Hardware for 4k recommendation please

Hi olli@,

thank you for the comment on the connectors. I have discovered that when I was browsing the different pictures.

Kindest regards,

M
 
Phishfry said:
Is anybody running a 4K FreeBSD desktop?
Click to expand...
I do, using this hardware:

vgapci0@pci0:65:0:0: class=0x030000 card=0x05211043 chip=0x67df1002 rev=0xe7 hdr=0x00
vendor = 'Advanced Micro Devices, Inc. [AMD/ATI]'
device = 'Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]'
class = display
subclass = VGA
bar [10] = type Prefetchable Memory, range 64, base 0x80000000, size 268435456, enabled
bar [18] = type Prefetchable Memory, range 64, base 0x90000000, size 2097152, enabled
bar [20] = type I/O Port, range 32, base 0x2000, size 256, enabled
bar [24] = type Memory, range 32, base 0x9ed00000, size 262144, enabled

and a DELL P4317Q 40" monitor with a 3840x2160 resolution.

There's one gotcha: I run it with FreeBSD 12.2 as I have not yet found a GPU that is supported with 13.0 8-(
 
mefizto said:
regarding the video card brand,I have had both. My understanding is that Nvidia has better drivers, the problem as I see it is that the drivers are closed, thus if/when Nvidia decides that the market is not worth it
Click to expand...

The only time nvidia let me down is when they dropped their logo display on X11 boot.

It is not that nvidia has better drivers, it does, but its not about that. It's about primary support from a hardware vendor. That relationship will soon be 20 years old.

Their effort is minimal. They have a proprietary blob, platform agnostic. The kernel module itself is just a loader (and open source). FreeBSD effort is on the kernel loader. The rest - meaning firmware and userland (X11 driver, libgl module, vulkan, nvidia-settings, etc.) are not FreeBSD specific effort. That means that a tiny fragment of the FreeBSD release is actually a FreeBSD primary effort. Which means as long as the graphics stacks of Linux and FreeBSD stay near compatible, there will be no foreseeable support issues with nvidia on FreeBSD.

So yes, if you want cutting edge discreet graphics, you can be sure that nvidia supports it here.

However be warned that large DPI on average X11 environment is very problematic. You may have software issues and therefore I do not believe you should be asking hardware questions. Go and check out every desktop application you're using and does it honour DPI desktop settings and can it redraw itself correctly with high DPI. A lot of desktop programs from ports are unreadable on 2K and while a lot of them can scale up either by manual settings or by environment, they don't look good.

So recheck your software and be sure that nvidia can drive any kind of monitor on FreeBSD.
 
Phishfry said:
I don't think the MSI brand is so important as they all use an NVidia reference design
Click to expand...
Not true, MSI has plenty of AMD cards.
As for monitors, TV's are getting cheaper and cheaper, but GPU prices are only climbing up. These days, you can get a 40 inch TV for about $300 USD (and even cheaper if you shop around). But for a decent GPU - $500 USD is rock bottom, even for used/old ones. An unfortunate reality of the GPU market. 😩
pi@ : Given the class of your GPU:
[QUOTE user=pi@]
vendor = 'Advanced Micro Devices, Inc. [AMD/ATI]'
device = 'Ellesmere [Radeon RX 470/480/570/570X/580/580X/590]'
[/QUOTE]
I'd think that it's not impossible to upgrade to 13-RELEASE - I have a Radeon RX 550 4 GB card running 13-RELEASE. :p
 
Zare said:
Their effort is minimal. They have a proprietary blob, platform agnostic. The kernel module itself is just a loader (and open source).
Click to expand...
This is neither technically correct nor close to the ground truth.

Zare said:
Which means as long as the graphics stacks of Linux and FreeBSD stay near compatible, there will be no foreseeable support issues with nvidia on FreeBSD.
Click to expand...
Well, in common usage "the Linux graphics stack" means things Nvidia doesn't want to implement, like DRI.
 
shkhln said:
This is neither technically correct nor close to the ground truth.
Click to expand...

Of course a kmod needs to do more than that. It's not meant to be technically correct, it to provide info to poster in simple format who believes that nvidia could drop support for FreeBSD. It's meant to be hyperbole, which I believe conveys the ground truth - there's 15k LOC in nvidia-driver src as opposed to 90 MB of delivered libs.

The point isn't what it does or doesn't, its about the size of platform specific part and the effort needed to maintain it.

My theory is that nvidia's FreeBSD marked might be small in absolute numbers but the effort behind it is even smaller. So one should not bother with "low market share" predictions of doom.

shkhln said:
Well, in common usage "the Linux graphics stack" means things Nvidia doesn't want to implement, like DRI.
Click to expand...

You are correct. It's about userspace libs developed mostly for GNU/Linux environment. I don't think Xorg or Mesa guys care what happens with FreeBSD per se, their reference platform is GNU/Linux.
 
I don't understand why the question. If you're steering into discussing about what nv-kernel.o does than I don't know and I'm not sure it has been fully reversed. The analysis several years old had compared Windows, Linux and Solaris stub and the claim was that the kernel blob is largely the same thus the platform agnostic part.
 
Analysis? I don't the think the Windows driver is that similar, actually. Anyway, the point is that the driver is not open-source, it's not at all minimal and while the shim part is rather small, we don't know for sure there isn't any OS-specific conditional logic in the blob (there bound to be at least a few fallbacks here and there), so it's not appropriate for us to speculate how difficult or easy would be for Nvidia to keep the FreeBSD support in shape.
 
You must log in or register to reply here.
Back
Top