Message boards :
Number crunching :
Anything relating to AstroPulse tasks
Message board moderation
Previous · 1 . . . 23 · 24 · 25 · 26 · 27 · 28 · 29 . . . 120 · Next
Author | Message |
---|---|
betreger Send message Joined: 29 Jun 99 Posts: 11359 Credit: 29,581,041 RAC: 66 |
Slowly they are flowing, how long will it last? |
qbit Send message Joined: 19 Sep 04 Posts: 630 Credit: 6,868,528 RAC: 0 |
Yeah, I got 3 APs or so..... Does anybody know the MB/AP ratio on a "tape"? |
Richard Haselgrove Send message Joined: 4 Jul 99 Posts: 14649 Credit: 200,643,578 RAC: 874 |
Yeah, I got 3 APs or so..... Josef Segur is your man for that sort of question. Each AP WU contains 8 MB of the raw data as recorded at Arecibo (possibly some replaced by shaped noise when needed). Each channel of a 50.2 GB tape file has enough data to produce around 400 AP WUs. But for MB, splitting produces 256 WUs from 64 MB of raw data and there's an overlap with the next group. The channel produces about 62 groups of 256 = 15872 WUs. |
qbit Send message Joined: 19 Sep 04 Posts: 630 Credit: 6,868,528 RAC: 0 |
Interresting, thx! I wonder if so many ppl are eager to get APs because they cover a wider range of science or if it's just because of the credits/Rac. |
Cosmic_Ocean Send message Joined: 23 Dec 00 Posts: 3027 Credit: 13,516,867 RAC: 13 |
Interresting, thx! Mostly the credits/processing-time ratio. I just picked an absolutely random wingmate from my tasks list and they have a zero-blanked AP that was crunched in 3511 seconds and got 593.35 credits for it. That same host, on the same GPU, an MB with an angle range of 0.377422 (I tried finding one around/near 0.41~4 (which is what I think is a "normal" or average WU), but that was as close as I could find) did it in 1152 seconds and got 117 credits for it. AP: 593.35/3511 = 0.168997 credits/second MB: 117/1152 = 0.1015625 credits/second Extrapolate both of those out, and you get: AP: 608.39 credits/hour MB: 365.625 credits/hour tl;dr: AP "pays" better. But it is also a completely different science than what MB is doing, so aside from the credit advantage, there is still good science being done either way, so in the end.. science gets the victory. If there was a third or fourth way to look at this same raw data, that would be even better. The more ways you can look at and analyze the data, the more possibility of getting some kind of answer out of it. Linux laptop: record uptime: 1511d 20h 19m (ended due to the power brick giving-up) |
qbit Send message Joined: 19 Sep 04 Posts: 630 Credit: 6,868,528 RAC: 0 |
Not sure why ppl care that much about credits, but anyway, they do. Would the ppl that run the project be able to raise the credits of MB work by a factor ~1.5, so that MB and AP give roughly the same amount of credits per hour? Or is that something that Boinc staff would have to do? |
petri33 Send message Joined: 6 Jun 02 Posts: 1668 Credit: 623,086,772 RAC: 156 |
Not sure why ppl care that much about credits, but anyway, they do. Would the ppl that run the project be able to raise the credits of MB work by a factor ~1.5, so that MB and AP give roughly the same amount of credits per hour? Or is that something that Boinc staff would have to do? An alternative is to optimize the MB app so that it does more work per hour. To overcome Heisenbergs: "You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones |
qbit Send message Joined: 19 Sep 04 Posts: 630 Credit: 6,868,528 RAC: 0 |
Not sure why ppl care that much about credits, but anyway, they do. Would the ppl that run the project be able to raise the credits of MB work by a factor ~1.5, so that MB and AP give roughly the same amount of credits per hour? Or is that something that Boinc staff would have to do? That would ofc be the best solution! But the AP app will also be improved sooner or later, so we may still have the same problem then. |
petri33 Send message Joined: 6 Jun 02 Posts: 1668 Credit: 623,086,772 RAC: 156 |
Not sure why ppl care that much about credits, but anyway, they do. Would the ppl that run the project be able to raise the credits of MB work by a factor ~1.5, so that MB and AP give roughly the same amount of credits per hour? Or is that something that Boinc staff would have to do? Yes You're right. And from the problems rise both the drive to improve and to learn. To overcome Heisenbergs: "You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones |
HAL9000 Send message Joined: 11 Sep 99 Posts: 6534 Credit: 196,805,888 RAC: 57 |
Interresting, thx! I like that they take longer. So I can build up a queue longer than a few hours on my fastest machines. SETI@home classic workunits: 93,865 CPU time: 863,447 hours Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[ |
betreger Send message Joined: 29 Jun 99 Posts: 11359 Credit: 29,581,041 RAC: 66 |
What a wonderful message since my cache is set to 10 days Aps only 10/7/2015 7:07:16 PM | SETI@home | Not requesting tasks: don't need |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
400 AP's / tape, no wonder it is like finding a needle in the ocean to get 150 tasks. And I'm like Hal, I can get about 2 days worth of GPU, 4 Days CPU with AP tasks, which holds me though almost any outage. And always try to make sure I'm full of MB tasks before maintenance, when I run off all the shorties and resends :) |
rob smith Send message Joined: 7 Mar 03 Posts: 22160 Credit: 416,307,556 RAC: 380 |
I'm not sure where you get "400 AP/tape". Each tape is 50GB of data, each AP is 8MB, so its 6250AP/tape - still not very many, making it a slightly larger needle.... Bob Smith Member of Seti PIPPS (Pluto is a Planet Protest Society) Somewhere in the (un)known Universe? |
Donald L. Johnson Send message Joined: 5 Aug 02 Posts: 8240 Credit: 14,654,533 RAC: 20 |
I'm not sure where you get "400 AP/tape". The actual quote from Joe is Each channel of a 50.2 GB tape file has enough data to produce around 400 AP WUs. So it's 400/channel, about 5600 if it's a full 50GB tape. Donald Infernal Optimist / Submariner, retired |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
Sorry, YES I did misread that .... 400/channel, 14 channels, plus the overlaps. |
qbit Send message Joined: 19 Sep 04 Posts: 630 Credit: 6,868,528 RAC: 0 |
Interresting, thx! Good point also! Problem here is the limit of 100 tasks, which doesn't make much sense in my opinion. Would be better if Seti would respect the cache settings in Boinc manager. |
Jimbocous Send message Joined: 1 Apr 13 Posts: 1849 Credit: 268,616,081 RAC: 1,349 |
Problem here is the limit of 100 tasks, which doesn't make much sense in my opinion. Would be better if Seti would respect the cache settings in Boinc manager. Agreed. Every time I mention this here, however, I get jumped on as being a dummy or selfish? |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
I still think it should be something like .... 150/TurnAroundTime (with max 150) There is no reason for people to have 100 files in their cache for 4 weeks. If they take 5 days to get them back to you, they only get 30. We all know that these lingering files just tie up the databases. Get the files out, back in, processed, and deleted. |
Bill G Send message Joined: 1 Jun 01 Posts: 1282 Credit: 187,688,550 RAC: 182 |
Seems to me that I remember when the DB was crashing all the time because the Results out in the Field was growing beyond reason. That is my understanding why there are limits. As an aside, I see a lot of WUs that sit and sit, until they time out. Seems I get a lot of people who join one day and probably delete BOINC from their computer as they just do not see the sense of participating in the project. I wait sometimes for 3-4 months before a WU finally clears my computer. SETI@home classic workunits 4,019 SETI@home classic CPU time 34,348 hours |
Herb Smith Send message Joined: 28 Jan 07 Posts: 76 Credit: 31,615,205 RAC: 0 |
I really like the idea of having limits based on the power of the machine. Two of my machines would take a week or two to clear 100 AP's. My big machine would work through in 8 to 12 hours if I ever got full cache. It currently works through a full cache of MB's in under 8 hours. I would love to be able to fill the cache with MB, set it to then download only AP's, and forget it for 24 hours. If no AP's come along then refill. As it is if I set to capture only AP's I run dry over night. If set to crunch MB's I end up full cache of MB's just as AP's are split and can't take them. The limits should be set so I can get at least 24 hours of work in a cache. Herb |
©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.