Posts by thimios

1) Message boards : Number crunching : App v0.10 (Message 8812) Posted 21 Jan 2009 by thimios Post: Definitely slower! On Phenom 9500 went to 63 minutes from 46 approx.
2) Message boards : Number crunching : New faster application? (Message 8641) Posted 19 Jan 2009 by thimios Post: I wonder what seti is going to do now with their cuda app. Will they cap it? What has a GPU to do with a CPU ? Its like saying a 3 GHz Quad core shouldnt be alowed to get more credits then a P II... And it doesnt matter how much a project improves over the time, and certainly not compared to other project. How you gonna compare or value the science output from 2 totaly different projects? exactly, you cant. And if MW stock app gets update to a new, faster level, admins might lower the credits again, like they did in the past, then youll be even far away from the numbers you are pulling now - so what? Im happy that I can run WUs so much faster now, and help the project getting to their goals a lot faster. Why dont see it this way? Its much more rewarding then pressing F5 on a statspage every 10mins looking how a stupid number keeps increasing. I don't understand what you are saying. Should a faster processing (on GPU or CPU) get more credit or not? I'm not comparing the value of science either. How can I? My only point is that more work should give more credit, which is not the case in MW currently. About that stupid number, credit, I like it (that's me). I like looking at what I've done over the years processing for scientific projects. Plus, it's fun to play with this number and compare to countries, teams, people, computer technology over time, programming evolution over time etc. It's a good pastime. Gives me a better understanding of the world around me. But if it's capped or otherwise manipulated...
3) Message boards : Number crunching : New faster application? (Message 8622) Posted 18 Jan 2009 by thimios Post: Each time MW has been optimized it has been ~ 6x faster. This I believe is the 3rd of this sort. The begining code from my knowledge was just basic coding to get started on since none of the staff are code experts. If you look at the real early posts they asked for someone to help with optimizing it since none of them know how. I don't know how adept they are now. I believe SETI was like that in their beginning. Just the science and only enough of the programming to run. Sounds logical. Why spend resources on something that might not have a future? I wonder what seti is going to do now with their cuda app. Will they cap it?
4) Message boards : Number crunching : New faster application? (Message 8618) Posted 18 Jan 2009 by thimios Post: I'm certainly NOT talking about shame.
5) Message boards : Number crunching : New faster application? (Message 8617) Posted 18 Jan 2009 by thimios Post: It's not exactly like that. I sure don't mean that. Suppose two projects start. At 1 credit/hour in the beginning (I suppose MW calibrated with SETI at their beginning, am I right?). Then project 2 got optimized to reach so many times the calculations they were getting at the beginning. Why wouldn't they get their credit? Unless of course the projects didn't start at the same level of optimization, which could be the case. If MW was a LOT less optimised than SETI when they started, then that's a way to get such a high optimization "rate" and thus credit. What has really happened here?
6) Message boards : Number crunching : New faster application? (Message 8614) Posted 18 Jan 2009 by thimios Post: What's all the fuss about? MW is handing out more credits than any other (CPU-) project (in my case nearly 2x more compared to seti optimized). Well, seti gets an optimization speedup of ~60% MW is 650% Credit for that 650% is lost, which I think is unfair to MW and its users.
7) Message boards : Number crunching : New faster application? (Message 8611) Posted 18 Jan 2009 by thimios Post: Yes, giving out candy should solve the problem!!!!!!!!!! Funny!!!
8) Message boards : Application Code Discussion : milkyway code license (Message 8610) Posted 18 Jan 2009 by thimios Post: This still leaves Travis and Co with the problem of debugging errors. This is not a license problem, theya have to do that anyway before putting apps on the project server. It also eventually leads to credit discrepancies and flame wars. The only way to avoid this is not to give out the code. To no one, never... Anyway, before releasing under some sort of free license, they have to make sure, that there are no non-free copy/paste pieces in the code. IMHO the above is wrong. So long as milkyway doesn't have a reason for keeping the code closed, it should be open for modification and optimization. Look at the speedup open source projects get (SETI) from donor optimizations. Also note that this is another kind of donoring in addition to cpu time and money. Personally, I don't really like closed source projects. Again IMHO (I'm not an expert), GPL should be enough, it works for linux.
9) Message boards : Number crunching : New faster application? (Message 8607) Posted 18 Jan 2009 by thimios Post: Not to forget. Congrats for the optimized apps. I mean, WOW!!!!!!!! I'm very impressed.
10) Message boards : Number crunching : New faster application? (Message 8605) Posted 18 Jan 2009 by thimios Post: That's really absurd. I wonder what their reasoning is. Why would a 35000 credit/day machine only get 10500. The reasoning is that 'TOO' many credits were being given out. And DA was complaining since that was more than seti standard is putting out. I still think it is not right, for two reasons: - I was planning to upgrade one of my machines just for BOINC. What's the reason now? - Milkyway got an optimization that seti didn't. From 45 minutes to 7 minutes it's 650% faster. What is wrong with that? Personally, I like credit. It's something to have in the place of VERY valuable scientific results that I am not going to see or have for the foreseeable future.
11) Message boards : Number crunching : New faster application? (Message 8603) Posted 18 Jan 2009 by thimios Post: to admins: do u plan to integrate the "new" optimalization in to the standard aplication? how about the workunits, will they be longer - more work would be done. Agree. Longer workunits would also remove some of the load off the server. Perhaps then you would be able to remove the 108 credit/core/hour limit.
12) Message boards : Number crunching : New faster application? (Message 8599) Posted 18 Jan 2009 by thimios Post: That's really absurd. I wonder what their reasoning is. Why would a 35000 credit/day machine only get 10500.
13) Message boards : Number crunching : New faster application? (Message 8595) Posted 18 Jan 2009 by thimios Post: Also, on my second computer, a xeon 5410 2.33mhz quad, the credit for each unit is 9.2 points, Which results in the same 10500 credit per day as the phenom 9500!!! What's up?
14) Message boards : Number crunching : New faster application? (Message 8594) Posted 18 Jan 2009 by thimios Post: Just installed the sse3 app and runs really fast. 7 minutes for each task instead of 45 on phenom 9500. But why is credit for each task ~13 points instead of ~40 that it was? Shouldn't it be the same?