Contribute
Register

NVIDIA Cuda Driver Version: 418.163 Is Now Available

Status
Not open for further replies.
Is it only for High Sierra? Will I get trouble if I install it in Mojave? I thought when I saw this that ’’now they have released an driver for Mojave, finally’’, but no...
Did you read the release highlights in the link?

New Release 418.105
  • CUDA driver update to support CUDA Toolkit 10.1 and macOS 10.13.6
Recommended CUDA version(s):
  • CUDA 10.1
Supported macOS
  • 10.13
 
On X299 systems, Cuda seems to be unstable. @Junologue has had this problem. A Kernel panic can happen just by clicking a file in finder. But it can be stable for a day or so... it is wierd...
So I wrote a little script to disable it until I need it for an app.

Code:
#!/bin/sh
if [[ $1 = u ]]; then
    echo "Unloading Cuda"
    sudo /bin/launchctl unload -w /Library/LaunchDaemons/com.nvidia.cudad.plist
    sudo kextunload -b com.nvidia.CUDA
    sudo mkdir -p /Library/{Extensions,LaunchDaemons,Frameworks,PreferencePanes,LaunchAgents}.disabled
    [ -e /Library/Extensions/CUDA.kext ] && sudo mv /Library/Extensions/CUDA.kext /Library/Extensions.disabled/
    [ -e /Library/Frameworks/CUDA.framework ] && sudo mv /Library/Frameworks/CUDA.framework /Library/Frameworks.disabled/
    [ -e "/Library/PreferencePanes/CUDA Preferences.prefPane" ] && sudo mv "/Library/PreferencePanes/CUDA Preferences.prefPane" /Library/PreferencePanes.disabled/
    [ -e /Library/LaunchDaemons/com.nvidia.cudad.plist ] && sudo mv /Library/LaunchDaemons/com.nvidia.cuda* /Library/LaunchDaemons.disabled/
    [ -e /Library/LaunchAgents/com.nvidia.CUDASoftwareUpdate.plist ] && sudo mv /Library/LaunchAgents/com.nvidia.CUDASoftwareUpdate.plist /Library/LaunchAgents.disabled/
    exit 0
fi
if [[ $1 = l ]]; then
    echo "Loading Cuda"
    [ -e /Library/Extensions.disabled/CUDA.kext ] && sudo mv /Library/Extensions.disabled/CUDA.kext /Library/Extensions/
    [ -e /Library/Frameworks.disabled/CUDA.framework ] && sudo mv /Library/Frameworks.disabled/CUDA.framework /Library/Frameworks/
    [ -e "/Library/PreferencePanes.disabled/CUDA Preferences.prefPane" ] && sudo mv "/Library/PreferencePanes.disabled/CUDA Preferences.prefPane" /Library/PreferencePanes/
    [ -e /Library/LaunchDaemons.disabled/com.nvidia.cudad.plist ] && sudo mv /Library/LaunchDaemons.disabled/com.nvidia.cuda* /Library/LaunchDaemons/
    [ -e /Library/LaunchAgents.disabled/com.nvidia.CUDASoftwareUpdate.plist ] && sudo mv /Library/LaunchAgents.disabled/com.nvidia.CUDASoftwareUpdate.plist /Library/LaunchAgents/
    sleep 1
    sudo kextload /Library/Extensions/CUDA.kext
    sleep 1
    sudo /bin/launchctl load -w /Library/LaunchDaemons/com.nvidia.cudad.plist
    [ -e /Library/LaunchDaemons.disabled ] && sudo rmdir /Library/{Extensions,LaunchDaemons,Frameworks,PreferencePanes,LaunchAgents}.disabled
    exit 0
fi
echo "OOPS!!! You need to run '$(basename "$0")' with u to unload and l to load.
IE: $(basename "$0") l"

I have is saved in my ~/bin folder, and it is in my path. I have it called cuda. so in a terminal, I just run `cuda u` to unload it, and `cuda l` to load it :)
 
@jaminmc — Literally made an account to come in and thank you. I was head-scratching and throwing myself at a wall for the past few days with a super unstable system. I thought it was the extra 64gb of RAM I added (not getting enough voltage, etc), but then when I saw your post I not only remembered I recently _also_ installed CUDA drivers to try and get an additional performance bump and that I had done zero troubleshooting on that front. That, combined with another note from someone on the forums re: AVX numbers related to OC'ing my CPU, ended up being the solve! I literally couldn't get a single thing exported out of Adobe Media Encoder without a kernel panic. Thanks so much for posting this script. Going to completely remove CUDA and just continue using Metal. I didn't see much of a performance boost in Premiere during the brief moments things were medium-stable anyway. Plus, this means I can go back and attempt the additional 64gb of RAM install now (I had pulled all the sticks thinking that was the cause).

Anyway, this literally saved my ass on a client delivery last night and you deserved hella praise for it! Cheers, friend!!
 
On X299 systems, Cuda seems to be unstable. @Junologue has had this problem. A Kernel panic can happen just by clicking a file in finder. But it can be stable for a day or so... it is wierd...
So I wrote a little script to disable it until I need it for an app.

Code:
#!/bin/sh
if [[ $1 = u ]]; then
    echo "Unloading Cuda"
    sudo /bin/launchctl unload -w /Library/LaunchDaemons/com.nvidia.cudad.plist
    sudo kextunload -b com.nvidia.CUDA
    sudo mkdir -p /Library/{Extensions,LaunchDaemons,Frameworks,PreferencePanes,LaunchAgents}.disabled
    [ -e /Library/Extensions/CUDA.kext ] && sudo mv /Library/Extensions/CUDA.kext /Library/Extensions.disabled/
    [ -e /Library/Frameworks/CUDA.framework ] && sudo mv /Library/Frameworks/CUDA.framework /Library/Frameworks.disabled/
    [ -e "/Library/PreferencePanes/CUDA Preferences.prefPane" ] && sudo mv "/Library/PreferencePanes/CUDA Preferences.prefPane" /Library/PreferencePanes.disabled/
    [ -e /Library/LaunchDaemons/com.nvidia.cudad.plist ] && sudo mv /Library/LaunchDaemons/com.nvidia.cuda* /Library/LaunchDaemons.disabled/
    [ -e /Library/LaunchAgents/com.nvidia.CUDASoftwareUpdate.plist ] && sudo mv /Library/LaunchAgents/com.nvidia.CUDASoftwareUpdate.plist /Library/LaunchAgents.disabled/
    exit 0
fi
if [[ $1 = l ]]; then
    echo "Loading Cuda"
    [ -e /Library/Extensions.disabled/CUDA.kext ] && sudo mv /Library/Extensions.disabled/CUDA.kext /Library/Extensions/
    [ -e /Library/Frameworks.disabled/CUDA.framework ] && sudo mv /Library/Frameworks.disabled/CUDA.framework /Library/Frameworks/
    [ -e "/Library/PreferencePanes.disabled/CUDA Preferences.prefPane" ] && sudo mv "/Library/PreferencePanes.disabled/CUDA Preferences.prefPane" /Library/PreferencePanes/
    [ -e /Library/LaunchDaemons.disabled/com.nvidia.cudad.plist ] && sudo mv /Library/LaunchDaemons.disabled/com.nvidia.cuda* /Library/LaunchDaemons/
    [ -e /Library/LaunchAgents.disabled/com.nvidia.CUDASoftwareUpdate.plist ] && sudo mv /Library/LaunchAgents.disabled/com.nvidia.CUDASoftwareUpdate.plist /Library/LaunchAgents/
    sleep 1
    sudo kextload /Library/Extensions/CUDA.kext
    sleep 1
    sudo /bin/launchctl load -w /Library/LaunchDaemons/com.nvidia.cudad.plist
    [ -e /Library/LaunchDaemons.disabled ] && sudo rmdir /Library/{Extensions,LaunchDaemons,Frameworks,PreferencePanes,LaunchAgents}.disabled
    exit 0
fi
echo "OOPS!!! You need to run '$(basename "$0")' with u to unload and l to load.
IE: $(basename "$0") l"

I have is saved in my ~/bin folder, and it is in my path. I have it called cuda. so in a terminal, I just run `cuda u` to unload it, and `cuda l` to load it :)

Literally a life saver! Thank you! I've been unstable for a year now and always thought it was a processor issue. I use Cinema 4D and Octane for 3D work. This script will be so helpful. Once Octane X comes out I will be able to go AMD.
 
Literally a life saver! Thank you! I've been unstable for a year now and always thought it was a processor issue. I use Cinema 4D and Octane for 3D work. This script will be so helpful. Once Octane X comes out I will be able to go AMD.

Hi!
This KPs happen in my system sometimes several times a day, or it can be stable for days as @jaminmc
says. 99% of the times it is triggered by a click in finder or a spacebar press to quickview some image.
:banghead:
Good to know this kind of KP is CUDA related, until now I had no certain clue of what was causing this panics, RAM? CPU? SSDs? I was wondering if it might be my Wacom Cintq22HD in combination with my 2nd monitor (LG ULtrawide).

My system is a i7 8700K though, and way back in 2014 I was getting very similar KPs on my i7 Apple Mac Mini with an external GTX970 egpu .

I have tried both iMac 18,3 and iMaPro1'1 (right now; with no igpu enabled) but it has been of no help, I am still getting this reboots.
I need Cuda for Octane, so I need nVidia and High Sierra. So I am locked until that Octane X you mention is released.

I think the script will not be usefull to me, as I keep C4D open with Octane running while paint in ClipStudioPaint and go through references, having to disable CUDA would make my workflow impractical. But million thanks @jaminmc for the info and the script.

My tipical panics:

*** Panic Report ***
panic(cpu 2 caller 0xffffff8013c87754): "thread_invoke: preemption_level 1, possible cause: blocking while holding a spinlock, or within interrupt context"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/xnu/xnu-4570.71.46/osfmk/kern/sched_prim.c:2231

or

*** Panic Report ***
mp_kdp_enter(): 3071, 1, 12 TIMED-OUT WAITING FOR NMI-ACK, PROCEEDING
panic(cpu 10 caller 0xffffff8015ab1fc9): "waitq deadlock - waitq=0xffffff80ee30b450, cpu=10\n"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/xnu/xnu-4570.71.46/osfmk/kern/waitq.c:1825

Mac mini was:

Mon Aug 17 16:08:03 2015

*** Panic Report ***
panic(cpu 2 caller 0xffffff8004816df2): Kernel trap at 0xffffff7f856ae02e, type 14=page fault, registers:
CR0: 0x000000008001003b, CR2: 0x000000060000003e, CR3: 0x0000000007d2a000, CR4: 0x00000000001626e0
RAX: 0x0000000000000006, RBX: 0x000000060000003e, RCX: 0x0000000000000007, RDX: 0x0000000000000034
RSP: 0xffffff81edf1bc90, RBP: 0xffffff81edf1bd20, RSI: 0x0000000000000071, RDI: 0xffffff81a2566008
R8: 0x0000000000000001, R9: 0x0000000000000020, R10: 0x00000001c95b808d, R11: 0xffffff7f855d1af2
R12: 0x0000000600000006, R13: 0xffffff802487c5c0, R14: 0xffffff81a2570008, R15: 0x0000000600000006
RFL: 0x0000000000010206, RIP: 0xffffff7f856ae02e, CS: 0x0000000000000008, SS: 0x0000000000000000
Fault CR2: 0x000000060000003e, Error code: 0x0000000000000000, Fault CPU: 0x2

and they share the:

Kernel Extensions in backtrace:
com.nvidia.web.NVDAResmanWeb(10.3.3)[9440CED0-E613-32B7-AAEB-13932263DD0F]@0xffffff7f88b81000->0xffffff7f891f9fff
dependency: com.apple.iokit.IOPCIFamily(2.9)[9C640A56-0FCC-39B4-B469-31F6ED15228E]@0xffffff7f88a94000
dependency: com.apple.iokit.IONDRVSupport(519.21)[8E0EE518-0D53-38C0-B308-2C6CD9EC5B7B]@0xffffff7f88b65000
dependency: com.apple.iokit.IOGraphicsFamily(519.21)[E5B53C5D-BF6E-3D72-9091-A80DFE5426C9]@0xffffff7f88b0f000
dependency: com.apple.AppleGraphicsDeviceControl(3.20.16)[163A3487-6927-3191-9A5A-F395956DDF72]@0xffffff7f88b75000
com.nvidia.web.NVDAGP100HalWeb(10.3.3)[76B91C29-8578-33E2-9C91-A6CB7C5E2C56]@0xffffff7f89377000->0xffffff7f8951dfff
dependency: com.nvidia.web.NVDAResmanWeb(10.3.3)[9440CED0-E613-32B7-AAEB-13932263DD0F]@0xffffff7f88b81000
dependency: com.apple.iokit.IOPCIFamily(2.9)[9C640A56-0FCC-39B4-B469-31F6ED15228E]@0xffffff7f88a94000
 
Hi!
This KPs happen in my system sometimes several times a day, or it can be stable for days as @jaminmc
says. 99% of the times it is triggered by a click in finder or a spacebar press to quickview some image.
:banghead:
Good to know this kind of KP is CUDA related, until now I had no certain clue of what was causing this panics, RAM? CPU? SSDs? I was wondering if it might be my Wacom Cintq22HD in combination with my 2nd monitor (LG ULtrawide).

My system is a i7 8700K though, and way back in 2014 I was getting very similar KPs on my i7 Apple Mac Mini with an external GTX970 egpu .

I have tried both iMac 18,3 and iMaPro1'1 (right now; with no igpu enabled) but it has been of no help, I am still getting this reboots.
I need Cuda for Octane, so I need nVidia and High Sierra. So I am locked until that Octane X you mention is released.

I think the script will not be usefull to me, as I keep C4D open with Octane running while paint in ClipStudioPaint and go through references, having to disable CUDA would make my workflow impractical. But million thanks @jaminmc for the info and the script.

My tipical panics:

*** Panic Report ***
panic(cpu 2 caller 0xffffff8013c87754): "thread_invoke: preemption_level 1, possible cause: blocking while holding a spinlock, or within interrupt context"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/xnu/xnu-4570.71.46/osfmk/kern/sched_prim.c:2231

or

*** Panic Report ***
mp_kdp_enter(): 3071, 1, 12 TIMED-OUT WAITING FOR NMI-ACK, PROCEEDING
panic(cpu 10 caller 0xffffff8015ab1fc9): "waitq deadlock - waitq=0xffffff80ee30b450, cpu=10\n"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/xnu/xnu-4570.71.46/osfmk/kern/waitq.c:1825

Mac mini was:

Mon Aug 17 16:08:03 2015

*** Panic Report ***
panic(cpu 2 caller 0xffffff8004816df2): Kernel trap at 0xffffff7f856ae02e, type 14=page fault, registers:
CR0: 0x000000008001003b, CR2: 0x000000060000003e, CR3: 0x0000000007d2a000, CR4: 0x00000000001626e0
RAX: 0x0000000000000006, RBX: 0x000000060000003e, RCX: 0x0000000000000007, RDX: 0x0000000000000034
RSP: 0xffffff81edf1bc90, RBP: 0xffffff81edf1bd20, RSI: 0x0000000000000071, RDI: 0xffffff81a2566008
R8: 0x0000000000000001, R9: 0x0000000000000020, R10: 0x00000001c95b808d, R11: 0xffffff7f855d1af2
R12: 0x0000000600000006, R13: 0xffffff802487c5c0, R14: 0xffffff81a2570008, R15: 0x0000000600000006
RFL: 0x0000000000010206, RIP: 0xffffff7f856ae02e, CS: 0x0000000000000008, SS: 0x0000000000000000
Fault CR2: 0x000000060000003e, Error code: 0x0000000000000000, Fault CPU: 0x2

and they share the:

Kernel Extensions in backtrace:
com.nvidia.web.NVDAResmanWeb(10.3.3)[9440CED0-E613-32B7-AAEB-13932263DD0F]@0xffffff7f88b81000->0xffffff7f891f9fff
dependency: com.apple.iokit.IOPCIFamily(2.9)[9C640A56-0FCC-39B4-B469-31F6ED15228E]@0xffffff7f88a94000
dependency: com.apple.iokit.IONDRVSupport(519.21)[8E0EE518-0D53-38C0-B308-2C6CD9EC5B7B]@0xffffff7f88b65000
dependency: com.apple.iokit.IOGraphicsFamily(519.21)[E5B53C5D-BF6E-3D72-9091-A80DFE5426C9]@0xffffff7f88b0f000
dependency: com.apple.AppleGraphicsDeviceControl(3.20.16)[163A3487-6927-3191-9A5A-F395956DDF72]@0xffffff7f88b75000
com.nvidia.web.NVDAGP100HalWeb(10.3.3)[76B91C29-8578-33E2-9C91-A6CB7C5E2C56]@0xffffff7f89377000->0xffffff7f8951dfff
dependency: com.nvidia.web.NVDAResmanWeb(10.3.3)[9440CED0-E613-32B7-AAEB-13932263DD0F]@0xffffff7f88b81000
dependency: com.apple.iokit.IOPCIFamily(2.9)[9C640A56-0FCC-39B4-B469-31F6ED15228E]@0xffffff7f88a94000

Hey @Goiniro,

Had any luck over the past few months? I still have to enable and disable as I use it for Octane. I still crashes and KP's as I'm rendering or using it. Its frustrating as hell, so I'm wondering if anyone has had luck aside from just switching to Windows.
 
Hey @Goiniro,

Had any luck over the past few months? I still have to enable and disable as I use it for Octane. I still crashes and KP's as I'm rendering or using it. Its frustrating as hell, so I'm wondering if anyone has had luck aside from just switching to Windows.
Hi! No luck at all, and same feeling over here.
The latest nVidia drivers for HS have not helped either.
I might try to install my GTX970 as second gpu, according to bepeg4d having one card is not the ideal https://render.otoy.com/forum/viewtopic.php?f=30&t=73326#p373767 .

We will have to wait for OctaneX and see what level of performance it achieves with ATI cards or move to windows :(.

Happy New year!
 
Hi! No luck at all, and same feeling over here.
The latest nVidia drivers for HS have not helped either.
I might try to install my GTX970 as second gpu, according to bepeg4d having one card is not the ideal https://render.otoy.com/forum/viewtopic.php?f=30&t=73326#p373767 .

We will have to wait for OctaneX and see what level of performance it achieves with ATI cards or move to windows :(.

Happy New year!

Interesting, I definitely have crashes that don't cause a whole reboot while using octane. I'll have to try the priority thing. Happy new year and hears hoping AMD will release a card that is comparable to the 2080ti. Octane X *fingers crossed*
 
Status
Not open for further replies.
Back
Top