Solved Xorg won't start with officially supported NVIDIA 5070 GPU?

cederom you will need to add hw.nvidia.registry.EnableGpuFirmware=1 to your /boot/loader.conf. The 5000 series cards require GSP firmware to be loaded but it is currently disabled by default in ports due to a suspend/resume issue I am looking into.
 
Yes ashafer I did that and have Xorg working stable :-)

T-Aoki CUDA/OpenCL does not work even with latest 570.169 drivers from ports :-(

Code: 
% uname -a
FreeBSD hexagon 14.2-RELEASE-p3 FreeBSD 14.2-RELEASE-p3 #0 releng/14.2-n269524-1eb03b059e56-dirty: Tue Jun 24 13:08:04 CEST 2025     root@hexagon:/usr/obj/usr/src/amd64.amd64/sys/GENERIC amd64

% pkg info -x nvidia
linux-nvidia-libs-570.169
nvidia-driver-570.169.1402000
nvidia-drm-61-kmod-570.169.1402000_2
nvidia-drm-kmod-570.169
nvidia-settings-570.169
nvidia-xconfig-570.169


% nv-sglrun nvidia-smi
/usr/local/lib/libc6-shim/libc6.so: shim init
Thu Jun 26 16:13:06 2025
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 570.169                Driver Version: 570.169        CUDA Version: 12.8     |
|-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 5070        Off |   00000000:02:00.0  On |                  N/A |
|  0%   38C    P0             31W /  250W |    1204MiB /  12227MiB |      4%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
+-----------------------------------------------------------------------------------------+


% nv-sglrun clpeak
/usr/local/lib/libc6-shim/libc6.so: shim init

Platform: NVIDIA CUDA
  Device: NVIDIA GeForce RTX 5070
    Driver version  : 570.169 (FreeBSD)
    Compute units   : 48
    Clock frequency : 2610 MHz

    Global memory bandwidth (GBPS)
zsh: segmentation fault (core dumped)  nv-sglrun clpeak
 
cederom said:
Hey @mfoacs if this is desktop computer you may want to go into BIO SETUP and configure these things below that also caused problems for me:

1. PCI-e x16 into 8x/8x mode if its 16x (this is crucial).
2. Try disabling PCI-e power management modes, as well as other bus power management modes.
3. Try disabling C-states for CPU and see if that helps too. Too low power stated are not handled by FreeBSD and may cause hiccups.
4. If there are some advanced power management modes you may want to disable them first, then set that these are managed by BIOS not the OS.
5. Make sure RAM has correct speed and SPDII (spd2) mode set so all parameters are taken from the chip. When in doubt try lower the RAM speed.

Modern BIOS Setup has hundreds of options, kinda hard and long way to tune.

Hope that helps :-)
Click to expand...
Intersting, yes.

I have tried with some of those before, namely 1 - 4. I usually don't touch RAM but will do some testing.
 
mfoacs said:
Intersting, yes.

I have tried with some of those before, namely 1 - 4. I usually don't touch RAM but will do some testing.
Click to expand...
And just to confirm, it will have no effect whatsoever. Fiddling with memory timings is couterproductive imo.
 
Currently, x11/nvidia-driver is at 570.169 and brand-new x11/nvidia-driver-devel (conflicts each other just like in legacy drivers) is at 575.64 on latest main branch of ports tree.
It would be included in upcoming 2025Q3 quarterly branch.
 
You must log in or register to reply here.
Back
Top