mfaktc: a CUDA program for Mersenne prefactoring

[DrobinsonPE]

Originally Posted by storm5510

I just began using Ubuntu 20.04 LTS. The archive mfaktc-0.21-linux64.cuda10.1-gpusievesize2047.tar.gz does not contain the libraries needed to run. Where can I find them?

try this:
sudo apt install nvidia-cuda-toolkit

[James Heinrich]

Originally Posted by Viliam Furik

You can often find CUDA DLLs by googling them.

All the relevant DLLs can be found here https://download.mersenne.ca/CUDA-DLLs
But they won't help on Linux.

[storm5510]

Originally Posted by DrobinsonPE

try this:
sudo apt install nvidia-cuda-toolkit

It replies with "Unable to locate package nvidia-cuda."

When I try to run it, I get, "Error while loading shared libraries: libcudart.so.10.1: cannot open shared object file..."

I have ran this in the past, but did not keep the archive containing all the libraries. I believe there is more than one.

Trying to find them on the web is like beating the dead horse. Jumping from site to site.




Edit: Disregard all the previous on this subject. I removed Ubuntu from my system.

[Aramis Wyler]

I had a good bit of trouble getting it to go on a debian based linux as well - seemed like now matter how many times I installed the nvidia drivers, be it from apt or the deb file, I couldn't get t to work. It ended up being exactly what DrobinsonPE said: after I got the nvidia kernal going and rebooted I needed to install the toolkit as well. Then everything worked.

EDIT: Turns out I have upgrades available:
sudo apt list |grep cuda |grep toolkit
[sudo] password for aramis:

WARNING: apt does not have a stable CLI interface. Use with caution in scripts.

nvidia-cuda-toolkit-doc/kali-rolling,kali-rolling 10.2.89-5 all [upgradable from: 10.2.89-4]
nvidia-cuda-toolkit-gcc/kali-rolling 10.2.89-5 amd64
nvidia-cuda-toolkit/kali-rolling 10.2.89-5 amd64 [upgradable from: 10.2.89-4]

[storm5510]

Originally Posted by storm5510

Edit: Disregard all the previous on this subject. I removed Ubuntu from my system.

Actually, the above is not really true. I still have Ubuntu on Windows. It is also 20.04 LTS. Using mprime does not pose visual issues like Prime95 does, but then, I should not mention this for fear of being berated, again. I will just say I much prefer console programs now and leave it go at that.

[kriesel]

Originally Posted by storm5510

Actually, the above is not really true. I still have Ubuntu on Windows. It is also 20.04 LTS. Using mprime does not pose visual issues like Prime95 does, but then, I should not mention this for fear of being berated, again. I will just say I much prefer console programs now and leave it go at that.

Looks like you're a little ahead of NVIDIA. The libraries are distributed as part of the CUDA toolkit, and NVIDIA offers nothing later than v18.04 for Ubuntu that I found, for CUDA10.x.

[DrobinsonPE]

Originally Posted by kriesel

Looks like you're a little ahead of NVIDIA. The libraries are distributed as part of the CUDA toolkit, and NVIDIA offers nothing later than v18.04 for Ubuntu that I found, for CUDA10.x.

That is strange. The computer I have my GTX 1650 super in is running Linux Mint 20.1 which is based on Ubuntu 20.04. I installed the CUDA toolkit on that computer and that is how I got mfaktc working.

[kriesel]

Originally Posted by DrobinsonPE

That is strange. The computer I have my GTX 1650 super in is running Linux Mint 20.1 which is based on Ubuntu 20.04. I installed the CUDA toolkit on that computer and that is how I got mfaktc working.

10.2 https://developer.nvidia.com/cuda-10..._distro=Ubuntu 18.04 or 16.04
11.1 https://developer.nvidia.com/cuda-do..._distro=Ubuntu 20.04 18.04 16.04

[storm5510]

Originally Posted by kriesel

Looks like you're a little ahead of NVIDIA. The libraries are distributed as part of the CUDA toolkit, and NVIDIA offers nothing later than v18.04 for Ubuntu that I found, for CUDA10.x.

In this case, is it using Linux drivers or Windows drivers? I believe it to be the latter. I don't see a reason why it would not. I have no need to run a GPU program with it.

[kriesel]

Originally Posted by storm5510

In this case, is it using Linux drivers or Windows drivers? I believe it to be the latter. I don't see a reason why it would not. I have no need to run a GPU program with it.

Don't know. Maybe it would be Windows on the Windows side and WSL-Ubuntu on the other. https://developer.nvidia.com/cuda-do...get_version=20

[aheeffer]

What are the odds? Mfaktc found two factors in the same 79-bit range:

387687056493187475315017 and 413121450787122123422543 divide M333114347