How to install CUDA 10.0 in B7119 with Nvidia Geforce RTX-2080ti

Hardware

B7119, BIOS v1.03, BMC v4.0 (SmartFan: Disable, active video: Onboard)

CPU: Xeon 8180 x2

RAM: DDR4 2666 32GB RDIMM x2

OS: CentOS 7.6 x64

VGA: PNY RTX-2080ti x10 (Driver: v410.93)

Cuda: v10.0

Gpu-burn: v0.9

Note:

1.      Nvidia Geforce RTX-2080ti requires dual 8-pin PCI-e DC power

2.      Connect LAN to DHCP environment before Install OS.

3.      Run as root account

Process

1.      Install RTX-2080ti in B7119

2.      Install CentOS 7.6 as Development and Creative Workstation. Add Additional Development, Compatibility Libraries and Development Tools.

3.      Copy VGA driver and cuda source files from USB disk

cp /usb/NVIDIA-Linux-x86_64-410.93.run .

cp /usb/cuda_10.0.130_411.3161_375.26_linux.run .

4.      Reboot B7119

Disable Nouveau

5.      Boot B7119 to Text mode

6.      vi /usr/lib/modprobe.d/dist-blacklist.conf

Add the follows

blacklist nouveau

options nouveau modeset=0

mv /boot/initramfs-$(uname -r).img /boot/initramfs-$(uname -r).img.bak

dracut /boot/initramfs-$(uname -r).img $(uname -r)

7.      Reboot B7119

Install Nvidia driver

8.      Boot B7119 to Text mode

9.      ./NVIDIA-Linux-x86_64-410.93.run

10.  Reboot B7119

Install Cuda

11.  Boot B7119 to Graphic mode

12.  ./cuda_10.0.130_410.48_linux.run

   Note: Do not select “install VGA driver” during cuda installation

13.  Set CUDA Environment

export PATH=/usr/local/cuda-10.0/bin:$PATH

export LD_LIBRARY_PATH=/usr/local/cuda-10.0/lib64:$LD_LIBRARY_PATH

14.  Reboot B7119

Test Cuda

15.  Boot B7119 to Graphic mode

16.  yum install mesa-libGLU-devel mesa-libGL-devel freeglut freeglut-devel

17.  cd /usr/local/cuda-10.0/samples/5-simulations/nbody

18.  make

19.  ./nbody –benchmark –numdevices=10

Nbody burnin-test

20.  Open ten terminals

21.  /usr/local/cuda-10.0/samples/5_Simulations/nbody/nbody -benchmark -numbodies=65536000 -device=n (n=0 ~9)