Message boards :
Number crunching :
I AM SWITCHING OFF
Message board moderation
| Author | Message |
|---|---|
|
Dark Angel Send message Joined: 26 Aug 01 Posts: 432 Credit: 2,673,754 RAC: 0
|
Two things. First, 5.8.0 has a revamped CPU scheduler. Second, the current version of BOINC is 5.4.11, not 5.4.9. Interesting... I knew he was running a i386 but as it doesn't tell what speed it was and it doesn't look to be running at 3Ghz compaired with my machine which is a Mac Pro Quad 3Ghz maching 4GB ram running the same client and the same app which is opptimized for the C2D Mac Pro. My estamate was and is based off of his crunch time on work units where his machine is noticably slower. RAM in this case doesn't do much for the SETI app as even though I have 4GB of RAM the app only taps about 1.5 GB when running full out. Note: I do remember him being in the top 20 at one point or there about.
|
John Clark Send message Joined: 29 Sep 99 Posts: 16515 Credit: 4,418,829 RAC: 0
|
Two things. First, 5.8.0 has a revamped CPU scheduler. Second, the current version of BOINC is 5.4.11, not 5.4.9. I think you will find David is running an Intel Mac Pro (G5 replacement) dual core dual processor system, running at 3.0GHz. It was delivered last August with 4Gb RAM, and he was planning to add another 8Gb, if I remember his data of a few monthes ago. On SETI only he rose to 17 in the Top Computers table (at that time) with a maximum RAC of about 3,400 - 3,500. Since then he has been away a number of times and pulled in SETI Beta. It's good to be back amongst friends and colleagues
|
|
Dark Angel Send message Joined: 26 Aug 01 Posts: 432 Credit: 2,673,754 RAC: 0
|
Two things. First, 5.8.0 has a revamped CPU scheduler. Second, the current version of BOINC is 5.4.11, not 5.4.9. Looks like he's running a quad 2.0 or 2.66Ghz machine I'm guessing a 2.66 by the time to crunch...5.4.9 is the most current version for the mac which is what I'm running. I only crunch for SETI and it looks like SETI beta may be part of the issue where it's balancing the resorces. Just based on that type of power I think that he is running likely a 50/50 share of his time. When I ran two projects on my computer which I did when SETI had it's last server issue about 4 months ago just to give my computer something to do...it was not unusual for me to see only three processors working on a given project at a given time however with careful checking I always found that the 4th was working on the other project. Looking at his numbers with that type of power and a 50/50 project distribution he has about the RAC that I would expect. Take my numbers for instance 3200 current RAC as I write this but running a Quad 3.0 if I shared my processor power with another project I would expect around 1600 RAC per project in a perfectly semetrical world which this is not take into consideration the power difference between the different processor speeds and I believe it would be safe to say that he's sitting about right. Apples for apples (no pun intended) he's running about a third again as much time as I am per work unit using the same app and same client for the same WU credit. seems just about right...now take also into consideration how much his machine runs...mine runs 24/7/365 and crunchs more durring the winter months as I have less work to otherwise occupy it. I was averaging 80% processor time to SETI but lately it's closer to 90-95% which shows in my current numbers.
|
kittyman ![]() Send message Joined: 9 Jul 00 Posts: 50494 Credit: 1,018,363,574 RAC: 2,276
|
I think that the 'bug' you have been discussing is one that I have brought up in the past, and have noticed even recently. It is a rare occurence, but I have seen it on my FX60, c2d and c2quad rigs. I am currently running 5.4.11 Boinc, and what happens (it has occured while I was watching Seti run) is that 2 WUs complete at exactly the same moment, and only 1 new WU starts running, leaving 1 core idling. I have seen this happen on my quad, where 4 WUs are running, 2 happen to complete at the same time, and only 1 restarts, leaving 3 cores running. Exiting Boinc and realaunching always sets things right, and a 4th WU starts when Seti starts to run again. As I said, it is a rare occurence, but it usually happens when a batch of new WUs are downloaded and Boinc goes into EDF mode, preempting what it was working on and starting 2 or 4 of the brand new WUs at the same time. Since the new batch of WUs are often all from the same tape, they are often of the same AR and crunch times, so the stage is set for 2 of them to finish in tandem. It's probably a case of split second timing that causes either Boinc or the OS to not start 2 new WUs at once, leaving 1 core without a task assigned. "Learn from yesterday. Live for today. Hope for tomorrow." Albert Einstein "With cats." kittyman
|
|
Dark Angel Send message Joined: 26 Aug 01 Posts: 432 Credit: 2,673,754 RAC: 0
|
Yep, always a source for great personal satisfaction. :-) Here is a thought for you...I used to play that game as a kid...loved it too...as I remember it was an arcade game.
|
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 63168 Credit: 55,293,173 RAC: 111
|
@Batman So did I, But Ya did have fun didn't ya? I know I did. It was worth the cost of the Mag, A free program, Considering CDs were a dream and the internet was still a top secret project back then. My Amazon Wishlist The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
|
Aurora Borealis Send message Joined: 14 Jan 01 Posts: 3075 Credit: 5,631,463 RAC: 0
|
@Batman Compute! magazine.... Finally recycled the last of them last year. I had them all from the first edition. Type in almost every program for the Commodore. Debuged, cleaned up the code to make it neat as possible and modified it to work to meet my needs and preferences. I sure did waste a lot of hours on those. Boinc V7.2.42 Win7 i5 3.33G 4GB, GTX470 |
Steven - KO4E Send message Joined: 21 Jun 99 Posts: 53 Credit: 2,434,487 RAC: 0
|
Two things. First, 5.8.0 has a revamped CPU scheduler. Second, the current version of BOINC is 5.4.11, not 5.4.9. According to the downloads page 5.4.9 is the current version for the Mac which is what he is running. 5.4.11 is a windows only thing or did I miss something? From one of his results http://setiathome.berkeley.edu/result.php?resultid=446238070 <core_client_version>5.4.9</core_client_version> <stderr_txt> OS X optimized S@H Enhanced application by Alex Kan Version info: OS X SSSE3 (Intel, Core 2-optimized v7.2-nographics) V5.13 by Alex Kan SETI@home classic workunits 5,429 SETI@home classic CPU time 73,472 hours |
zoom3+1=4 Send message Joined: 30 Nov 03 Posts: 63168 Credit: 55,293,173 RAC: 111
|
Yep, always a source for great personal satisfaction. :-) What no swizzle stick?? Horrible thought. Finding bugs by accident, I can sympathise as I had to type in thousands of lines of code in Basic from Compute! magazine, Anyone remember them? For an Atari 400 computer and yeah I had to store that on an audio cassette drive(named the Atari 410), Finding an error there was time consuming, But worth It. :D This was around 1980-1982 of course, Not long after Star Wars came out in fact, there was one game where one would fly over the Deathstars Trench and blast the Deathstar like Luke Skywalker did(In color w/sound effects) and all with only 48k ram too. :D My Amazon Wishlist The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's |
|
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0
|
Yep, always a source for great personal satisfaction. :-) Unfortunately the timing is usually such that Locutus comes along and insists you must upgrade or be destroyed, and you're back to ground zero without a swizzle stick! :-D Alinator |
|
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0
|
Sigh.... And then there is the feeling you get when you finally find the bug that has been around for 5 or 6 years - by code inspection looking for a completely different bug. We had one of those at my last job - the program would crash after about 8 hours of continuous use. It was a resource that was not being cleaned up correctly - enough times and windows 1, 2, and 3 would just quit. BOINC WIKI |
|
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0
|
Sigh.... Yeah I've been there. On one hand you're happy the problem seems to have gone away, but on the other it leaves you with a knot in your stomach because you don't have a reason why it was acting up in the first place (and could rear it's ugly head again anytime). ;-) HAH, and we race headlong to turn our lives and confidential information over to machines running commercial proprietary software. :-) Alinator |
|
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0
|
The part I'm not getting is how can one project be running on 4 and the other on only three? Shouldn't the latest preferences set for one eventually migrate to all, unless they are nonlinked accounts. I don't believe that we ever tracked it down. It just seems to have disappeared when we re-wrote a chunk of code... BOINC WIKI |
|
Odysseus Send message Joined: 26 Jul 99 Posts: 1808 Credit: 6,701,347 RAC: 14
|
From my understanding, each project has it's own preferences. And, just in case it isn’t clear from the foregoing, the number-of-CPUs setting is in the general prefs. The only functional project-specific setting here that directly concerns running the SETI app, aside from those that configure the graphic display, is the resource share. |
|
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0
|
The part I'm not getting is how can one project be running on 4 and the other on only three? Shouldn't the latest preferences set for one eventually migrate to all, unless they are nonlinked accounts. Yes and no. The general preferences are global across projects. The project preferences and the forum preferences are specific to a particular project. BOINC WIKI |
|
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0
|
The part I'm not getting is how can one project be running on 4 and the other on only three? Shouldn't the latest preferences set for one eventually migrate to all, unless they are nonlinked accounts. OK, I didn't make the connection because I was figuring that once the problem showed up for one project it affected all of them. Sheesh, this one must have been a real joy to track down. ;-) Alinator |
Michael ![]() Send message Joined: 21 Aug 99 Posts: 4603 Credit: 7,427,891 RAC: 41
|
The part I'm not getting is how can one project be running on 4 and the other on only three? Shouldn't the latest preferences set for one eventually migrate to all, unless they are nonlinked accounts. From my understanding, each project has it's own preferences. |
Diego -=Mav3rik=- Send message Joined: 1 Jun 99 Posts: 333 Credit: 3,587,148 RAC: 0 |
... all of my machines until Berkeley get things going right. Please Bodley, let me know when you turn your machines back on so I can stop crying my eyes out and be happy again. /Mav We have lingered long enough on the shores of the cosmic ocean. We are ready at last to set sail for the stars. (Carl Sagan) |
|
John McLeod VII Send message Joined: 15 Jul 99 Posts: 24806 Credit: 790,712 RAC: 0
|
The part I'm not getting is how can one project be running on 4 and the other on only three? Shouldn't the latest preferences set for one eventually migrate to all, unless they are nonlinked accounts. 5.4 has a bug where occasionally a CPU will go idle for no apparent reason. This appears to not be a problem with 5.8.0 (it hasn't been seen yet). BOINC WIKI |
|
Alinator Send message Joined: 19 Apr 05 Posts: 4178 Credit: 4,647,982 RAC: 0
|
The part I'm not getting is how can one project be running on 4 and the other on only three? Shouldn't the latest preferences set for one eventually migrate to all, unless they are nonlinked accounts. In that line of thought, can you set things like that locally in the override file? Alinator |
©2020 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.