User talk:Jesse V.

en:User talk:Jesse V.

Hi Jesse

I'm interested in your sources for the graphs you made on the Folding@home article comparing the the Rmax speeds (TFLOPS) to the fastest top500 supercomputers.

Thanks, Chris — Preceding unsigned comment added by Coffeefilter (talk • contribs) 18:43, 22 February 2019 (UTC)

If you are at it changing the colors would benefit the color blind as well. I'm unable to distinguish the different lines. Theking2 (talk) 12:03, 25 April 2021 (UTC)

This is honestly a very good point. I was a much younger man when I created it and I did not consider that at the time. As for the sources, I was pulling the data from the OS Stats page on the Folding@home site. I found various records of this on the Wayback Machine, various forums that recorded screenshots, and my own observations of the page. The challenge with the OS Stats is that this FLOPS estimation is based on an estimated conversion factor between the operations to compute something like a square root on x86 and on a GPU. I don't think this conversion factor is reliable anymore. Lately the stats have also suffered from an overestimation of the number of active clients, since there were some large groups of people that ran F@h inside a Docker container that did not preserve the client ID, so the servers would think that they were a new user every time they rebooted the container, which was often, and this led to an inflation of the numbers. In short, I trust the OS Stats page at the time but I personally no longer consider it accurate in my opinion. I hope that helps! Jesse V. (talk) 21:09, 5 October 2022 (UTC)

File:Folding@home and Supercomputer Computational Powers.png requires updating

Hi Jesse V.,
Thank you for your contributions to Commons. I noticed File:Folding@home and Supercomputer Computational Powers.png is out of date. The last data point in the graph is from 2013. Could you update it? Thanks again. -- Veikk0.ma (talk) 08:42, 13 March 2020 (UTC)

First, sorry for the very late reply! Yes, it's now 9 years old. The challenge, as I noted in a reply above, is that I was pulling from the OS Stats page on the F@h website, and I no longer personally consider this page trustworthy in its metrics for more recent measurements of the network in the past few years. I'm not sure if there's an accurate method and I personally haven't seen any internal signs from F@h to improve the page. In short, I'm not sure how to get reliable numbers to create a more modern graph. Jesse V. (talk) 21:12, 5 October 2022 (UTC)