| |
| Your personal background. | |
| I'm a retired GE electrical engineer with a keen interest in astronomy. | |
| Your opinions about Einstein@Home | |
| I run E@H because I can, at least for now. I am excited about the project, but feel that like SETI, the work units are much too large and take way too much time, compared with the short deadlines. We must remember that the compute resources are using "spare" CPU cycles and are not 110% dedicated to the E@H task. As such, units that take too long actually waste CPU time because either the user gives up due to lack of progress or (more often) the system gives up on the user before completion. That is the main reason I quit doing SETI. After a while, there's just no point. While I agree that this tends to "weed out" the slower CPU's from the hunt, it seems a shame that there isn't some way that the tens of thousands of us with less that CRAY mainframes can contribute. The estimates of completion time that the software makes when requesting work units is WAY off, I mean by at least an order of magnitude or more in my own case. Suggestions: 1) Make work "units" no more that 1 hour CPU time in duration, sized in proportion to the user's capabilities. That way impatient geeks like myself can see daily progress and are more likely to stick with the program. An agonizingly slow 7-day countdown measured in hundredths of a percent without completion does not give the participant an impression of progress. 2) Don't make participants feel that they are in some kind of competition. Isn't the goal massive parallelism? To get as many CPU's as possible on line, rather than to "weed out" the slower ones? Already, however, the website is showing "top performers" with over 500,000 credits, and the program just got started!!! After two days of computing and so far "0" credits, I already feel like I'm wasting my (and my CPU's) time. If you MUST treat it like an competition, at least be fair and don't make everyone play in the same class as the CRAY clusters or those geeks at Google. 3) The CPU over-heating issue is a potential barn-burner, and is my pet peeve with most all of today's distributed computing schemes. There is really no excuse for this, it is is easy to fix, and should be fixed ASAP. Remember that it is the participant community that you must depend upon to achieve truly massive parallelism, and as such you MUST do whatever you can to minimize a participant's risk. I will be watching my temp and if it climbs above 40 deg C (normally runs 32-38) I'm gone, as this constitutes system abuse in my book. | |
| Your feedback on this profile | |
| Recommend this profile for User of the Day: | I like this profile |
| Alert administrators to an offensive profile: | I don't like this profile |
| Account data | View |
| Team | None |
| Message boards | 1 posts |