequency you can reach until it gets unstable during a full benchmark run in Valley. Most cards should reach 1030MHz just fine. This way you can likely boost your scores up to 210k in Geekbench. Another thing you can try is increasing clocks for P4 to P6. P6 can also be the same as P7. This also gives a slight score boost.

 

While playing with clocks and voltages please also have a look at your UI smoothness. I find that some frequency and voltage levels result in a smoother subjective UI performance than others, totally independent from benchmark results.

Thank you for the detailed explanation Chris! I will play with it and try to find the sweet spots.

Great job with coming up with such a tool for the community!

Share this post


Link to post
Share on other sites
On 1/30/2019 at 11:28 AM, yigpig said:

Just tried the settings suggested with PP_DisableDIDT=1 and PP_DisablePowerContainment=1.

 

Geekbench: 200302

LuxMark: 31101

 

The results are pretty much the same maybe a slightly lower than before. Coil whine noise sound level is the same (not too bad). I will monitor these settings for a while.

 

I guess it is impossible to get rid of the coil whine 100% with Sapphire Nitro+ Rx Vega 64.

 

And should I keep PP_DisableDIDT=1 and PP_DisablePowerContainment=1 settings there even if I switched back to previous numbers?

 

Thank you so much Chris!

And what about the temperature regime

Share this post


Link to post
Share on other sites

Got a second Vega 64 in my build. Works great despite just having a 850W power supply. I recommend using min. 1000W tho.

Bildschirmfoto 2019-02-01 um 16.46.48.jpg

51472155_2237047009897699_7209610181276073984_o.jpg

Share this post


Link to post
Share on other sites
On 2/1/2019 at 12:31 PM, pitrysha said:

And what about the temperature regime

What do you mean exactly by temperature regime?

Share this post


Link to post
Share on other sites
21 minutes ago, yigpig said:

What do you mean exactly by temperature regime?

Temperature in idle and under load

Снимок экрана 2019-02-04 в 11.21.10.png

Снимок экрана 2019-02-04 в 11.22.52.png

Share this post


Link to post
Share on other sites
1 minute ago, pitrysha said:

Temperature in idle and under load

Снимок экрана 2019-02-04 в 11.21.10.png

Снимок экрана 2019-02-04 в 11.22.52.png

 

Idle -> 36 - 38 degrees

Under luxmark load -> 55 - 58 degrees

Share this post


Link to post
Share on other sites

Hi everyone, I'm looking for suggestions for further improvement.  I have coil while (not terrible but there), fuzzy text in some apps (skype for business in the chat windows) and poor performance (LuxMark v3.1 = 6384), although I don't game in MacOS.

 

 The column on the left is what I'm currently running and the column on the right is @CMMChris optimal settings from his Jan 29 post.  When I tried Chris's settings, the fan spun up every 60 seconds and the performance in LuxMark was still poor. (btw, if I use stock macOS settings, my Lexmark score is 30k).

 

I have a Phantom Gaming X Radeon RX VEGA 64 8G (PG X R RX VEGA 64 8G) with:
Engine Clock / Memory Clock
Boost Clock: 1546 MHz / 945 MHz
Base Clock: 1247 MHz / 945 MHz

compare.png

Edited by douirc

Share this post


Link to post
Share on other sites

Extract PPT from your cards VBIOS and implement it into the Excel generator. Now you can tune your default PPT which should resolve your issue.

Share this post


Link to post
Share on other sites

Please help. I'm very new to Hackintosh, I've been over a week figuring how to set this up. Now finally when I got the operating system up and running smoothly I noticed my graphic card (ASUS RADEON RX VEGA 64 8GB) Is not running its fans and the GPU is very slow. I checked my information and this is what I see. I've tried searching everywhere no results
I've installed:
 AMD10000Controler.kext 
 Whatevergreen
 Lilu

 RadeonLowFanSpeed.kext


My system:
Motherboard: ASUS Prime X299 Deluxe 

GPU: ASUS Strix RADEON RX VEGA 64 8GB
CPU: i9 7940x

DRAM: Corsair Vengeance LPX 4x16GB (64GB)

Screenshot 2019-02-06 at 16.31.08.png

Share this post


Link to post
Share on other sites

You say you installed "AMD10000Controller.kext". This is a system component pre-installed on every copy of macOS and part of the AMD graphics driver. No need to install this. Whatever you did, get rid of it.

Share this post


Link to post
Share on other sites
On 1/2/2019 at 10:44 PM, skinmunster said:

Hi CMMChris,

I own a ASUS Radeon RX Vega 56 ROG Strix OC, which is a custom design. I extracted first my PP_PhmSoftPowerPlayTable from the Win-Registry / org. BIOS to compare. (card is switched to quiet-profile)

But I was surprised to recognize that my raw-value table has a different length (679 byte, vs. 694 in your excel) and not only has diifferent values.

Some byte positions/functions seems to be equal (like power limit%), but some are located on different one (like max temperature, D24/E24 vs. S24/T24.)

 

I will try the injection tomorrow ..

 

KR

Hello skinmunster, did you have any success with injection? I just bought the same card and I'm starting to find out the best settings. Thanks.

Share this post


Link to post
Share on other sites
On 2/1/2019 at 10:51 AM, CMMChris said:

Got a second Vega 64 in my build. Works great despite just having a 850W power supply. I recommend using min. 1000W tho.

Bildschirmfoto 2019-02-01 um 16.46.48.jpg

51472155_2237047009897699_7209610181276073984_o.jpg

looks like geek bench cant use both cards together...

Share this post


Link to post
Share on other sites
On 2/16/2019 at 10:00 AM, darthsian said:

Hello skinmunster, did you have any success with injection? I just bought the same card and I'm starting to find out the best settings. Thanks.

Sorry, i did not really checked it, because i thought  the different length of the raw value tables will not work

Edited by skinmunster

Share this post


Link to post
Share on other sites

Hi Vega users,

 

I just put together a new build consisting of a Asus Z390i, 9900K, and Sapphire Vega 64 8GB HBM2 Reference Card running a triple boot with Ubuntu 18.04, Windows 10, and macOS Mojave’s 10.14.3. Everything is working really well with the exception of the GPU. I originally noticed the fans were on full blast and used VGTab to generate my device properties in clover, after tweaking the settings dozens of times I have found the ones that seem to make the system the most stable in terms of noise and temp. I am not unhappy with my benchmark scores, and I have never seen a crash during high load activity.

 

The card worked out of box (aside from the fan issue) and appears in system info correctly. I am using iMac18,3 SMBIOS, and have my iGPU running headlessly with quick sync, DRM, HEVC all working (full hw acceleration)

 

I plan on using Chris’ spreadsheet to see if I can improve things further, as I am not sure how accurate the VGTab PPT is for my sapphire reference model.

 

Anyways, the actual problem I am having is that my build seems to have a GPU related crash after many hours (12-48 hrs) and I know it is the GPU since I can see it in Console as well as the system logs. It reads just like some other users who have posted here, com.apple.DumpGPURestart appears many times. Here is what I see over and over in the system logs

 

com.apple.xpc.launchd[1] (com.apple.DumpGPURestart): Service only ran for 0 seconds. Pushing respawn out by 10 seconds.

 

The error is very similar to the one Mentioned here

 

Some other users have mentioned they experienced something similar as well, such as this user who mentioned they had it happen with dual monitors (I am only using one monitor right now via HDMI at 1080p).

 

I can even ssh into my machine in this crashed state and check that the machine is in fact idle and not doing anything heavy. (I have tried disabling screensaver/sleep/monitor sleep in case that was related, all of those things work fine regularly though)

 

My first guess was that this must be a hardware problem, so I left windows and Ubuntu on for 72 hours each without letting them go to sleep, and they both remained stable while macOS does not. So in order to learn more I setup istats menu and a couple scripts to monitor the performance statistics in ioreg. It seems the temps are very normal and the fan and GPU are not doing anything intense prior to the sudden crash. Another interesting thing I noticed is that over ssh I cannot use vclist, get the vbios output from ioreg, or even check the stats via ioreg after it has crashed, ioreg just hangs.

For example here are the last two ioreg statistics readings prior to the GPU crash
 

"Fan Speed(%)"=20
"Fan Speed(RPM)"=1001
"Temperature(C)"=42
"Total Power(W)"=25

"Fan Speed(%)"=20
"Fan Speed(RPM)"=984
"Temperature(C)"=42
"Total Power(W)"=25

 

If I run `ioreg -l -n GFX0@0 -r` while the gpu is crashed, it prints out the log and the last line I see before ioreg hangs is

 

  +-o AMDRadeonX5000_AMDVega10GraphicsAccelerator  <class AMDRadeonX5000_AMDVega10GraphicsAccelerator, id 0x1000003e0, registered, matched, active, busy 977 (122238179$
    | {

 

I saw one user here mentioned a kind of memory leak for the Vega GPU, I am very interested in hearing more about that as I need to investigate and see if something similar is happening to me. Perhaps after left running for a long time unable to free up memory our GPUs just crash? At this point I’m just wildly guessing, but I would love to learn or investigate this issue further. Even though it seems this problem could be unrelated to PPT and speed/temps.

These are the device properties I am using right now (created with VGTab) that keep temps and power down.

 

<dict>
	<key>PP_DisableClockStretcher</key>
	<integer>1</integer>
	<key>PP_DisablePowerContainment</key>
	<integer>0</integer>
	<key>PP_DisableULV</key>
	<integer>1</integer>
	<key>PP_EnableUploadFirmware</key>
	<integer>0</integer>
	<key>PP_Falcon_QuickTransition_Enable</key>
	<integer>1</integer>
	<key>PP_GfxOffControl</key>
	<integer>0</integer>
	<key>PP_PhmSoftPowerPlayTable</key>
	<data>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</data>
	<key>PP_PhmUseDummyBackEnd</key>
	<integer>0</integer>
	<key>PP_ToolsLogSpaceSize</key>
	<integer>524288</integer>
</dict>

 

Any guidance or advice on this problem would be greatly appreciated. If anyone wants to see my config.plist, kexts, EFI drivers or similar just let me know. Again, like I said everything else works really really well, even using the GPU on Heaven benchmark (> 1 hr), Luxmark, Geekbench, Cinebench or similar reveal great results that I am happy with. I am just stuck on this stability issue.

Update
I did some extra reading on the subject, and I think this has something to do with HWMonitor2, as I scrolled through the logs I came across these repeated lines
 

com.apple.xpc.launchd[1] (org.slice.HWMonitorSMC2-Helper[33725]): Service exited with abnormal code: 78
com.apple.xpc.launchd[1] (org.slice.HWMonitorSMC2-Helper): Service only ran for 0 seconds. Pushing respawn out by 10 seconds.
com.apple.xpc.launchd[1] (com.apple.DumpGPURestart): Service only ran for 0 seconds. Pushing respawn out by 10 seconds.
com.apple.xpc.launchd[1] (org.slice.HWMonitorSMC2-Helper[33728]): Could not find and/or execute program specified by service: 155: Refusing to execute/trust quarantined program/file: org.slice.HWMonitorSMC2-Helper


All of this sounds identical to what is being described here. But I am not completely certain.

Edited by Vinnymac
Add HWMonitor2 Update

Share this post


Link to post
Share on other sites
3 hours ago, Vinnymac said:

Anyways, the actual problem I am having is that my build seems to have a GPU related crash after many hours (12-48 hrs) and I know it is the GPU since I can see it in Console as well as the system logs. It reads just like some other users who have posted here, com.apple.DumpGPURestart appears many times. Here is what I see over and over in the system logs

 


com.apple.xpc.launchd[1] (com.apple.DumpGPURestart): Service only ran for 0 seconds. Pushing respawn out by 10 seconds.

 

 

I have exactly the same problem with my Sapphire Vega 64 Nitro+ 8GB. I am still trying to figure out where it comes from. One question first: Do you have a Vega kext installed in your Clover folder (not matter which configuration it has)? This could be causing the issue perhaps. I have experienced no issues at least during the last two days without any kext and no changed Vega settings at all, though this might be just coincidentally. Anyway, with no adjustments at all the GPU fans remain on way too long once they started. I am now trying out some other settings with copying the code from either a kext file or the Excel sheet into the config.plist and see if I find a solution here.

Edited by MatthiasSt
typo

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

  • Recently Browsing   0 members

    No registered users viewing this page.

×