On 24/08/17 05:58 AM, Luca Boccassi wrote:
I don't see a 367 release on:
http://www.nvidia.com/object/unix.html
And there hasn't been a 367 release in almost a year.
I have a feeling then its destined to be an abandoned poor cousin. If you
checkout one of the newer READMEs about supported NVIDIA GPU products you will
see it listed right before 340.xx list of supported GPUs
```
Below are the legacy GPUs that are no longer supported in the unified driver.
These GPUs will continue to be maintained through the special legacy NVIDIA GPU
driver releases.
The 367.xx driver supports the following set of GPUs:
NVIDIA GPU product Device PCI ID* VDPAU features
GRID K340 0FEF D
GRID K1 0FF2 D
GRID K2 11BF D
The 340.xx driver supports the following set of GPUs:
...
```
http://us.download.nvidia.com/XFree86/Linux-x86_64/375.66/README/supportedchips.html
Picking one of the above GRID GPUs (K1 in my case) from their download box also
kicks you through to a 367.xx release
Version: 367.57
Release Date: 2016.10.10
Operating System: Linux 64-bit
Language: English (US)
File Size: 73.55 MB
Realistically though, I'm wondering if it is worth the bother. The GRID cards
aren't something individuals are going to have, and it doesn't look like NVIDIA
is serious about GRID customers anyway when they strand just them and then
release no updates. I can definitely confirm that 367.57 needs significant
patching to compile against newer kernels.
I see the latest 340.xx legacy release
Version: 340.102
Release Date: 2017.2.14
Operating System: Linux 64-bit
Language: English (US)
File Size: 66.72 MB
README lists all the stranded GRID cards as supported hardware
```
NVIDIA GRID GPUs
NVIDIA GPU product Device PCI ID* VDPAU features
GRID K340 0x0FEF D
GRID K1 0x0FF2 D
GRID K520 0x118A D
GRID K2 0x11BF D
```
http://us.download.nvidia.com/XFree86/Linux-x86_64/340.102/README/supportedchips.html
so it should work. At least that way you get newer kernel updates. Personally
I would like to just ditch NVIDIA all together for a full open-source in-tree
solution. Unfortunately they seem to have captured the HPC market place with
their CUDA.
Thanks! -Tyson