BOINC opinion poll

[graeme]

What do people think of the BOINC project? Is this something we should consider implementing an eon client for? Are there any pros or cons?

[black_civic55]

i enjoy BOINC, very easy to use I think. And i don't really enjoy EON the way it is now. Not much else i can say, i'm not too detailed with computers.

[em99010pepe]

I have to be honest. I hate BOINC project.
Pros or cons? Until today always had problems with credit pending, several errors when downloading new jobs, bandwith limits, etc.

Carlos

[Bok]

From when I've tried it (with the LHC project) it was very simple, but that kind of takes away from the fun of it for me.

I also hate the 'pending' credit feature.

I'll probably revisit the boinc linux client at some point and see if it's improved at all. It has to be a simple commandline client for me :)

Bok

[magnav0x]

I would have to say, that boinc is a rather good and bad idea. I find it impossible to remember my "account" urls for projects on boinc when setting up on computers, forcing me to goto my e-mail or the projects website in order to get the info.

All in all I love the EON command line client. Simple, elegant, and doesn't use much memory. The current client is also ideal for installing as a service. Just say no to BOINC!

[PCZ]

No thanks

I have nothing good to say about Boinc.
I give it a fair chance but walked away extremely dissapointed.

The credit system is awful.
Sometimes you wait weeks to get credit, if you get any at all.
The software is updated constantly and not all the projects are able to keep up.
There are lots of problems with machine ratings being wrong in a big way.
The users are paranoid and acuse each other of cheating.
The forums are ruined by trolls, which the admins do nothing about.

How can you trust something being championed by an organisation that farmed out the same WU's over and over again and allowed cheating to continue unchecked.

[Longbow]

I dislike Boinc and am not a contributer to any projects under that system. I could see your potential attraction to it for adding to your user base, but I honestly think a more native windows client would work just as well.

The only real gripe I have with the current windows client (besides hearing it is slower than the linux varient) is the fact that it minimizes to the task bar rather than the tray.

[black_civic55]

yea same as longbow said, id rather have it minimize in the tray then the taskbar

[ECL]

As it stands, Boinc is an exceptionally poor implementation of what was originally a very good idea. The infrastructural demands and learning curve (on the project side) are also pretty steep. Boinc is still quite immature, so it's probably better to improve your existing clients and let other projects do the heavy lifting for Berkeley.

Currently, Linux users pay a heavy performance penalty on Boinc, and you may have noticed a certain coolness towards Windows from many in the DC community.

You can always do a Boinc client some time down the road. When and if Boinc works better.

[willy1]

I'll add an emphatic :bs: thumbs-down on BOINC, for most or all of the reasons others have listed above.

[rcoulter]

I have to agree with much that has been said. First off Magnav0x is right about tring to remember your "account". With Eon it's easy (well relatively easy, I still managed to get 2 rcoulter accounts).

Next I like the very simple, quick install. I can have this running on a system in about 2 minutes, which is real useful for me. Since I am constantly add`ing and removing computers before selling them, I can install, run for a few days and easily delete.

Lastly, the only good reason to move to Boinc is to increase the number of clients. The better way is just what you are going now, asking us for feedback and suggestions, letting us know that our results are important, fixing stats and software errors and being involved with those providing the computers. If you want to see the number of users drop off all you have to do is ignore us. This has worked for every project that I have run in the last 6 years or so, including this one in the past.

Keep making improvements, keep us in the loop, and your next problem will most likely be a lack of server power and bandwidth, not users.

[graeme]

Thanks for the great response. It doesn't sound like putting effort into BOINC should be a high priority right now. We'll work on keeping the project alive, listening to your comments, making changes, and keeping you in the loop. Omer is working on a windows gui client which will minimize to the taskbar. I'll also use a different fortran compiler and see if we can improve the client efficiency.

[omer]

Yea, thanks a lot for all the feedback, I am working on our website these days, which should be updated by this weekend, after that i think it should take me another week or two to get the Windows client GUI ready, let me know if any specific features you would like to have (besides the fact that it should minimize to the tray). It will be a lot easy to work on it from scratch than to do changes later!

Once again, i want to thank you all for all the appreciation/comments/suggestions, it really felt nice to read all the replies :)

[IronBits]

The closer to the way the DNET client works, the more users you will get :D
I don't like GUI clients of any kind, and sending it to the tray is always the better thing to do for windows clients.
Especially if the GUI takes even .000001% performance from the work the client can really do...
I don't care about what it looks like as much as I do about the quality of the work the client returns, but more important than that is the reliabliity of the Server, and the clients ability to recover when the Servers do go out of touch for many different reasons. (networks outages for instance)
Keep up the good work! :thumbs:

[Mustard]

Being a hard core *nix fan, it is really nice to have a project for a change where linux/unix can excel. The commandline client works, is not a resource hog, and evidently gives you the answers you are after. You are currently receiving some good input to your system because of the linux systems. Move it to boinc, and you will lose several hard core supporters. Maybe you would get more help from the windows crowd, who knows.

However after your note to me Graeme regarding the issues working against caching, I don't really see how you could do a boinc client without having to re-design your approach in attacking your project. But who knows, maybe you have already figured out a more elegant way of doing that???? :)

But I would personally hate to see boinc come along as others have said above. I'd be another one that would not move to boinc and would move to another *nix friendly effort.

Bruce

[PCZ]

I should have added this to my original post.

On a positive note the source code is available for the boinc platform so you can compile it yourself.
This certainly helped to bridge the performance gap between the nix and windows clients.

Note:
This is just the base boinc framework not the clients that run in it.

[Mustard]

And of course, there is nothing that says that you couldn't run *two* projects, "Eon original" and "Eon Boinc." Of course this might put a bit of a squeeze on the financial aspect of your project. Finish getting the kinks worked out of the current clients, and then start in on a boinc effort.

It's not like the boinc world has an ample supply of solid projects with a non-ending supply of work units.... seems like running out of work is a common complaint of the boinc crowd with the exception of a couple projects that seem to be doing very well on boinc. In that regard I think you would be a very welcome addition to the boinc fans. And you could really concentrate on a highly optimized windows client for boinc. I do think you'd pick up some highly dedicated boinc fans.

Meanwhile those of us that really like the command line stuff, and using linux/*nix would have a project that would be optimized for our O/S for a change.

Then you'd have the best of both worlds with a ton of computing power to help find your answers.