Hi Fellow BOINCers,
I'm considering adding one or more projects and would like to know what other crunchers think of Einstein, Malaria, uFluids, and Lattice. Rather than the relative merit of what they're doing or the underlying science, I'm more interested in thoughts on how the project is run, problems, available work, admin response, etc. I've kept an eye on the websites' message boards and news, but would like to hear what folks here at BOINCstats think.
Thanks for your input!
MarkR
I run Einstein and Malaria, on 32-bit Windows, 64-bit Linux, and Mac G5 (PowerPC, not Intel). I like both.
Einstein is stable. There is usually, but not always, work. The work units tend to be large. They start, crank for hours and hours, then report. Problems are rare. They have occurred (rarely), but usually the errors are silent and I find them after the fact. They also are pretty good about posting beta clients, if you know where to look on their site. I've rarely had issues with the beta clients. If credit is a big motivator to you, Einstein is also one of the most generous with credit per cobblestone. Einstein supports many platforms. It also has several different kinds of work unit, which have different computation time, so you cannot assume or say "most work units take 10-12 hours". Some models take 4-5 hours on my Linux box, some 30-35, but most 10-20.
Malaria (no PPC client) always has work. Always. Always always always. For me, it has also historically been one of the three buggiest projects. The bugs are not frequent, but they tend to be a bit more dramatic. If a malaria WU crashes, you have to acknowledge the error message (Windows only), and go through the "do you want to send an error report?" before the BOINC client will release it. That means the core it's using will sit idle until you acknowledge it, which I don't like. Old Malaria WU's would occasionally sink BOINC entirely, but it hasn't done that to me in a long time. Also, I stopped running the Java based data optimizer client. On my PC, the processes it would spawn would run at normal priority, and the processor drag would bring my PC to a crawl. I would have to halt all Malaria work until I was finished doing whatever I was doing. I disabled these work units a month or so after they were released, so I have not tried them lately. (This is similar to a problem I would also sometimes have with LHC@Home.) I was able to get into one of their open beta enrollments, so I've been using it for a while. The error rates have improved greatly over time. (Aside from the Java optimizer client, Windows only, it's become pretty stable on Windows on AMD Athlon 32-bit, Fedora Core Linux on AMD 64 bit, and Ubuntu Linux on Via C7-D 32-bit.) I would call it stable and recommend it, but it may crash on occasion. (I had a Windows Malaria WU crash last week, but none for moths before that.)
Both are good. Malaria WU's are much smaller. If you run Einstein, check the web site occasionally to make sure your results are checking OK. If you run Malaria, and you don't touch the PC often, check it occasionally for errors, as well as checking the web site.