GPU72/Colab Error

[chalsall]

Originally Posted by TheJudger

Hi, do you have CUDA toolkit 11.x installed on that system? If so just grab the code, perhaps adjust some pathes in the makefile and run make. That is the prefered way. P.S. this isn't specific to A100 cards, every Ampere card needs a CUDA 11 compiled binary of mfaktc.

Thanks greatly, Oliver.

I have no idea what is currently "under" current Colab instances. I'll climb into one tomorrow by way of reverse SSH, and take a look.

There's nothing quite like talking directly to the developer to get high-bandwidth guidance!

[chalsall]

Originally Posted by chalsall

While I'm writing though... For the last couple of weeks, I have been getting nothing *BUT* T4s! For ~ 3.5 hours per day.

I found this article interesting...

Correlation does not mean causality... 9-)

[chalsall]

Originally Posted by chalsall

I have no idea what is currently "under" current Colab instances. I'll climb into one tomorrow by way of reverse SSH, and take a look.

Hey Oliver et al... So, to answer your question, it looks like the current (at least "Free Tier") Colab instances are installed with cuda-10.0. It /might/ be possible to replace this with a different version during the ephemeral run...

root@Colab_CVH:/# lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 18.04.5 LTS
Release: 18.04
Codename: bionic

root@Colab_CVH:~# lspci
00:00.0 Host bridge: Intel Corporation 440FX - 82441FX PMC [Natoma] (rev 02)
00:01.0 ISA bridge: Intel Corporation 82371AB/EB/MB PIIX4 ISA (rev 03)
00:01.3 Bridge: Intel Corporation 82371AB/EB/MB PIIX4 ACPI (rev 03)
00:03.0 Non-VGA unclassified device: Red Hat, Inc. Virtio SCSI
00:04.0 3D controller: NVIDIA Corporation Device 1eb8 (rev a1)
00:05.0 Ethernet controller: Red Hat, Inc. Virtio network device
00:06.0 Unclassified device [00ff]: Red Hat, Inc. Virtio RNG

root@Colab_CVH:/usr/local/cuda-10.0/bin# ./nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2018 NVIDIA Corporation
Built on Sat_Aug_25_21:08:01_CDT_2018
Cuda compilation tools, release 10.0, V10.0.130

root@Colab_CVH:~# nvidia-smi
Failed to initialize NVML: Driver/library version mismatch

Does anyone have any suggestions?

[LaurV]

Yarrr...

Isn't that a waste?...

BTW, my smi shows cuda 11 available ?? (are you in the wrong folder? hihi)
(you are talking foreign languages anyhow, so sorry if I am out of line)