[QUOTE=moebius;554346][B]Colab or Google Drive seem to be playing crazy again, the automatic saving does not work properly.[/B][/QUOTE]

Yes, I can confirm this behavior too. Has been happening for the past couple of hours.

[QUOTE=axn;554347][SIZE="1"]Yes, I can confirm this behavior too. Has been happening for the past couple of hours.[/SIZE][/QUOTE]
[B][SIZE="2"]apparently fixed[/SIZE][/B]

[SIZE="1"][QUOTE=Fan Ming;549346]Complied new version of gpuowl (commit E5A8F2C:[url]https://github.com/preda/gpuowl/commit/e5a8f2c5942b3bdb88d92c160bc5dcd953ae33cf[/url]) on Google colab.Not tested, but should work.[/QUOTE]
[/SIZE]
[I][B]I need your expertise again regarding a current gpuowl binary with proofs so that I am up to date with the current status of the project. Thank you very much![/B][/I]

[QUOTE=moebius;555291][SIZE="1"]
[/SIZE]
[I][B]I need your expertise..[/B][/I][/QUOTE]

[COLOR="Indigo"]Does somebody have an executable of the latest version (wich generates Proofs)? I tried to run make under colab but got strange error messages. I don't have a Linux installation with rocm / nvidia drivers available either.
Fan Ming hasn't posted in the forum for a while, I think.[/COLOR]

Wow! I just got Tesla V100-SXM2 instead usual P100. :tu:


V100 around 1.75 faster than P100 on PRP.

[QUOTE=Lorenzo;555620]Wow! I just got Tesla V100-SXM2 instead usual P100. :tu:


V100 around 1.75 faster than P100 on PRP.[/QUOTE]

I wonder if we will get the same run time with these faster GPUs.

V100 first day
 

My four concurrent sessions on the paid tier of Colab each ran for 12.5 hours, each with a Tesla V100. My GPU72 statistics really got a boost from this.

I don't know yet how long I have to "rest" before I can restart the paid sessions.

The increase starting in February is due to switching to the paid tier.

[QUOTE=moebius;555373][COLOR=Indigo]Does somebody have an executable of the latest version (wich generates Proofs)? I tried to run make under colab but got strange error messages. I don't have a Linux installation with rocm / nvidia drivers available either.
Fan Ming hasn't posted in the forum for a while, I think.[/COLOR][/QUOTE]
Isn't ROCm both linux specific and AMD gpu (no NVIDIA) specific? [URL]https://github.com/RadeonOpenCompute/ROCm[/URL]
The attached zip file is the result of a build of gpuowl-v6.11-366 on Google Colab on August 1 2020.
Probably built by the following:[CODE]#draft Notebook to set up a gpuowl Google drive folder for a future Colab session
import os.path
from google.colab import drive
import sys
if not os.path.exists('/content/drive/My Drive'):
drive.mount('/content/drive')
%cd '/content/drive/My Drive//'
!chmod +w '/content/drive/My Drive'

if not os.path.exists('/content/drive/My Drive/gpuowl'):
!mkdir '/content/drive/My Drive/gpuowl'

%cd '/content/drive/My Drive/gpuowl//'
print('git clone')
!git clone https://github.com/preda/gpuowl

%cd '/content/drive/My Drive/gpuowl/gpuowl//'
!apt install libgmp-dev
!update-alternatives --remove-all gcc
!update-alternatives --remove-all g++
!apt-get install gcc-8 g++-8
!update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-8 10
!update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-8 10
!update-alternatives --config gcc
!update-alternatives --config g++
!g++ --version
!chmod +w '/content/drive/My Drive/gpuowl/gpuowl'
print('make gpuowl')
!make gpuowl

!echo create config.txt, worktodo.txt before continuing
print('customize to make a specific new gpu model\'s folder ready')[/CODE]There's also Fan Ming's somewhat earlier version at [URL]https://mersenneforum.org/showpost.php?p=549346&postcount=958[/URL], made weeks after the introduction of PRP proof capability in gpuowl. It may not have the niceties of file cleanup etc. In general, the place to check for this sort of thing is [URL]https://www.mersenneforum.org/showpost.php?p=488539&postcount=4[/URL], where I log and link to what versions have been posted. Linux posts are rare but they do happen.

[QUOTE=kriesel;555882].Probably built by the following [/QUOTE]

[B]All right, I already got it up and running with the help of ATH,

[URL="https://mersenneforum.org/showpost.php?p=555557&postcount=37"]https://mersenneforum.org/showpost.php?p=555557&postcount=37[/URL][/B]

Many reported google drive outage Tuesday
 

Might have affected Google Colab users using google drive also. Teachers complaining. [url]https://www.cnn.com/2020/09/08/tech/google-drive-outage/index.html[/url]

One of my accounts suddenly stopped getting GPU resources at all. For the last 72 hrs, it could not connect to GPU runtime for even once (usually it can use about 30~80 minutes every 24 hours).


I don't know if this is related to the fact that I started to use its CPU for P-1 factoring just shortly before this happened. (Used GPU for TF at ~13:00 one day, started a new notebook to do P-1 at ~21:00, then when I stopped P-1 and tried to use GPU at ~14:00 the next day, it couldn't connect to runtime and never ever could since then.) Interesting enough, I do P-1 on another free Colab account too, but access to GPU on that account is normal (at least until now).