You are viewing a single comment's thread from:

RE: My "Yes" on delisting Moo

in #gridcoin7 years ago

Are we looking at selective participation to the detriment of users with older or less powerful systems or are we going to continue to be objective about this and allow full and equal participation?

Perhaps, with a little initiative ( say talking to them ), we can get Moo to create another OpenCl project that the community as a whole considers more useful. That way less fortunate users may also participate in a 'useful' project ..

Maybe, as they have been whitelisted for such a long time, the foundation could create a bounty for either a replacement OpenCL project crunching useful data by Moo Wrapper or a new project using OpenCL as a replacement?

"SCIENTIA HUMANA LIBERTAS"

boinc
Courtesy of @joshoeah

Sort:  

I briefly heard something about older hardware when I joined in the middle of the hangout discussion but I didn't quite catch what you guys were talking about. Is it that they support very old compute engines or very early OpenCL versions? If they can create an application which does something useful then I'd gladly embrace them.

I don't think we should let hardware support influence project selection at all unless the project in question only supports some super obscure platform. However, we so have a large community and should be able help the open source projects supporting older hardware by providing them with code and test assistance if that is something we want to strive for.

i don't really think it's about hardware support, as much as it is about Gridcoin becoming the deciding factor as to what science is valuable/worth crunching. this is very dangerous territory, and the removal of moo has set precedent to a degree.

i know the argument is "they don't have to crunch Gridcoin-whitelisted projects", but that's also very out of line with "we encourage all new users to join Gridcoin"... and i contest that we removed a large portion of magnitude from lower-end GPU users, and shunted it to people that could already afford more capable cards, i.e. whales.

i can't get my older AMD cards to run on anything but moo. even collatz is throwing errors lately... that said, i would argue collatz is a more useless project than moo, as it doesn't have a finite goal. this is exactly why personal opinion has to be kept out of it. otherwise, we'll be left crunching 3 "approved" projects, and the laughing stock of the distributed community.

we should not seek to remove projects outside of extenuating circumstances such as sourcefinder, or there is evidence of cheating/spoofing.

also, you also touched on something @fkinglag and i were talking about recently... we are trying to get together some developers to make opencl and cuda apps for even simple parallelization of cpu-only BOINC projects. it could end up being one of our more valuable contributions to BOINC. <3

Which AMD card do you have? I have a Radeon 6950 which I use for Milkyway. It's not great but it chugs on.

I think as long as we have the reward structure we have with projects taking rewards from the shared pool we do need some restrictions on which ones to include. I would much more prefer a shared pool with rewards based on individual relative contribution but we don't have a solution for that :(

i have a 4870, a 5450, a 5670, a 5770, an r9 270, and an rx 560. last two do amicable.