Message boards :
Number crunching :
How much improvement do optimized clients give?
Message board moderation
Author | Message |
---|---|
[STS]LoB Send message Joined: 23 May 99 Posts: 53 Credit: 3,375,888 RAC: 144 |
Hello! I've just installed these optimized binaries for my Core Duo processor. How much of an improvement in terms of crunch times will they give me? |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
Hello! Hey you are in my start date class. Welcome back. It doubled my Opterons RAC and made a HUGE difference to the rest. Official Abuser of Boinc Buttons... And no good credit hound! |
[STS]LoB Send message Joined: 23 May 99 Posts: 53 Credit: 3,375,888 RAC: 144 |
Hello :) Thanks for the answer, class mate ;) *edit* One more question: If optimization improves crunch times so much, why not detect the client's cpu and send it optimized binaries right from the start? |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
Hello :) That I do not know, other than optimised clients are made by private people that are dedicating tons of their own time and money to help out. Here is a link to our class page stat in case you haven't found it and are interested. Class of May 23, 1999 Official Abuser of Boinc Buttons... And no good credit hound! |
Saenger Send message Joined: 3 Apr 99 Posts: 2452 Credit: 33,281 RAC: 0 |
Hello :) That's done on some projects. The newer BOINCs support this, as they detect the type of CPU quite extensive. Perhaps Seti will follow this as well. But I expect the next credits uproar, because this will lead to less credits per time in comparsion to the optimized ones yet, as everything coming from the project direct is just standard and has to recieve the same credits like other projects. Gruesse vom Saenger For questions about Boinc look in the BOINC-Wiki |
[STS]LoB Send message Joined: 23 May 99 Posts: 53 Credit: 3,375,888 RAC: 144 |
Here is a link to our class page stat in case you haven't found it and are interested. Thanks. I still know that one :D |
[STS]LoB Send message Joined: 23 May 99 Posts: 53 Credit: 3,375,888 RAC: 144 |
Good point, but on the other hand the most important thing is science ;) And moreover it's even more unfair now, as only people who know of the optimized binaries can benefit from them... |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
Here is a link to our class page stat in case you haven't found it and are interested. Hey thanks I hadn't seen that one in years...I ended up #53...was a lot lower but quit about the time they started talking about Boinc...Just started again last May. Official Abuser of Boinc Buttons... And no good credit hound! |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
At least they all have the button to take them here to learn about the apps... Official Abuser of Boinc Buttons... And no good credit hound! |
Clyde C. Phillips, III Send message Joined: 2 Aug 00 Posts: 1851 Credit: 5,955,047 RAC: 0 |
One could look at the crunch times of the default cruncher and compare them to the times of the optimized cruncher. It is very important to compare several units of each credit range; for example, 12-, 18-, 25-30-, 45-50- and 60-65-credit units. Usually the angle range (how far the scope drifts through the stars during the unit's 107-second collection time) determines the number of credits assigned to each unit. It's been more than a double for me (unless the default itself has been improved since mid-2006). |
[STS]LoB Send message Joined: 23 May 99 Posts: 53 Credit: 3,375,888 RAC: 144 |
One could look at the crunch times of the default cruncher and compare them to the times of the optimized cruncher. It is very important to compare several units of each credit range; for example, 12-, 18-, 25-30-, 45-50- and 60-65-credit units. Usually the angle range (how far the scope drifts through the stars during the unit's 107-second collection time) determines the number of credits assigned to each unit. OK, thanks! But to do this, I have to be able to switch back to the default app. How can I do this? |
Odysseus Send message Joined: 26 Jul 99 Posts: 1808 Credit: 6,701,347 RAC: 6 |
If optimization improves crunch times so much, why not detect the client's cpu and send it optimized binaries right from the start? This project doesn’t have the resources to develop optimized apps; there are only a handful of staff and grad-student programmers, who are already busy developing & debugging apps for the Beta project. |
Odysseus Send message Joined: 26 Jul 99 Posts: 1808 Credit: 6,701,347 RAC: 6 |
But to do this, I have to be able to switch back to the default app. Not necessarily: you can look for similar systems on the stats sitesâ€â€or among your quorum partnersâ€â€to compare to yours. How can I do this? If you really want to: set your host to No new tasks, let it finish the work on board, stop BOINC, remove the app_info.xml file from the project folder, relaunch BOINC. It will download the stock app next time it gets work. |
[STS]LoB Send message Joined: 23 May 99 Posts: 53 Credit: 3,375,888 RAC: 144 |
If optimization improves crunch times so much, why not detect the client's cpu and send it optimized binaries right from the start? Yeah, I guessed that. But then again - there are obviously working apps which are optimized. Why not check and then integrate those? |
Pooh Bear 27 Send message Joined: 14 Jul 03 Posts: 3224 Credit: 4,603,826 RAC: 0 |
Yeah, I guessed that. But then again - there are obviously working apps which are optimized. Why not check and then integrate those? A few reasons: 1: The current checking on the CPU only uses the OS to do it, so on newer machines and older OS, it does not see the best optimization, yet. 2: Time, the crew of SETI is small, and they need to take time to make changes. They are busy squashing other issues, doing hardware upgrade necessary, etc. 3: Money, compilers cost money. SETI is on a shoestring budget. 4: SETI is open source, and it is encouraged for others to do better with it. Some of those optimizations may end up in newer versions (they did in the past), as they have time to implement them. Those are just a few reasons. I am sure there are more. My movie https://vimeo.com/manage/videos/502242 |
Josef W. Segur Send message Joined: 30 Oct 99 Posts: 4504 Credit: 1,414,761 RAC: 0 |
... In addition to the other good replies to this question, I'll note that the versions in Beta Test do have code which tests various alternate functions and chooses which is best for the host. Several of the optimizations we've used in the KWSN lunatics.at versions can be recoded to work with that system, and I hope to get that done before the new versions come here. Joe |
[STS]LoB Send message Joined: 23 May 99 Posts: 53 Credit: 3,375,888 RAC: 144 |
Thanks for all the answers, and keep up the good work! :) |
hiamps Send message Joined: 23 May 99 Posts: 4292 Credit: 72,971,319 RAC: 0 |
... Thank you Josef for all your hard work...Everytime Simon gets praise it is also meant for you and Ben all the rest of you great Guys and Gals...You really are appreciated. Official Abuser of Boinc Buttons... And no good credit hound! |
Clyde C. Phillips, III Send message Joined: 2 Aug 00 Posts: 1851 Credit: 5,955,047 RAC: 0 |
One could look at the crunch times of the default cruncher and compare them to the times of the optimized cruncher. It is very important to compare several units of each credit range; for example, 12-, 18-, 25-30-, 45-50- and 60-65-credit units. Usually the angle range (how far the scope drifts through the stars during the unit's 107-second collection time) determines the number of credits assigned to each unit. I would never do that but for the computer-savvy I suppose it's not out of the question. I think I would remember my former RAC and watch it rise slowly to the terminal value, which takes several weeks. One could subtract total outputs of each machine and average the daily differences but then the pending credit and "Ready to Reports" would would cause those values to vary. Hours on per day is a factor, too. |
[STS]LoB Send message Joined: 23 May 99 Posts: 53 Credit: 3,375,888 RAC: 144 |
I think I would remember my former RAC and watch it rise slowly to the terminal value, which takes several weeks. One could subtract total outputs of each machine and average the daily differences but then the pending credit and "Ready to Reports" would would cause those values to vary. Hours on per day is a factor, too. Hehe, good idea, unless your RAC was nearly 0, and you were starting right away with the optimized client. Like I did ;) |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.