Hello people,

am present since 2001 with Seti and has experienced, hence, already a lot here. I find the new and renewed slowing down of the arithmetic times WU without announcement or vote once again a cheek of Seti. Now since beginning the arithmetic times were slowed down around the at least 8-fold. It is clear that the Pc's become quicker and quicker and which thereby more and more WU's are calculated.
I am of the opinion, Seti make simply her job not properly and not use the amounts from 3 million ready calculated WU's any more. To push it, however, on justice of different quick PC's, is audacious. Where are we in a few years??? Should weak Monocore-Pc's count then then 120 or more hours? Only to reach less ready WU's?

My consideration to the justice would be there to make it of the number of the processor cores dependent:

Example:
Monocore: WU time / 1 = 100 Credits
Dualcore: WU time / 2 = 50 Credits
Quadcore: WU time / 4 = 25 Credits
8 Core: WU time / 8 = 12.5 Credits


In 2001 we had been sometimes 3 million active users. Today it are worldwide only approx. 0.9 million active users. What this probably lies with???

Greetings

---

i don't know about the WU credits, but the number of active uses is off. i have been working sence 2001 and i dont remember there being more than 500k active (one WU in 30-45 days) at any one time, till we went Bionic.

---

Credits are simply a measure of how much work your machine has done. Different work units have different amounts of credit because some take longer than others.

But giving more credits based on clock speed or # of cores? May as well just go back to counting CPU hours, where a PII would be equal to an I7. But in the real world the I7 does 100X the work, so gets 100X the credits.

Yes at some stage the old PCs just wont be capable of running the new work units any more, thats just the way things go. Thats why no one runs 8086 PCs any more. The flash new I7s will become museum pieces one day too.

Ian

For July and August of 2001, my 200 MHz. Pentium MMX system averaged 41 hours 51 minutes per Classic WU (min. 35:15, max. 48:49). The only work we've recently had with similar processing is VHAR, before the chirp resolution doubling that host was taking about 39 hours for those. Enhanced processing and the recent doubling produce better science, I would be very unhappy if the project didn't use increasing compute capability to extract as much detail from the data as possible. It certainly doesn't bother me that a midrange WU with a 46 day deadline will take about 10 days on that host.

The notional reference computer would earn 100 Cobblestones a day, that's how those are defined. With a RAC around 10, my 200 MHz. host is probably being overpaid.

                                                                  Joe

For July and August of 2001, my 200 MHz. Pentium MMX system averaged 41 hours 51 minutes per Classic WU (min. 35:15, max. 48:49). The only work we've recently had with similar processing is VHAR, before the chirp resolution doubling that host was taking about 39 hours for those. Enhanced processing and the recent doubling produce better science, I would be very unhappy if the project didn't use increasing compute capability to extract as much detail from the data as possible. It certainly doesn't bother me that a midrange WU with a 46 day deadline will take about 10 days on that host.

The notional reference computer would earn 100 Cobblestones a day, that's how those are defined. With a RAC around 10, my 200 MHz. host is probably being overpaid.

                                                                  Joe

Just for fun I put BOINC on this machine. It's been chewing on 1 task for 2 days and I think it still has 100+ hours to go to complete it. Should be fun to see what find of credit it can get lol.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

SETI has recently increased the sensitivity of the calculations doing more science, taking more CPU time and generating more credit / task.

---

BOINC WIKI

Task ID Computer Time reported Run time (sec) granted credit

1315471862 500xxxx 28 Jul 2009 15:26:14 UTC 44,435.33 113.26

1315471863 458xxxx 29 Jul 2009 2:11:55 UTC 12,598.97 113.26


and now the justice ???? ;-)

it is the same WU

---

Task ID Computer Time reported Run time (sec) granted credit

1315471862 500xxxx 28 Jul 2009 15:26:14 UTC 44,435.33 113.26

1315471863 458xxxx 29 Jul 2009 2:11:55 UTC 12,598.97 113.26


and now the justice ???? ;-)

it is the same WU

Cobblestone = amount of work. Your machine 5001711 with the 2.60Ghz Celeron will take more time to preform the same amount of work as your 4781122 2.66Ghz Pentium4.

One of my machines, located here, took 263,549.70 seconds to get 92.43 credit.

This might be helpful.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

yes same work unit but two differet computers. you did that on a celeron and he did it on a dual core. you both did the same work but he had a faster computer. I see no injustice in that. My old P4 with op apps takes about 7 to 8 hours now with the longer Work units. My dual core mac does them in 4 to 5 hours with stock apps. I dont understand why you would want to punish newer computers for being more efficient Me, im happy the old P4 can still crank them out.

---

[/quote]

Old James

Task ID Computer Time reported Run time (sec) granted credit

1315471862 500xxxx 28 Jul 2009 15:26:14 UTC 44,435.33 113.26

1315471863 458xxxx 29 Jul 2009 2:11:55 UTC 12,598.97 113.26


and now the justice ???? ;-)

it is the same WU

Yes, exactly. It is the same work unit with the same amount of work done on both.

Equal pay for equal work has been a long standing concept behind BOINC credit.

Others have pointed out that the two machines are different, and their performance varies alot. The faster machine simply got through the work alot more quickly.

The injustice would be paying them differently.

---

Task ID Computer Time reported Run time (sec) granted credit

1315471862 500xxxx 28 Jul 2009 15:26:14 UTC 44,435.33 113.26

1315471863 458xxxx 29 Jul 2009 2:11:55 UTC 12,598.97 113.26


and now the justice ???? ;-)

it is the same WU

Missing the point - you get credits for doing actually calculations, not for time spent.

If your machine completes work units faster, you get more credits.

If you have a slow machine you get exactly the same credit for each work unit. You just cant do as many in a day.

Completely fair.

Ian