netkas.org forum
March 25, 2017, 04:54:21 AM *
Welcome, Guest. Please login or register.
Did you miss your activation email?

Login with username, password and session length
News: Information for registering users http://forum.netkas.org/index.php/topic,2246.0.html
 
   Home   Help Search Login Register  
Pages: 1 ... 9 10 [11]
  Print  
Author Topic: Sierra builds of nVidia Web Driver  (Read 22984 times)
eVasilis
Jr. Member
**
Offline Offline

Posts: 83


« Reply #150 on: January 25, 2017, 10:28:31 PM »

I can't understand why nvidia ties its drivers' releases to particular os builds. Why not one for all?
Logged
eVasilis
Jr. Member
**
Offline Offline

Posts: 83


« Reply #151 on: January 25, 2017, 10:45:42 PM »

I found that at tonymacx86's:

http://us.download.nvidia.com/Mac/cuda_8_0/cudadriver_8.0.63_macos.dmg
Logged
Asgorath
Full Member
***
Offline Offline

Posts: 246


« Reply #152 on: January 25, 2017, 11:55:00 PM »

I can't understand why nvidia ties its drivers' releases to particular os builds. Why not one for all?

http://forums.macrumors.com/threads/frequently-asked-questions-about-nvidia-pc-non-efi-graphics-cards.1440150/

Quote
27) Why does every security update break the NVIDIA web driver?

NVIDIA provides a set of binary driver components that must interact with a binary interface from the Apple components, both Frameworks like Metal or OpenGL and kernel extensions like IOKit. As far as I can tell, NVIDIA has two options:
  • Tie each driver release to a specific build of the OS, so that they can guarantee that those binary interfaces remain static and thus their binary drivers are compatible. Every time Apple releases a new build of the OS, NVIDIA must release a new build of the drivers to match.
  • Assume that Apple won't change these binary interfaces in a security update and allow one driver build to work with any version of a specific OS release (like 10.12.2). If Apple does change one of these interfaces, the driver would continually crash or cause kernel panics, thus rendering the system basically unusable.
NVIDIA has chosen the first of these two options in order to avoid the huge support burden that would be generated any time Apple changes an interface that their drivers depend on. Having the desktop continually crash or the system go into a kernel panic boot loop can be very difficult to recover from, so NVIDIA has chosen the lesser of two evils and will generally require a new driver build to exactly match the specific OS build that Apple releases to the public.

Think about the support burden NVIDIA would have if Apple broke binary compatibility and then everyone's systems started kernel panicking on boot (i.e. the dreaded "reboot loop" problem).
Logged
eVasilis
Jr. Member
**
Offline Offline

Posts: 83


« Reply #153 on: January 26, 2017, 07:42:49 AM »

Thanks!
Logged
netkas
Administrator
Hero Member
*****
Offline Offline

Posts: 798



« Reply #154 on: January 29, 2017, 04:45:38 PM »

10.12.3's webdriver "works" in 10.12.4 beta1

framebuffer works. accelerationb doesnt - kxld[com.nvidia.web.GeForceWeb]: The vtable '__ZTV24nvDisplayPipeTransaction' is malformed. Make sure your kext has been built against the correct headers.
Logged
Spacedust
Full Member
***
Offline Offline

Posts: 206


« Reply #155 on: February 01, 2017, 01:29:03 PM »

10.12.3's webdriver "works" in 10.12.4 beta1

framebuffer works. accelerationb doesnt - kxld[com.nvidia.web.GeForceWeb]: The vtable '__ZTV24nvDisplayPipeTransaction' is malformed. Make sure your kext has been built against the correct headers.


Maybe that would be a good indicator of some new stuff coming (Pascal support) Wink
Logged
mysticalos
Hero Member
*****
Offline Offline

Posts: 569


« Reply #156 on: February 01, 2017, 08:01:42 PM »

10.12.3's webdriver "works" in 10.12.4 beta1

framebuffer works. accelerationb doesnt - kxld[com.nvidia.web.GeForceWeb]: The vtable '__ZTV24nvDisplayPipeTransaction' is malformed. Make sure your kext has been built against the correct headers.


Maybe that would be a good indicator of some new stuff coming (Pascal support) Wink

No, that just means 10.12.4 had enough changes to break driver, which goes back to why nvidia build locks them.
Logged
Squishy Tia
Jr. Member
**
Offline Offline

Posts: 85



« Reply #157 on: February 02, 2017, 01:38:19 PM »

The sad part is, nVidia doesn't have to build lock in Windows under any version of that OS. So the onus here still falls on Apple. And it's a major reason support is such a nightmare for Mac OS. Well that and the yearly abandonment of each OS version. Apple is more fractured than Android, if you can believe that.
Logged
Asgorath
Full Member
***
Offline Offline

Posts: 246


« Reply #158 on: February 02, 2017, 08:11:41 PM »

The sad part is, nVidia doesn't have to build lock in Windows under any version of that OS. So the onus here still falls on Apple. And it's a major reason support is such a nightmare for Mac OS. Well that and the yearly abandonment of each OS version. Apple is more fractured than Android, if you can believe that.

Right, it's a side effect of the Apple driver model, where vendor drivers are closely intertwined with Apple frameworks/kernel extensions.
Logged
Pages: 1 ... 9 10 [11]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines
SMFAds for Free Forums
Valid XHTML 1.0! Valid CSS!