next steps
-
yoyo
- Vereinsvorstand

- Beiträge: 8126
- Registriert: 17.12.2002 14:09
- Wohnort: Berlin
next steps
Next planned steps:
1) have 2 weeks holliday
2) Suspend/Boinc restart works not properly. Check what the wrapper is doing here and what the cruncher. Every help to check the wrapper code is welcome.
3) create a own dnet account for each boinc user.
4) create a option in user prefs to use the boinc mail address as dnet account.
*) further improvements of automatic work creation/assimilation
1) have 2 weeks holliday
2) Suspend/Boinc restart works not properly. Check what the wrapper is doing here and what the cruncher. Every help to check the wrapper code is welcome.
3) create a own dnet account for each boinc user.
4) create a option in user prefs to use the boinc mail address as dnet account.
*) further improvements of automatic work creation/assimilation
-
SnakeKaplan
- Vereinsmitglied

- Beiträge: 382
- Registriert: 24.07.2007 11:36
-
jrt
- Fingerzähler

- Beiträge: 1
- Registriert: 15.08.2007 14:07
Re: next steps
4) create a option in user prefs to use the boinc mail address as dnet account.
================
Not sure it is a good idea to use mail address as an account. Long time ago, I joined one of the DC projects using my email address as the account name, within 30 minutes I started getting spam emails. I eventually had to create a new email account.
An idle way is to allow 'boincers' to create a dnet account manually or use their exisitng dnet accounts (if they have been crunching dnet) then link to it via yoyo@home. If they prefer not to, then they can continue to contribute to as yoyo@home.
================
Not sure it is a good idea to use mail address as an account. Long time ago, I joined one of the DC projects using my email address as the account name, within 30 minutes I started getting spam emails. I eventually had to create a new email account.
An idle way is to allow 'boincers' to create a dnet account manually or use their exisitng dnet accounts (if they have been crunching dnet) then link to it via yoyo@home. If they prefer not to, then they can continue to contribute to as yoyo@home.
-
yoyo
- Vereinsvorstand

- Beiträge: 8126
- Registriert: 17.12.2002 14:09
- Wohnort: Berlin
-
zombie67
- Prozessor-Polier

- Beiträge: 121
- Registriert: 26.07.2007 21:58
- Wohnort: Reno, NV
Here is a list of items I would like to see (in no particular order):
1) WUs are much longer now. I think I read that it is because each one contains 8 OGR-25 tasks. I had one take 151,702 seconds (42 hours) on a Pentium D 2.8ghz. 42 hours is too long without checkpointing. So until checkpointing is implemented, can the multiple of OGR-25 tasks be reduced?
2) Mac applications: Currently, the CPU Time stays at 0. For scoring, it just means that the wing man sets the score, since the Mac claims 0. No big deal. However, I am concerned about Debt on the Macs. I am not positive, but I believe that since no time is accrued, the LTD continues to increase to the point where no other projects ever run. Is this true about the Debt?
3) Linux application: It looks like the CPU Time runs at half speed. For example, after running for 20 hours, CPU Time shows only 10 hours. Seconds click off very slowly. Unlike the Mac situation, where it claims 0, so the wing man's claim is given, I think this might be worse. The current scoring method is lesser of the two claims (unless one is 0), I believe. So if a Linux box is claiming half of what it should, due to the CPU time being half, then both people are awarded half credits.
1) WUs are much longer now. I think I read that it is because each one contains 8 OGR-25 tasks. I had one take 151,702 seconds (42 hours) on a Pentium D 2.8ghz. 42 hours is too long without checkpointing. So until checkpointing is implemented, can the multiple of OGR-25 tasks be reduced?
2) Mac applications: Currently, the CPU Time stays at 0. For scoring, it just means that the wing man sets the score, since the Mac claims 0. No big deal. However, I am concerned about Debt on the Macs. I am not positive, but I believe that since no time is accrued, the LTD continues to increase to the point where no other projects ever run. Is this true about the Debt?
3) Linux application: It looks like the CPU Time runs at half speed. For example, after running for 20 hours, CPU Time shows only 10 hours. Seconds click off very slowly. Unlike the Mac situation, where it claims 0, so the wing man's claim is given, I think this might be worse. The current scoring method is lesser of the two claims (unless one is 0), I believe. So if a Linux box is claiming half of what it should, due to the CPU time being half, then both people are awarded half credits.
Team: SETI.USA
-
zombie67
- Prozessor-Polier

- Beiträge: 121
- Registriert: 26.07.2007 21:58
- Wohnort: Reno, NV
It looks like 1.6 fixed this, sorta. The CPU time still does not move. But it does jump every time there is the 1/8th incremental checkpoint. For example, the CPU time is "0" until the first checkpoint, at which time it jumps to (say) 1:42:33, when the progress bar moved to 12.5%.zombie67 hat geschrieben:2) Mac applications: Currently, the CPU Time stays at 0. For scoring, it just means that the wing man sets the score, since the Mac claims 0. No big deal. However, I am concerned about Debt on the Macs. I am not positive, but I believe that since no time is accrued, the LTD continues to increase to the point where no other projects ever run. Is this true about the Debt
Team: SETI.USA
-
zombie67
- Prozessor-Polier

- Beiträge: 121
- Registriert: 26.07.2007 21:58
- Wohnort: Reno, NV
Yep. Looks normal to me.
Here are a couple from intel:
http://www.rechenkraft.net/yoyo/result. ... ltid=45457
http://www.rechenkraft.net/yoyo/result. ... ltid=42831
Here are a couple from PPC:
http://www.rechenkraft.net/yoyo/result. ... ltid=46984
http://www.rechenkraft.net/yoyo/result. ... ltid=44663
Here are a couple from intel:
http://www.rechenkraft.net/yoyo/result. ... ltid=45457
http://www.rechenkraft.net/yoyo/result. ... ltid=42831
Here are a couple from PPC:
http://www.rechenkraft.net/yoyo/result. ... ltid=46984
http://www.rechenkraft.net/yoyo/result. ... ltid=44663
Team: SETI.USA