AMD/Nvidia Primary Display with AirPlay Mirroring

[toleda]

help me to solve the full speed lock on IGPU?

IOReg shows GFX0/max power state; consider changing GFX0 to IGPU. Clover can do that, however the current ssdt-1 would no longer work, Clover/9series/ssdt/HDMI audio is the preferred solution, see HDMI Audio AppleHDA. Edit the framebuffer as noted in Post #1 of this thread.

No idea if HD4600 PM is related to the Safari crashes.

 

[toleda]

the video stream for 2 monitor freezes

What is 2 monitor?

 

[oculto]

IOReg shows GFX0/max power state; consider changing GFX0 to IGPU. Clover can do that, however the current ssdt-1 would no longer work, Clover/9series/ssdt/HDMI audio is the preferred solution, see HDMI Audio AppleHDA. Edit the framebuffer as noted in Post #1 of this thread.

No idea if HD4600 PM is related to the Safari crashes.

I use clover to rename GFX0 to IGPU and adjust my ssdt-1, everything is ok. it seems it fixes the safari crashes on videos (at least it seem until now) but the full speed on IGPU isnt fix yet.

i dont know if it's related to simbios 14,2 or to the framebufer 04120004 but it all set correctly, and the power management still on maximum state. I have the same problem on other machine, a H87M-D3H and i have it since mavericks.

intel power gadgets shows always at full speed.

 

[toleda]

i dont know if it's related to simbios 14,2 or to the framebufer 04120004

Without an IOReg, no idea if anything changed. The configuration is native imac14,2.

 

[oculto]

see ioreg attached.
View attachment oculto-12-1-2015.ioreg.zip

UPDATE: i think the problem is the 0x04120004 framebuffer. see the scenarios bellow:

1) clover ig-platform-id = 04120004 (post #1 this guide), airplay NOT work, PM work and clover inject <c4 dd 3e 00> as ig-platform-id

2) clover ig-platform-id = 0x04120004 (note 0x in the beginning), airplay WORK, PM MAX SPEED and clover inject <04 00 12 04> as ig-platform-id

3) clover ig-platform-id = 0x0D220003 (note 0x in the beginning), airplay WORK, PM WORK and clover inject <03 00 22 0d> as ig-platform-id

so, for some reason, on Z97X-SLI, 0x0D220003 works without conecting a display, with airplay and power management. just follow your guide (inject intel, ad ig-plataform-id, etc) but using 0x0D220003 instead 04120004 for HD4600.

see the new ioreg here... View attachment oculto-0x0D220003.ioreg.zip


UPDATE2: this setup works, but my performance on FCPX worst a bit, i believe the second 280x is not render anymore. BruceX benchmark from 26s to 35s

I remember someone said IGPU + 2x GPUs could not be the best for FCPX, just 2x GPU is better.

UPDATE3: in fact, this setup is not good for someone who has 2x GPU video cards...
1) using 0x04120004 = BruceX = 25s
2) using 0x0D220003 = BruceX = 35s
3) IGPU Disable on bios = BruceX = 26s (could be 25s, i measure with stopwatch).
4) i remember using only one 280x = 48s approximately

So, at least for now, enabling airplay with 0x0D220003 is not good for who has more than one videocard.

 

[toleda]

on Z97X-SLI, 0x0D220003 works without conecting a display, with airplay and power management. just follow your guide (inject intel, ad ig-plataform-id, etc) but using 0x0D220003 instead 04120004 for HD4600.

Case 2, IOReg, no change from previous IOReg (i.e., IGPU did not solve problem)
Case 3, IGPU/AGPM loads correctly. For most systems, boot hangs; not recommended

 

[oculto]

Case 2, IOReg, no change from previous IOReg (i.e., IGPU did not solve problem)
Case 3, IGPU/AGPM loads correctly. For most systems, boot hangs; not recommended

No problem on boot, see the bios, clover menu, apple logo, all works fine to me, but as i said, it is not the best case when user have twin graphics cards to work on final cut pro x.

and rename GFX0 -> IGPU solve my problem on Safari. i had several errors on safari when HTML5 need video conversion, and I suspect it was because the apple framework didnt find intel qsync on the right place. rename GFX0 to IGPU solve this problem, finally. I don't use flash or silverlight, only pure safari, thats why this issue was certainly related to hardware address.

but please, verify my notes on clover when inject ig-platform-id. without 0x at the beginning, clover didn't inject the value properly. maybe it is the case to warn the users on post #1

 

[toleda]

but please, verify my notes on clover when inject ig-platform-id. without 0x at the beginning, clover didn't inject the value properly. maybe it is the case to warn the users on post #1

Post #1 config.plist values have been verified. Missed a step:
1. Bootloader Settings

2. Clover/config.plist​

1. Graphics/Inject/Intel/YES​

Setting config.plist/Graphics/ig-platform-id with Graphics/Inject/Intel/NO is not recommended; the reason Case 1 did not work.

With config/Graphics/Inject/Intel/YES, testing shows Clover handles 0x04120004 and 04120004 the same, both work. That is a problem specific to Clover. The AppleIntelFramebufferAzul binary uses little endian byte ordering meaning 04120004 is represented as 0x04001204 (see Post #1 dsdt edits). However, config.plist/Graphics/ig-platform-id/0x04001204 does not work. This behavior is unexpected.

 

[oculto]

in all situations i use on clover INJECT INTEL = TRUE

1. Bootloader Settings2. Clover/config.plist
1. Graphics/Inject/Intel/YES (i did it every time)

and i confirm with 04120004 (without 0x at beginning) clover did not inject ig-platform-id correctly.

if i send to you a config.plist with inject = NO it is my mistake. all tests will be done with inject YES.

 

[toleda]

if i send to you a config.plist with inject = NO it is my mistake. all tests will be done with inject YES.

My assessment is based on the files attached to Post #209 and the IORegs posted subsequently.
Looks like ssdt-1 loaded during your Clover/Intel/injection experiments; expect the unexpected.