I have 3 Telsa P100s on Dell 740xd with RHEL 8.10
kernel=4.18.0-553.81.1.el8_10.x86_64
3b:00.0 3D controller: NVIDIA Corporation GP100GL [Tesla P100 PCIe 12GB] (rev a1)
Subsystem: NVIDIA Corporation Device 11da
Recommended by nvidia is to use
Nvidia 440 driver and cuda 10.2
with either the RPM install or the binary install it hangs at creating kernel modules with DKMS and without DKMS getting assorted errors
kernel-devel is loaded but failures are listed in the binary install of
CC [M] /tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/nvidia/nv-frontend.o
CC \[M\] /tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/nvidia/nv.o
CC \[M\] /tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/nvidia/nv-pci.o
CC \[M\] /tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/nvidia/nv-acpi.o
CC \[M\] /tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/nvidia/nv-cray.o
CC \[M\] /tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/nvidia/nv-dma.o
CC \[M\] /tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/nvidia/nv-i2c.o
In file included from /tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/nvidia/nv-frontend.c:13:
/tmp/selfgz32526/NVIDIA-Linux-x86_64-440.95.01/kernel/common/inc/nv-linux.h:118:10: fatal error: asm/kmap_types.h: No such file or directory
#include <asm/kmap_types.h> /\* page table entry lookup \*/
asm/kmap_types no such file seems to indicate no kernel-headers but kernel-headers is part of kernel-devel.
modprobe nvidia shows no drivers
lsmod |grep nvidia shows nothing
Any suggestions for getting just the drivers installed. Am i using the wrong version?
the chips are older but should be still usable the system was running fine as older 7 kernel until rebuild with 8.10.