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

MP 1,1-5,1 Activate AMD hardware acceleration

Status
The first post of this thread is a WikiPost and can be edited by anyone with the appropiate permissions. Your edits will be public.
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
Code: 
@@ -207,10 +206,20 @@ void WEG::processKernel(KernelPatcher &patcher) {
                    }
                    if (agdp) {
                        processGraphicsPolicyStr(agdp);
                        break;
                        return true;
                    }
                }

                return false;
            };

            size_t extNum = devInfo->videoExternal.size();
            for (size_t i = 0; i < extNum; i++) {
                if (getAgpdMod(devInfo->videoExternal[i].video))
                    break;
            }
            if (devInfo->videoBuiltin != nullptr && graphicsDisplayPolicyMod == AGDP_DETECT) /* Default detect only */
                getAgpdMod(devInfo->videoBuiltin);
        }

        // Do not inject properties unless non-Apple
This is probably due to the complain that OC fiddles with the private NVRAM.
We asked for it :)
 
h9826790

h9826790

macrumors P6
Original poster
Apr 3, 2014
16,656
8,587
Hong Kong
startergo said:
Code: 
@@ -207,10 +206,20 @@ void WEG::processKernel(KernelPatcher &patcher) {
                    }
                    if (agdp) {
                        processGraphicsPolicyStr(agdp);
                        break;
                        return true;
                    }
                }

                return false;
            };

            size_t extNum = devInfo->videoExternal.size();
            for (size_t i = 0; i < extNum; i++) {
                if (getAgpdMod(devInfo->videoExternal[i].video))
                    break;
            }
            if (devInfo->videoBuiltin != nullptr && graphicsDisplayPolicyMod == AGDP_DETECT) /* Default detect only */
                getAgpdMod(devInfo->videoBuiltin);
        }

        // Do not inject properties unless non-Apple
This is probably due to the complain that OC fiddles with the private NVRAM.
We asked for it :)
Click to expand...

Expected, and if they are responsible developers, and both of they and us have no way to tell why some extra binary blobs show up in private NVRAM when boot via OC. All they can do at this moment is just post a warning.

Anyway, I think it's just a coincident that they decided to remove the shikigva option, but not because we "reminded" them to do that.
 
C

cdf

macrumors 68020
Jul 27, 2012
2,256
2,583
My understanding is that it is the legacy -shikigva alias that has been dropped, shikigva=N is still there.
 
  • Like
Reactions: h9826790
L

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
Ludacrisvp said:
... Using 'lsof' I can also see that on the 1,1 VTEncoder is using it and on the 3,1 it is not. VideoProc and Handbrake *try* to use it but they fail. ...
Click to expand...
I suppose I could have / should have mentioned this was 10.14.6 and was using my 1.3.6 WEG version (PXS1 to GFX0 rename does work)
I couldn't tell any differences between 1.3.5 and 1.3.6 but I don't have a ton of testing time.
 
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
Ludacrisvp said:
I suppose I could have / should have mentioned this was 10.14.6 and was using my 1.3.6 WEG version (PXS1 to GFX0 rename does work)
I couldn't tell any differences between 1.3.5 and 1.3.6 but I don't have a ton of testing time.
Click to expand...
Just to clarify:
With this WEG there is no need for manual rename like this:
Code: 
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>ACPI</key>
    <dict>
        <key>Patch</key>
        <array>
            <dict>
                <key>Comment</key>
                <string>PXS1 to GFX0</string>
                <key>Count</key>
                <integer>0</integer>
                <key>Enabled</key>
                <true/>
                <key>Find</key>
                <data>
                UFhTMQ==
                </data>
                <key>Limit</key>
                <integer>0</integer>
                <key>Mask</key>
                <data>
                </data>
                <key>OemTableId</key>
                <data>
                </data>
                <key>Replace</key>
                <data>
                R0ZYMA==
Or do we still have to patch it?
 
L

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
startergo said:
Just to clarify:
With this WEG there is no need for manual rename like this:
Code: 
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
    <key>ACPI</key>
    <dict>
        <key>Patch</key>
        <array>
            <dict>
                <key>Comment</key>
                <string>PXS1 to GFX0</string>
                <key>Count</key>
                <integer>0</integer>
                <key>Enabled</key>
                <true/>
                <key>Find</key>
                <data>
                UFhTMQ==
                </data>
                <key>Limit</key>
                <integer>0</integer>
                <key>Mask</key>
                <data>
                </data>
                <key>OemTableId</key>
                <data>
                </data>
                <key>Replace</key>
                <data>
                R0ZYMA==
Or do we still have to patch it?
Click to expand...
No need to patch if using 1.3.5+ of WEG if you use the “-wegtree” boot arg. It renamed my device in the 3,1 with no issue.
 
  • Like
Reactions: h9826790
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
Ludacrisvp said:
No need to patch if using 1.3.5+ of WEG if you use the “-wegtree” boot arg. It renamed my device in the 3,1 with no issue.
Click to expand...
Did they reflect the change in the readme of shiki cause I don't see a change in usage?
Where is the -wegtree readme?
 
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
Ludacrisvp said:
No need to patch if using 1.3.5+ of WEG if you use the “-wegtree” boot arg. It renamed my device in the 3,1 with no issue.
Click to expand...
Using a copy of your WEG with -wegtree I don't see renamed PXS1:
Code: 
<string>-v  debug=0x144 -no_compat_check agdpmode=pikera -wegtree shiki-id=Mac-7BA5B2D9E42DDD94</string>
 
L

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
startergo said:
Using a copy of your WEG with -wegtree I don't see renamed PXS1:
Code: 
<string>-v  debug=0x144 -no_compat_check agdpmode=pikera -wegtree shiki-id=Mac-7BA5B2D9E42DDD94</string>
Click to expand...

Regarding '-wegtree' not sure if it made it to any readme files yet, considering 1.3.5 isn't officially released yet I'm sure documentation will be updated later. It was mentioned in the commit.

Are you sure it loaded the new version?

Code: 
ludacrisvp-macpro-osx:~ ludacrisvp$ kextstat |grep Wha
   18    0 0xffffff7f823d4000 0x83000    0x83000    as.vit9696.WhateverGreen (1.3.6) C77F4534-7B0D-312D-85A2-561BB38992F5 <17 13 8 6 5 3 2 1>

You can see here it is renamed to GFX0 on my 3,1 via WEG ... I still can't get OC to boot on my 3,1 (haven't had a lot of time to mess with that though)

Code: 
ludacrisvp-macpro-osx:~ ludacrisvp$ ioreg -l|grep GFX0
    | |   |   +-o GFX0@0  <class IOPCIDevice, id 0x1000001b6, registered, matched, active, busy 0 (1185 ms), retain 32>
    | |   |   | |   "compatible" = <"pci1da2,e387","pci1002,67df","pciclass,030000","PXS1","GFX0">
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/NRP5@5/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/NRP5@5/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/NRP5@5/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |       "IODisplayPrefsKey" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/NRP5@5/IOPP/GFX0@0/ATY,AMD,RadeonFramebuffer@2/AMDFramebufferVIB/display0/AppleDisplay-10ac-a098"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/NRP5@5/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/NRP5@5/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | | | |   "Config2" = {"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |   | | | |   "Config4" = {"IGPU"={"unload"=Yes},"display"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |   | | | |   "Config3" = {"IGPU"={"unload"=Yes},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
ludacrisvp-macpro-osx:~ ludacrisvp$ nvram -p|grep arg
boot-args	shikigva=48 shiki-id=Mac-7BA5B2D9E42DDD94 -v debug=0x144 -wegtree
 
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
Ludacrisvp said:
Are you sure it loaded the new version?
Click to expand...
That is funny. It turns out I had an older version in /L/E 128+135 Lilu. So all these tests I did with the modified WEG inside the LILU. I need to reevaluate the results now that I removed that kext
[automerge]1573754837[/automerge]
Now -wegtree works for renaming:
Code: 
-----------------------------------------------------------------------
System Info
-----------------------------------------------------------------------
Host                           G5.local
OS                             macOS Mojave Version 10.14.6 (Build 18G2007)
Kernel                         Darwin 18.7.0 x86_64
RAM                            128.00 GB
Model Identifier               MacPro5,1
CPU                            Intel(R) Xeon(R) CPU           X5690  @ 3.47GHz
Intel Generation               ???
Platform ID                    0x00000000
Board ID                       Mac-F221BEC8
FW Version                     144.0.0.0.0
Serial Number               
Hardware UUID               
System ID                   
ROM                         
Board Serial Number         
VDA Decoder                    Fully Supported
-----------------------------------------------------------------------
Serial Info
-----------------------------------------------------------------------
Country                        Unknown
Year                           2011
Week                           12.17.2011-12.23.2011
Line                           707 (copy 1)
Model                          MacPro5,1
Valid                          Possibly
-----------------------------------------------------------------------
GFX1
-----------------------------------------------------------------------
GPU Name                       GK110 [GeForce GTX TITAN]
GPU Device ID                  0x100510DE
Quartz Extreme (QE/CI)         Yes
Metal Supported                Yes
Metal Device Name              NVIDIA GeForce GTX TITAN
Metal Default Device           Yes
Metal Low Power                No
Metal Headless                 No
-----------------------------------------------------------------------
GFX0
-----------------------------------------------------------------------
GPU Name                       Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]
GPU Device ID                  0x67DF1002
-----------------------------------------------------------------------
Now the display@ is properly replaced by GFX1 even though it is in the expansion not directly in slot 2
Code: 
kextstat |grep Wha
   38    0 0xffffff7f85304000 0x86000    0x86000    as.vit9696.WhateverGreen (1.3.6) C77F4534-7B0D-312D-85A2-561BB38992F5 <37 13 8 6 5 3 2 1>
G5% kextstat |grep Lilu
   37    1 0xffffff7f852db000 0x29000    0x29000    as.vit9696.Lilu (1.3.9) 6E2022BA-255E-38F8-975E-46DCD3B040F9 <8 6 5 3 2 1>
G5% ioreg -l|grep GFX0
    | |   |   +-o GFX0@0  <class IOPCIDevice, id 0x1000001c4, registered, matched, active, busy 0 (1057 ms), retain 49>
    | |   |   | |   "compatible" = <"pci1028,1701","pci1002,67df","pciclass,030000","PXS1","GFX0">
    | |   |   | |   "attached-gpu-control-path" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/ATY,AMD,RadeonFramebuffer@0/AMDFramebufferVIB/AGPM"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | | | |   "Config2" = {"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |   | | | |   "Config4" = {"IGPU"={"unload"=Yes},"display"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |   | | | |   "Config3" = {"IGPU"={"unload"=Yes},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |       |   | | | |   "Config2" = {"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |       |   | | | |   "Config4" = {"IGPU"={"unload"=Yes},"display"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |       |   | | | |   "Config3" = {"IGPU"={"unload"=Yes},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
G5% ioreg -l|grep GFX1
    | |   |   | | | |   "Config1" = {"GFX1"={"unload"=Yes}}
    | |   |       |   +-o GFX1@0  <class IOPCIDevice, id 0x100000209, registered, matched, active, busy 0 (506 ms), retain 47>
    | |   |       |   | |   "IOReportLegend" = ({"IOReportGroupName"="Interrupt Statistics (by index)","IOReportChannels"=((5291294645182070784,4295098369,"               First Level Interrupt Handler Count"),(5291294645182070785,4295098369,"              Second Level Interrupt Handler Count"),(5291294645182070786,4295098369,"        First Level Interrupt Handler Time (MATUs)"),(5291294645182070787,4295098369,"   Second Level Interrupt Handler CPU Time (MATUs)"),(5291294645182070788,4295098369,"Second Level Interrupt Handler System Time (MATUs)")),"IOReportChannelInfo"={"IOReportChannelUnit"=0},"IOReportSubGroupName"="GFX1 1"})
    | |   |       |   | |   "attached-gpu-control-path" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU1@7/IOPP/PXS2@0/IOPP/pci-bridge@8/IOPP/GFX1@0/NVDA,Display-A@0/NVDA/AGPM"
    | |   |       |   | |   "compatible" = <"pci3842,2790","pci10de,1005","pciclass,030000","GFX1">
    | |   |       |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU1@7/IOPP/PXS2@0/IOPP/pci-bridge@8/IOPP/GFX1@0/nvAccelerator"
    | |   |       |   | |   |       "IODisplayPrefsKey" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU1@7/IOPP/PXS2@0/IOPP/pci-bridge@8/IOPP/GFX1@0/NVDA,Display-A@0/NVDA/display0/AppleDisplay-4c2d-68b"
    | |   |       |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU1@7/IOPP/PXS2@0/IOPP/pci-bridge@8/IOPP/GFX1@0/nvAccelerator"
    | |   |       |   | |   |       "IODisplayPrefsKey" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU1@7/IOPP/PXS2@0/IOPP/pci-bridge@8/IOPP/GFX1@0/NVDA,Display-B@1/NVDA/display0/AppleDisplay-4c2d-320"
    | |   |       |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU1@7/IOPP/PXS2@0/IOPP/pci-bridge@8/IOPP/GFX1@0/nvAccelerator"
    | |   |       |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU1@7/IOPP/PXS2@0/IOPP/pci-bridge@8/IOPP/GFX1@0/nvAccelerator"
    | |   |       |   | | | |   "Config1" = {"GFX1"={"unload"=Yes}}
 
Last edited:
  • Like
Reactions: h9826790
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
startergo said:
Code: 
G5% ioreg -l|grep GFX0
    | |   |   +-o GFX0@0  <class IOPCIDevice, id 0x1000001c4, registered, matched, active, busy 0 (1057 ms), retain 49>
    | |   |   | |   "compatible" = <"pci1028,1701","pci1002,67df","pciclass,030000","PXS1","GFX0">
    | |   |   | |   "attached-gpu-control-path" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/ATY,AMD,RadeonFramebuffer@0/AMDFramebufferVIB/AGPM"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | |   |   "IOAccelTypes" = "IOService:/AppleACPIPlatformExpert/PCI0@0/AppleACPIPCI/IOU0@3/IOPP/GFX0@0/AMDRadeonX4000_AMDEllesmereGraphicsAccelerator"
    | |   |   | | | |   "Config2" = {"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |   | | | |   "Config4" = {"IGPU"={"unload"=Yes},"display"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |   | | | |   "Config3" = {"IGPU"={"unload"=Yes},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |       |   | | | |   "Config2" = {"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |       |   | | | |   "Config4" = {"IGPU"={"unload"=Yes},"display"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
    | |   |       |   | | | |   "Config3" = {"IGPU"={"unload"=Yes},"GFX0"={"EDID"={"index"=0},"FeatureControl"=12,"unload"=No}}
Click to expand...

Somebody knows how to interpret these config 1,2,3,4,5? Are these configurable or are they automatic from the WEG?
 
L

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
startergo said:
That is funny. It turns out I had an older version in /L/E 128+135 Lilu. So all these tests I did with the modified WEG inside the LILU. I need to reevaluate the results now that I removed that kext
Click to expand...
I too was bitten by this (which is what made me think to suggest it) ... however it was the reverse for me, I had put the new kexts in /L/E but apparently i had an ancient copy of them in /S/L/E that was getting in the way.
 
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
Ludacrisvp said:
I too was bitten by this (which is what made me think to suggest it) ... however it was the reverse for me, I had put the new kexts in /L/E but apparently i had an ancient copy of them in /S/L/E that was getting in the way.
Click to expand...
I guess the priority is /S/L/E-->/L/E-->/Kexts in EFI folder
 
L

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
startergo said:
I guess the priority is /S/L/E-->/L/E-->/Kexts in EFI folder
Click to expand...
I think it is actually based on which folder that kextcache decides to hit first ... as i had previously updated to 1.3.4 WEG without an issue, only when I went to use 1.3.5 or higher did i have issues.
 
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
Ahaa changing only the board-id in SMBIOS and updating only SMBIOS (Automatic=No) Fools the VideoProc to believe the HEVC is enabled. Unfortunately this is only a cosmetic change as playing the HEVC with selected HEVC codec in VideoProc uses the CPU.
[automerge]1573773687[/automerge]
@Ludacrisvp by the way using your WEG the system froze. I had to ask somebody to reboot the system. I used latest Pavo's OCBuilder which builds the WEG under 1.3.5 Version. So far no freezes.
 
Last edited:
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
This is how VideoProc treats hardware acceleration:
1573777291605.png

Intel really?
 
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
tsialex said:
We should stop using VideoProc and go for ffmpeg + videotoolbox. I have noticed this weird behavior too and OC devs told about it too.
Click to expand...
How to test with ffmpeg (I was having issues with it)?
[automerge]1573777734[/automerge]
Just FYI changing only the board ID in SMBIOS keeps the cMP identity as far as the SMC is concerned. MFC sees the sensors and fans of the actual board. Not sure how facetime and imessage will be affected.
 
startergo

startergo

macrumors 603
Sep 20, 2018
5,020
2,282
ffplay is struggling to play HEVC (uses CPU). It plays fine H264
[automerge]1573786570[/automerge]
I might be chasing my tail with Kepler and ffmpeg:
1573786560994.png

developer.nvidia.com

FFmpeg

developer.nvidia.com developer.nvidia.com
 
Last edited:
L

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
startergo said:
Ahaa changing only the board-id in SMBIOS and updating only SMBIOS (Automatic=No) Fools the VideoProc to believe the HEVC is enabled. Unfortunately this is only a cosmetic change as playing the HEVC with selected HEVC codec in VideoProc uses the CPU.
[automerge]1573773687[/automerge]
@Ludacrisvp by the way using your WEG the system froze. I had to ask somebody to reboot the system. I used latest Pavo's OCBuilder which builds the WEG under 1.3.5 Version. So far no freezes.
Click to expand...
Perhaps the legacy flags they removed created some instability (and this is one of the reasons it was removed)?

Eitherway it really seems like all they do is allow you to use “-shikigva” instead of the proper “shikigva=value” same thing as the other one. So they really aren’t needed to be included. I Should have looked closer at the code change they did before undoing it.
[automerge]1573788098[/automerge]
startergo said:
ffplay is struggling to play HEVC (uses CPU). It plays fine H264
[automerge]1573786570[/automerge]
I might be chasing my tail with Kepler and ffmpeg:
View attachment 877173
developer.nvidia.com

FFmpeg

developer.nvidia.com developer.nvidia.com
Click to expand...
That’s because Kepler can’t do HEVC decode ... it doesn’t have the hardware.
My guess is that the graph from activity monitor that showed GPU usage inaccurately (usage for the wrong GPU) perhaps caused by limited support for Nvidia or mixing of GPU makers.
 
Last edited:
L

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
startergo said:
So only -shikigva? just like -wegtree ? No other options?
Click to expand...
Well they also removed “-shikifps” which was for FairPlay 1.0 support because it was integrated into the shikigva=number style.
in any case you should be using “shikigva=number” (where the number is actually a number ... the two legacy flags are part of a bit value that you add together to get the number used in the argument. Doesn’t seem to be documented well outside the source code.

Im going to remove the 1.3.6 kext from my post as it’s adding back legacy options we dont need and should not be using.
 
  • Like
Reactions: cdf and h9826790
tsialex

tsialex

Contributor
Jun 13, 2016
13,454
13,601
I found that there are FIVE different, (1) DSDT and (4) SSDT, tables that we need to do PXS* renames inside the BootROM.

I'm doing it initially for my own config and these are the tables and offsets for the PXS2->GFX0:

Code: 
5F1h - APPLEMacPro - 0xCAF
1AD29h - HdaSlt1 - 0x70A
1B505h - HdaSlt2 - 0x5F8
1BCE1h - HdSlt12 - 0x713
1C5E1h - NoHdau - 0x6C5
 
  • Like
Reactions: kohlson, cdf, octoviaa and 1 other person
L

Ludacrisvp

macrumors 6502a
May 14, 2008
797
363
Has anyone here run this before?
Code: 
/System/Library/Extensions/AppleGraphicsControl.kext/Contents/MacOS/AGDCDiagnose -a
It has a lot of info about your GPU(s) in it, temp, clock speeds, even the mac board ID under the "PD - Start" section.
Makes me think that we need to find more places where the iMacPro board-id needs to be injected
 
  • Like
Reactions: w1z, cdf, tsialex and 2 others
h9826790

h9826790

macrumors P6
Original poster
Apr 3, 2014
16,656
8,587
Hong Kong
startergo said:
This is how VideoProc treats hardware acceleration:
View attachment 877150
Intel really?
Click to expand...

This is a known cosmetic error. Obviously, there is no Intel GPU on our Mac, but only the AMD GPU has the hardware to do the job.

startergo said:
ffplay is struggling to play HEVC (uses CPU). It plays fine H264
[automerge]1573786570[/automerge]
I might be chasing my tail with Kepler and ffmpeg:
View attachment 877173
developer.nvidia.com

FFmpeg

developer.nvidia.com developer.nvidia.com
Click to expand...

There are three problems for the Titian to provide HEVC HWAccel.

1) As per your finding. Kepler GPU has no associated hardware for HEVC HWAccel.

2) There is no web driver support in Mojave.

3) Nvidia never provide NVENC and NVDEC support in any of their MacOS web driver / CUDA package.

Any one of this already makes your Titan impossible to decode that Sony Swordsmith HEVC sample. And at this moment, all three exist together.
 
  • Like
Reactions: Flacko
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom