nvidia gpu

My lord, you are probably going to look at me and scowl when I say this but, the intuition you require can be sourced by reading FreeBSD handbook section 5.4.5.

zader said:
rtfm is a perfectly acceptable answer. thanks ...
cheers
Click to expand...
Dr zadar, If you read this, I have a burning question. Is it possible for you to lock the thread?. I ask simply because I have not had my hand in creating a thread yet. :)
 
zader said:
rtfm is a perfectly acceptable answer. thanks ...
Click to expand...

I wish someone had told me RTFM back then but I was the black sheep of that feckless flock. I don't remember reading the Handbook early on for some reason, only learning by trail and error and a hard way to go of it to get where I am today.

The box I'm on now has a Nvidia Quadro 1000M with Optimus Technology and is supported out-of-the-box. I have installed FreeBSD on it before but issues with scrambled install screens and screen resolution once I hit the desktop made OpenBSD more attractive at the time IMO. I have 2 others that use Nvidia Quadro NVS 140 that work well, though all of them are Win7 era machines and not new hardware.
 
At one point, there was a nice tutorial on these forums. Ah, found it.
https://forums.freebsd.org/threads/howto-setup-xorg-with-nvidias-driver.52311/#post-314579

Other things to note. (Some mentioned in the thread). The Linux compatibility option is on by default, so if you make no changes to the configuration, if you use ports, or if you use pkg, you will have to enable linux compatibility. I've found that after installation, using nvidia-xconfig is sometimes necessary to produce an /etc/X11/xorg.conf that will work with the card. (And sometimes not, it may work as soon as the driver's installed, necessary options are added to rc.conf or loader.conf and the system is rebooted.

There's also a GUI nvidia-settings package that can be useful, especially with multiple monitors.
 
zader said:
just wondering how well the nvidia drivers are supported on laptops / desktops ..
Click to expand...

Do you have any specific use case in mind? The drivers exist, whether they are supported is debatable. That is, they are provided "as is" by Nvidia and if you have any issues with them you might as well complain to a wall.
 
Trihexagonal said:
The box I'm on now has a Nvidia Quadro 1000M with Optimus Technology and is supported out-of-the-box. I have installed FreeBSD on it before but issues with scrambled install screens and screen resolution once I hit the desktop made OpenBSD more attractive at the time IMO. I have 2 others that use Nvidia Quadro NVS 140 that work well, though all of them are Win7 era machines and not new hardware.
Click to expand...

Optimus supported out of the box? OpenBSD more attractive desktop with Nvidia? I'm going to need more detail on these. Out of the box, so you didn't install the binary nvidia drivers and were using the vesa driver?

And OpenBSD has no support for Nividia as there is not an nvidia binary driver and the nv driver will not do any higher resolutions and was depreciated in 2010, but since the nvs 140 is older than that maybe that's why it works for you. I ran Nvidia exculsively for years, but switched to Radeon card so I can run OpenBSD as a desktop.
 
roddierod said:
Optimus supported out of the box? OpenBSD more attractive desktop with Nvidia? I'm going to need more detail on these. Out of the box, so you didn't install the binary nvidia drivers and were using the vesa driver?
Click to expand...

Of course I needed to install the correct driver from ports and specify which one to use in the Notebook Compatibility thread. The vesa driver is always installed as a backup no matter the GPU.

When I installed FreeBSD on this W520 before you were met with multiple renditions of the install screen and had to use a boot option to be able to proceed. Once I did hit the desktop it was at something like 640 resolution.

OpenBSD never had the install screen issue and when I hit the desktop it was at the correct resolution with no tweaking necessary before moving on to what needed done. I don't recall needing to specify a driver during the build but I have a screenshot of it running OpenBSD and FreeBSD:

johnnywinter.png
demoni.png

Now FreeBSD proceeds normally during the install process and only needs to same steps to make Optimus work as I use with my other 2 laptops with Nvidia. I used the same driver and settings I do with them, the Nvidia screen appeared during boot and all was well at the right resolution. I knew it would be right when I didn't see scrambled screens at the install screen and went with the build.
 
Optimus is a just a marketing name for an ability to transfer rendered images from Nvidia to Intel GPU (presumably using some WDDM features) + some power management. It's a pure software solution and doesn't have any hardware components. The presence/absence of a hardware multiplexor doesn't have any fancy marketing name and is totally unrelated to the "Optimus" buzzword. The closest thing Nvidia's Unix Driver has is a RandR 1.4 support introduced in 319.12.

Is that clear enough?
 
My T400 has Switchable Graphics with Intel GMA 4500MHD and ATI Mobility Radeon HD 3470 that defaults to the Radeon driver in what must be the forerunner of Optimus. It's what I've been using on a regular basis till now.
 
Trihexagonal said:
OpenBSD never had the install screen issue and when I hit the desktop it was at the correct resolution with no tweaking necessary before moving on to what needed done. I don't recall needing to specify a driver during the build but I have a screenshot of it running OpenBSD and FreeBSD:
Click to expand...

But you have integrated Intel graphics running on your OpenBSD as shown on your neofetch output, not Nvidia. Am I right?
 
shkhln said:
It's a pure software solution and doesn't have any hardware components.
Click to expand...

It sounded like absolutely unnecessary complexity to me so I avoided it like the plague but does it really not have any hardware (i.e some connection between GPUs) to facilitate this?

Grabbing data from the GPU is fairly slow; would this not be an issue each frame?

Otherwise it sounds like I fell for the marketing! XD
 
kpedersen said:
but does it really not have any hardware (i.e some connection between GPUs) to facilitate this?
Click to expand...

That would be a good bragging point for marketing materials. Do you see one? Plus a dedicated link would require a new socket from Intel with more pins.

kpedersen said:
Grabbing data from the GPU is fairly slow; would this not be an issue each frame?
Click to expand...

I imagine this eats a bit into PCI-E bandwidth (not sure how much), other than that I don't see what's slow about it. Laptop hardware is less powerful than desktop hardware anyway, so that should at least partially mask the performance drop due to reduced bandwidth. It's hard to be concerned with theoretical inability to push 240+ FPS when your CPU is already overheating at 20-30 FPS.
 
You must log in or register to reply here.
Back
Top