WhatEverGreen Support Topic

[Babiloe]

Upgraded whatever green got 2 mins to boot..

old one takes 2 secs.. 

checked via "log show --predicate "processID == 0" --start $(date "+%Y-%m-%d") --debug > kenapalama.txt"
 

Quote

2021-01-20 09:01:10.614034-0800 0x74       Default     0x0                  0      0    kernel: (IOGraphicsFamily) GTrace synchronization point 9bc
2021-01-20 09:01:55.626200-0800 0x74       Default     0x0                  0      0    kernel: IOConsoleUsers: gIOScreenLockState 3, hs 0, bs 0, now 0, sm 0xe0000280
2021-01-20 09:01:57.381804-0800 0x74       Default     0x0                  0      0    kernel: PM response took 1755 ms (53, powerd)
...

2021-01-20 09:02:02.599443-0800 0x74       Default     0x0                  0      0    kernel: (IOGraphicsFamily) GTrace synchronization point 9f7
2021-01-20 09:02:48.026304-0800 0x74       Default     0x0                  0      0    kernel: IOConsoleUsers: gIOScreenLockState 3, hs 0, bs 0, now 0, sm 0xe0000280
2021-01-20 09:02:49.475176-0800 0x74       Default     0x0                  0      0    kernel: PM response took 1448 ms (53, powerd)
2021-01-20 09:02:49.475512-0800 0x74       Default     0x0                  0      0    kernel: (IOGraphicsFamily) GTrace synchronization point 9fd
....

2021-01-20 09:02:54.616414-0800 0x74       Default     0x0                  0      0    kernel: (IOGraphicsFamily) GTrace synchronization point a32
2021-01-20 09:03:40.025445-0800 0x74       Default     0x0                  0      0    kernel: IOConsoleUsers: gIOScreenLockState 3, hs 0, bs 0, now 0, sm 0xe0000280
2021-01-20 09:03:41.550168-0800 0x74       Default     0x0                  0      0    kernel: PM response took 1524 ms (53, powerd)
2021-01-20 09:03:41.550504-0800 0x74       Default     0x0                  0      0    kernel: (IOGraphicsFamily) GTrace synchronization point a38
...

2021-01-20 09:04:38.601774-0800 0x74       Default     0x0                  0      0    kernel: (IOGraphicsFamily) GTrace synchronization point aa8
2021-01-20 09:05:24.025678-0800 0x74       Default     0x0                  0      0    kernel: IOConsoleUsers: gIOScreenLockState 3, hs 0, bs 0, now 0, sm 0xe0000280
2021-01-20 09:05:25.617760-0800 0x74       Default     0x0                  0      0    kernel: PM response took 1591 ms (53, powerd)
 

 

new whatevergreen long boot.zip

 

Whatevergreen.kext 

 

NEW  1.4.6  SLOW runme  https://www.mediafire.com/file/tzhlvwfgcle8afg/Send_me_MacBook-Pro_newslow.zip/file

 

OLD FAST runme https://www.mediafire.com/file/9fz3zhmnfg34bjw/Send_me_MacBook-Pro_-oldfast.zip/file

Edited January 20, 2021 by Babiloe
link rune

[polkaholga]

On 1/17/2021 at 3:05 PM, Neblogas said:

I'Ve some strange issue with the latest kext. System boots normal in verbose, but at the time when apple logo appears(see it very dimmed) the display goes black and stays so for few minutes and gets back to normal again.

 

With older versions is everything ok.

Same on my "MBP15,2" ( see sig ) and external display through HDMI does't work anymore

 

On my Z390 Hack no problems though !

Edited February 5, 2021 by polkaholga

[Planet X]

On 12/13/2020 at 5:26 PM, MacNB said:

 

Well it is not "normal" but consistent with what I observe with my RX580. If I run the GB5 straight after booting, I get around 25000 score but if  macOS goes to sleep and wakes, and then I run GB, I get around 42000. I have had this for a long time. It is the same whether I boot via Clover or OC so it is not boot loader related. It does not appear to be WEG related either since I get the same result if I disable WEG.

 

Interestingly, LUXMARK scores are NOT affected - that is, same score before and after a system wake.

 

Many folks have this problem and some do not and so the cause is obvious.

 

Today I did update my SMBIOS to iMac 20,2 for a tryout and figured out that it does fix the Geekbench score for my RX 590 after boot. It always has been 20.000 points less than after wake from sleep. I have learned that switching to iMac 20,2, and enable the internal graphics with Multi Monitor enabled in the BIOS, did fix that issue with Geekbench for the Graphics performance. Now I always have a ~ 55.000 Metal score after boot. Before that has been around 37.000. I have also tried to see if enabling iGfx HD630 was cause, but on iMac Pro 1,1 the issue kept the same.

But now, because of the switch to iMac 20,2 SMBIOS, I do have to use CPUFriend.kext for having an CPU idle frequency at 800MHz, as it should be. It's 1300MHz idle without using CPUFriend. What is the idle frequency on a real iMac? All SMBIOS after iMac1,1 have the same issue with CPU idle @ 1,2 or 1,3 GHz. May be all newer iMac's are using this as default? Could somebody please shed some light om this?

Edited February 13, 2021 by Planet X

[MacNB]

On 2/7/2021 at 2:28 PM, Planet X said:

 

Today I did update my SMBIOS to iMac 20,2 for a tryout and figured out that it does fix the Geekbench score for my RX 590 after boot. It always has been 20.000 points less than after wake from sleep. I have learned that switching to iMac 20,2, and enable the internal graphics with Multi Monitor enabled in the BIOS, did fix that issue with Geekbench for the Graphics performance. Now I always have a ~ 55.000 Metal score after boot. Before that has been around 37.000. I have also tried to see if enabling iGfx HD630 was cause, but on iMac Pro 1,1 the issue kept the same.

But now, because of the switch to iMac 20,2 SMBIOS, I do have to use CPUFriend.kext for having an CPU idle frequency at 800MHz, as it should be. It's 1300MHz idle without using CPUFriend. What is the idle frequency on a real iMac? All SMBIOS after iMac1,1 have the same issue with CPU idle @ 1,2 or 1,3 GHz. May be all newer iMac's are using this as default? Could somebody please shed some light om this?

 

Thanks for the suggestion. My BIOS does not have a Multi Monitor option.

But I found why it is happening. 

 

When I have just ONE monitor attached to Display Port 1, the RX580's VRAM frequency varies from 300Mhz to 2000Mhz and Geekbench 5 give a low GPU score.

As soon as I hot plug a SECOND monitor to Display Port 2, the VRAM frequency INCREASES to 2000Mhz and STAYS FIXED at the that frequency (as long there's a second monitor).

Now, straight after booting GB5 gives a score of ~ 49K for OpenCL and ~ 53K for Metal.

As soon as I unplug ANY monitor (leaving just ONE), the GB5 goes DOWN to ~ 29K and the GPU VRAM frequency varies up & down again (between 300 & 2000Mhz).

 

With the VRAM Frequency fixed at 2000Mhz, the power consumption increases and the Fan comes ON more regularly.

 

When booting Windows 10 with TWO monitors, the VRAM frequency is also FIXED @ 2000Mhz too. 

As as soon as one monitor is removed, the VRAM frequency reduces to 300Mhz. HOWEVER, when GB5 on Windows runs, the VRAM frequency shoots up to 2000Mhz for the duration of the run and gives me a score of ~48K. 

 

Apparently, it is a known "feature" on AMD GPU's !!

If you google "AMD VRAM fixed to 2000Mhz", it will return a some results of people observing this behaviour (on Windows).

 

So, it appears to be a BUG in Apple's macOS AMD GPU driver that somehow does NOT scale the VRAM frequency UP when demanded by load while ONLY one monitor is used. 

Some kind of "setting" is not being applied to the GPU on boot to enable VRAM Frequency scaling BUT it is being applied AFTER waking from Sleep. 

[Planet X]

Very nice find. But Multi Monitor, it's so called on Asus boards, means to enable the internal Intel graphics from your CPU too in BIOS. Leave it at 64MB. That helps, but only on iMac 19,x or 20,x SMBIOS.

Edit: I don't mean a 2nd monitor.

Edited March 10, 2021 by Planet X
not 2nd monitor!

[yapan4]

18 hours ago, MacNB said:

 

Thanks for the suggestion. My BIOS does not have a Multi Monitor option.

But I found why it is happening. 

 

When I have just ONE monitor attached to Display Port 1, the RX580's VRAM frequency varies from 300Mhz to 2000Mhz and Geekbench 5 give a low GPU score.

As soon as I hot plug a SECOND monitor to Display Port 2, the VRAM frequency INCREASES to 2000Mhz and STAYS FIXED at the that frequency (as long there's a second monitor).

Now, straight after booting GB5 gives a score of ~ 49K for OpenCL and ~ 53K for Metal.

As soon as I unplug ANY monitor (leaving just ONE), the GB5 goes DOWN to ~ 29K and the GPU VRAM frequency varies up & down again (between 300 & 2000Mhz).

 

With the VRAM Frequency fixed at 2000Mhz, the power consumption increases and the Fan comes ON more regularly.

 

When booting Windows 10 with TWO monitors, the VRAM frequency is also FIXED @ 2000Mhz too. 

As as soon as one monitor is removed, the VRAM frequency reduces to 300Mhz. HOWEVER, when GB5 on Windows runs, the VRAM frequency shoots up to 2000Mhz for the duration of the run and gives me a score of ~48K. 

 

Apparently, it is a known "feature" on AMD GPU's !!

If you google "AMD VRAM fixed to 2000Mhz", it will return a some results of people observing this behaviour (on Windows).

 

So, it appears to be a BUG in Apple's macOS AMD GPU driver that somehow does NOT scale the VRAM frequency UP when demanded by load while ONLY one monitor is used. 

Some kind of "setting" is not being applied to the GPU on boot to enable VRAM Frequency scaling BUT it is being applied AFTER waking from Sleep. 

I tried to find a solution to this problem in https://www.insanelymac.com/forum/topic/346539-gpuvmem-stuck-to-full-speed-when-2nd-display-connected/ topic. No results.
Maybe someone will clarify this issue here....

 

Edited March 4, 2021 by yapan4

[LockDown]

In Whatevergreen documentation, it says

agdpmod=vit9696 disables check for board-id (or add agdpmod property to external GPU).

I know its a boot flag and it works for me.

I curious about this though:

(or add agdpmod property to external GPU).

Does it mean i can add it under Properties instead of a boot flag?

[theroadw]

@vit9696

 

Sorry to bug you with this as I know your time is precious, but I need some help with a WX-4170 MXM card booting in Catalina.

My laptop is a ZBook G5 and it has an UHD630 as well as the AMD DGPU. There is a Mux chip in the laptop and I can successfully boot and use either one or both GPU's in High Sierra and Mojave.

However all my previous tests with Catalina resulted in a boot stalled at the Apple logo just as the verbose boot turns off and the GPU drivers kick in.

 

After days of many different attempts to make it work, today I managed to edit the connectors of a ROM for another card (WX-4150) and I managed to boot and get Catalina working.

This comes at a price though, as the performance of this ROM is way below what the 4170 can do, and the ZBook appears to have a whitelist of some sorts that prevents the card from working in discrete mode using this ROM.

 

Now my test proved the hardware is compatible and can work, so it's just a matter of patching the driver with Whatevergreen so it will load with my card, as I suspect the problem probably lies in some variable that the real 4170 ROM is missing or different from what the driver wants.

I read of a similar problem with NAVI cards where the driver has a bug and looks for AMD instead of ATI or something like that, and it causes what seems to be the same stall as my laptop.

I've tested different SMBIOS's device-id's etc... all without any change, but the different Rom "worked" without changing/faking anything else.

 

How can I get all the information the drivers ask from the card so I can compare both and find which piece is missing?

 

Thanks!

Edited May 10, 2021 by theroadw

[Eniac]

I have a problems whit Whatevergreen version 1.5.2 . 

With this version i don't have hardware acceleration though  the GPU is recognized exactly ,  on OC 0.7.0 - 0.7.1 and 0.7.2

My graphics card is R9 270 2gb 

 

With whatevergreen versione 1.5.1 it's work fine.

Edited August 3, 2021 by Eniac
add image

[MacKonsti]

Hello everyone, I decided to write here before raising an issue on the tracker, for WhateverGreen.

 

My Lenovo IdeaPad is using Intel UHD Graphics 620 [8086:3ea0] and when enabling msgbuf=1048576 I see a flood of errors related to IGFB, which could explain the pause of the boot process (the line under the Apple logo stops for some seconds completely).

 

An extract of the repeated errors is the following:

 

localhost kernel[0]: (AppleIntelCFLGraphicsFramebuffer) [IGFB][ERROR  ] [AGDC] Failed with status -536870212 for stream access aux
localhost kernel[0]: (AppleIntelCFLGraphicsFramebuffer) [IGFB][ERROR  ] FB2: WriteAUX Timeout for addr 600 with Status = 0xe00002d6
localhost kernel[0]: (AppleIntelCFLGraphicsFramebuffer) [IGFB][ERROR  ] FB2: ReadAUX Timeout for addr 600 with Status = 0xe00002d6
localhost kernel[0]: (AppleIntelCFLGraphicsFramebuffer) [IGFB][ERROR  ] FB2: WriteAUX Timeout for addr 600 with Status = 0xe00002d6

 

Could it be an issue with my carefully set properties for IGPU? A conflict? A lack of support? Can anyone kindly check the log and tell me if there is some idea on the cause, e.g. incompatibility?

 

I am injecting the following device properties in my config (using latest kexts and OpenCore 0.7.4) with the following boot-arguments igfxonln=1 -igfxblr -igfxnohdmi msgbuf=1048576

 

			<key>PciRoot(0x0)/Pci(0x2,0x0)</key>
			<dict>
				<key>AAPL,GfxYTile</key>
				<data>AQAAAA==</data>
				<key>AAPL,ig-platform-id</key>
				<data>AACbPg==</data>
				<key>device-id</key>
				<data>mz4AAA==</data>
				<key>framebuffer-patch-enable</key>
				<data>AQAAAA==</data>
				<key>framebuffer-fbmem</key>
				<data>AACQAA==</data>
				<key>framebuffer-stolenmem</key>
				<data>AAAwAQ==</data>
				<key>enable-hdmi20</key>
				<data>AQAAAA==</data>
				<key>enable-lspcon-support</key>
				<data>AQAAAA==</data>
				<key>framebuffer-con0-enable</key>
				<data>AQAAAA==</data>
				<key>framebuffer-con0-alldata</key>
				<data>AAAIAAIAAACYAAAA</data>
				<key>framebuffer-con1-enable</key>
				<data>AQAAAA==</data>
				<key>framebuffer-con1-alldata</key>
				<data>AQEJAAAIAADHAQAA</data>
				<key>framebuffer-con1-has-lspcon</key>
				<data>AQAAAA==</data>
				<key>hda-gfx</key>
				<string>onboard-1</string>
				<key>igfxfw</key>
				<data>AgAAAA==</data>
				<key>model</key>
				<string>Intel UHD Graphics 620</string>
			</dict>

Does anyone have any idea or came across this too? Do you understand what these errors mean? Anything more I can provide as logs?

 

In order to detect the HDMI port (seems only via hot-plugging though) I am using the following port patching:

[0]  busId: 0x00,  pipe: 8,  type: 0x00000002,  flags: 0x00000098,  Connector LVDS
framebuffer-con0-alldata  →  00000800 02000000 98000000  →  AAAIAAIAAACYAAAA

[1]  busId: 0x05,  pipe: 9,  type: 0x00000800,  flags: 0x000001C7,  Connector HDMI
framebuffer-con1-alldata  →  01010900 00080000 c7010000  →  AQEJAAAIAADHAQAA

Thank you in advance.

IGFB Errors.zip

Edited October 24, 2021 by MacKonsti

[soumaxetuirk]

IG PLATFORM ID : 0x59168086

GPU Name : Intel HD Graphics 620 (Intel i5-7200U)

My laptop display is showing as bulletin display but no brightness slider.

 

I am using OpenCore with PNLF SSDT. Ioreg shows appleBacklight driver loaded under PNLF device.

can anyone suggest to fix brightness .thank you .

Opencore config is attached.config.plist

[MacKonsti]

Hello everyone, I'd like to ask a question that's in my mind for many months and I am hoping to get a reply from the developers of WhateverGreen...

 

On older platforms such as Broadwell and Haswell especially not chipsets directly supported by the Apple IGPU kexts, whenever the screensaver runs (either in Preview or full-screen mode) the IGPU usage is very high, according to Intel Power Gadget and CPU temperature raises without a real need... I am talking about simple animated screensaver like Flurry Arabesque etc. This is kinda forcing me to set the simplest screen-saver of all, the "Message" one.

 

In contrast to my Intel NUC Coffee Lake [8086:3ea5] (even Comet Lake) there is no such phenomenon i.e. I leave the hackintosh run into animated screen-saver before the screen goes to sleep. IGPU usage is not high and CPU temperatures remain normal.

 

Any idea why this would be happening? Coud it be related to SMBIOS too or just the Platform ID forced as device property?

 

I am referring to my Intel NUC D54250WYKH with Intel HD Graphics 5000 (Haswell) graphics [8086:0a26] and Beelink L55 with (Broadwell) Intel HD Graphics 5500 [8086:1616] chipsets.

I leave WhateverGreen decide the best match, after having tried to "force" Platform ID as 0x0D220003 (for Haswell) and 0x16160002 (for Broadwell) respectively (so I get some decent acceleration). The latter seems to be mobile chipset.

 

Your feedback welcome, let me know if more info is needed.... Thanks!

Edited November 7, 2021 by MacKonsti

[deeveedee]

Developers,  I was looking in ./WhateverGreen/kern_igfx.hpp at the definition of FirmwareLoad to get a better understanding of boot-arg igfxfw:

        enum FirmwareLoad {
                FW_AUTO    = -1 /* Use as is for Apple, disable for others */,
                FW_DISABLE = 0, /* Use host scheduler without GuC */
                FW_GENERIC = 1, /* Use reference scheduler with GuC */
                FW_APPLE   = 2, /* Use Apple GuC scheduler */
        };

I've only seen documentation that references the use of boot-arg igfxfw=2 (igfxfw=FW_APPLE).  Is there any reason that we would want to use any of the other FirmwareLoad enum values with Big Sur or Monterey?

Edited December 10, 2021 by tonyx86

[Allan]

@tonyx86 your post is more WEG-related questions, not OpenCore.

So I've moved your topic to here 😄

[deeveedee]

@Allan I should have realized sooner that I was asking about WEG boot-args.  Feel free to move my other questions/answers, too.  I am noticing iGPU performance improvements on my HackMini8,1 (CFL UHD630) with boot-arg igfxfw=2 and my HackBookPro15,2 (KBL-R UHD620) with boot-arg  igfxrpsc=1.

Edited December 10, 2021 by tonyx86

[matgeo]

16 hours ago, tonyx86 said:

@Allan I should have realized sooner that I was asking about WEG boot-args.  Feel free to move my other questions/answers, too.  I am noticing iGPU performance improvements on my HackMini8,1 (CFL UHD630) with boot-arg igfxfw=2 and my HackBookPro15,2 (KBL-R UHD620) with boot-arg  igfxrpsc=1.

Don't you have the problem with igfxrpsc=1 , where igpu frequency stays on max after sleep?

[deeveedee]

@matgeo No.  Sleep/Wake still works perfectly with igfxrpsc=1 (no different than without).

 

before sleep with igfxrpsc=1

Spoiler

 

after sleep with igfxrpsc=1

Spoiler

 

Edited December 11, 2021 by tonyx86

[matgeo]

3 hours ago, tonyx86 said:

@matgeo No.  Sleep/Wake still works perfectly with igfxrpsc=1 (no different than without).

 

before sleep with igfxrpsc=1

  Hide contents

 

 

after sleep with igfxrpsc=1

  Hide contents

 

Is it possible to send me your EFI?

 

Edited December 11, 2021 by matgeo

[deeveedee]

@matgeo I'm not ready to share my laptop EFI, but I'm maintaining my HackMini8,1 thread with my methodology here.  If you create a thread for your rig and post your work, ping me and I'd be happy to take a look.

[matgeo]

@tonyx86, here is my repo : https://github.com/Mateo1234454545/DELL-5570-Hackintosh

[deeveedee]

@matgeo In your repo comments: "note 1 : I use an older version of whatevergreen.kext , whick works on Big Sur but doesn't have the issue where aster waking from sleep gpu hangs at max frequency."

 

Does this comment mean that you can "fix" your gfxrpsc=1 max GPU frequency problem after sleep by reverting to an olderer version of WEG?

[matgeo]

@tonyx86, no . This comment doesn't affect  igfxrpsc=1 (and has been solved in later versions)

[deeveedee]

@matgeo I started looking at your EFI.  It looks like you have done a very thorough job.  I recognize some of the same patches that I've implemented on my HackBookPro15,2 (HP Envy KBL-R).  At first glance, the only obvious difference to me is that I did not attempt to enable I2CHID (only needed for my touchscreen and I don't need it).  Also, I don't need ECEnabler.kext, but that looks like some very cool work by @1Revenger1.  

 

I will continue to review when I get time - just wanted you to know that I didn't forget.

 

EDIT: There is one difference that I should have noticed immediately: You are not enabling WEG framebuffer patching with framebuffer-patch-enable (since apparently you don't need it).  I am spoofing device-id = 0x5916 as you are, but I also enable framebuffer patching with framebuffer-patch-enable=1 and AAPL,ig-platform-id 0x591B0000.  I'm just pointing that out, since I have never used WhateverGreen.kext without enabling framebuffer patching.

 

I'll continue to review when I get time.

Edited December 14, 2021 by tonyx86

[matgeo]

@tonyx86, thank you for your time.

I have tried  framebuffer-patch-enable , with no diffirence at all.

I'm waiting for you be ready to upload your laptop efi.

[deeveedee]

For those using DP->DVI video adapters with their iGPUs (CFL UHD630 in my case), in order to run Monterey 12.1, I needed to change my patched connector types from dual-link DVI (<04000000>) to HDMI (<00080000>).  My rig would boot 12.1 to black screens with connector types set to DVI.  When I first started patching the HP EliteDesk 800 G4 Mini (and now G5 Mini), guides at the time indicated that video converters/adapters were not supported.  Through brute force experimentation, I found only connector type dual-link DVI would enable my rig's displays with DP->DVI adapters.  I continued to use patched connector type DVI for Catalina and Big Sur.  Now, with Monterey, I need to switch to HDMI connector types to enable my displays with DP->DVI adapters.

 

My rig is documented here.  Note that the only change I need to make for Monterey 12.1 is the patched connector type.  The rest of my posted OC 0.7.6 EFI remains unchanged.

 

EDIT: Note that in order to use DP->DVI video adapters with my CFL UHD630 iGPU, I must use a "mobile" framebuffer (I'm using AAPL,ig-platform-id 0x3E920000).  A "desktop" framebuffer (e.g., AAPL,ig-platform-id 0x3E9B0007) with DP->DVI video adapters results in booting to black screens on my rig.  This has been true for as long as I have been patching my HP EliteDesk 800 G4/G5 Minis.

 

About This Mac: Monterey 12.1

Spoiler

 

Edited December 15, 2021 by tonyx86