Timeline for Power apps for S5R4 Run?

rbpeake
rbpeake
Joined: 18 Jan 05
Posts: 266
Credit: 968400345
RAC: 1167812

RE: ...No problem. Sorry

Message 83422 in response to message 83421

Quote:
...No problem. Sorry for the name confusion. Trying to do a million things at once here and my eyes saw Bob and my brain processed Rob. Must have been a random cosmic ray hit causing a bit flip on one of my neurons because I'm not normally dyslexic. ;)


Rob works just fine! And I think a bigger cosmic ray hit a bunch of my neurons! ;)

Arion
Arion
Joined: 20 Mar 05
Posts: 147
Credit: 1626747
RAC: 0

RE: Wow, I can understand

Message 83423 in response to message 83418

Quote:

Wow, I can understand the confusion on this though because I'd say the tasks have gotten quite a bit longer.

I've only had time to look in detail at a couple of computers but a P4D that used to take ~9.75 hours on 4.36 is going to take about 13.5 hours for it's first S5R4 and a Xeon 5120 that took about 6.75 hours (4.36) will take about 11.5 hours for it's first S5R4.

Since that takes the run times close to what the pre-SSE apps took for S5R3, I'm not surprised that folks are going to initially think it's no longer an SSE app.

Oh, I did check both of these to make sure they were running the _1 executable and they are. :)

[edit] I do foresee at least one problem though... the estimated run times are 1h46m for this Xeon that is actually going to take 13h to complete it's first one. I'm waaaaaay overloaded on work on this machine so I doubt I'll be downloading more for well over a week. The settings should have grabbed .25 days of work, but I'm sure E@H has accumulated a healthy LTD vs other projects, so perhaps that explains why it actually downloaded what it thinks will be exactly 2 days worth.[/edit]

[edit] As always, thanks for the good information Bikeman! [/edit]

Mine did the same thing.. It grabbed 32 units then said I hit my daily quota. they were showing up at 1 1/2 hours and right now I'll 11hrs in and only at 80%. These systems are all Win XP systems - 2 dual cores and 1 single core.

Brian Silvers
Brian Silvers
Joined: 26 Aug 05
Posts: 772
Credit: 282700
RAC: 0

RE: RE: Wow, I can

Message 83424 in response to message 83423

Quote:
Quote:

Wow, I can understand the confusion on this though because I'd say the tasks have gotten quite a bit longer.

I've only had time to look in detail at a couple of computers but a P4D that used to take ~9.75 hours on 4.36 is going to take about 13.5 hours for it's first S5R4 and a Xeon 5120 that took about 6.75 hours (4.36) will take about 11.5 hours for it's first S5R4.

Since that takes the run times close to what the pre-SSE apps took for S5R3, I'm not surprised that folks are going to initially think it's no longer an SSE app.

Oh, I did check both of these to make sure they were running the _1 executable and they are. :)

[edit] I do foresee at least one problem though... the estimated run times are 1h46m for this Xeon that is actually going to take 13h to complete it's first one. I'm waaaaaay overloaded on work on this machine so I doubt I'll be downloading more for well over a week. The settings should have grabbed .25 days of work, but I'm sure E@H has accumulated a healthy LTD vs other projects, so perhaps that explains why it actually downloaded what it thinks will be exactly 2 days worth.[/edit]

[edit] As always, thanks for the good information Bikeman! [/edit]

Mine did the same thing.. It grabbed 32 units then said I hit my daily quota. they were showing up at 1 1/2 hours and right now I'll 11hrs in and only at 80%. These systems are all Win XP systems - 2 dual cores and 1 single core.

Not only that, but I think Bernd made a very large overshoot on the credit adjustment.

Looking at this Mac host, one can see that runtimes are significantly longer, while claimed is significantly less. Looks like to me that instead of a 20-25% reduction, what will be seen is a 75-80% reduction, that is if claimed = granted...

Edit: Data on this is very hard to come by right now, but I did find a Linux host among the top hosts list that had already started returning S5R4. Similar reductions in credit (claims of roughly 25% of S5R3), with runtime also increasing 40-50%. Nothing has been granted yet, so I don't know if maybe this project is an early adopter of the new-fangled thingamajig that SETI proposed, but if grants == claims, then this project just went even lower than LHC and Orbit in the credit department, and I thought that was next to impossible...

Something seems to be very wrong...

Arion
Arion
Joined: 20 Mar 05
Posts: 147
Credit: 1626747
RAC: 0

RE: Not only that, but I

Message 83425 in response to message 83424

Quote:

Not only that, but I think Bernd made a very large overshoot on the credit adjustment.

Looking at this Mac host, one can see that runtimes are significantly longer, while claimed is significantly less. Looks like to me that instead of a 20-25% reduction, what will be seen is a 75-80% reduction, that is if claimed = granted...

If those figures stand up he didn't overshoot them, he has the correction reversed. I'm no CW, but 14+ hours for 50 or 60 credits is not suffient for the amount of time, electricity, or headaches that go along with maintaining my systems to produce valid results.

I hope this is rectified ASAP.

Is there anyway to get confirmation from him that these are correct credit levels for the work done?

[edit] I ran seti over the weekend to keep my systems running and used the stock seti app. and for roughly 4 or 5 hours runnign time I was being granted aprox. 50+ credits. Based on what we were told a 20 to 25% reduction should have cut us down to around 180 or so per unit.[/edit]

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686138939
RAC: 552243

RE: RE: Not only that,

Message 83426 in response to message 83425

Quote:
Quote:

Not only that, but I think Bernd made a very large overshoot on the credit adjustment.

Looking at this Mac host, one can see that runtimes are significantly longer, while claimed is significantly less. Looks like to me that instead of a 20-25% reduction, what will be seen is a 75-80% reduction, that is if claimed = granted...

If those figures stand up he didn't overshoot them, he has the correction reversed. I'm no CW, but 14+ hours for 50 or 60 credits is not suffient for the amount of time, electricity, or headaches that go along with maintaining my systems to produce valid results.

I hope this is rectified ASAP.

Is there anyway to get confirmation from him that these are correct credit levels for the work done?

[edit] I ran seti over the weekend to keep my systems running and used the stock seti app. and for roughly 4 or 5 hours runnign time I was being granted aprox. 50+ credits. Based on what we were told a 20 to 25% reduction should have cut us down to around 180 or so per unit.[/edit]

Don't panic :-). Bernd has repeatedly stated that the yardstick for credit is some kind of cross-project equality for the major projects (a comparison with SETI yields this 25% overpay cited before). This indeed seems to be way too low.

Also note that (at least as I write this), the 60 or so credits are just claimed credits, not granted credits. AFAIK the E@H servers will completely disregard claimed credit, it will award the credit that was computed for the WU when it was generated. Remember, the credit scheme here is NOT benchmark based. Clients can claim whatever they like :-)

Please have some faith & patience, and stay tuned.

CU
Bikeman

Bernd Machenschalk
Bernd Machenschalk
Moderator
Administrator
Joined: 15 Oct 04
Posts: 4267
Credit: 244932206
RAC: 16344

Please don't give anything on

Please don't give anything on the claimed credit right now. We're still adjusting parameters; as the granted credit is determined by the server and independent of the claimed credit, the latter is one of our lesser worries.

BM

BM

Arion
Arion
Joined: 20 Mar 05
Posts: 147
Credit: 1626747
RAC: 0

RE: Please have some faith

Message 83428 in response to message 83426

Quote:

Please have some faith & patience, and stay tuned.

CU
Bikeman

I've been through a transistion here before and this is my pet project. Faith & Patience in this project is what has kept me here amoung a lot of other good things such as the fellow crunchers. :-)

I don't make hasty decisions and I'll wait to see what happens. Considering what it is like on other projects why would I want to get thrown to the dogs..

Brian Silvers
Brian Silvers
Joined: 26 Aug 05
Posts: 772
Credit: 282700
RAC: 0

RE: Please don't give

Message 83429 in response to message 83427

Quote:

Please don't give anything on the claimed credit right now. We're still adjusting parameters; as the granted credit is determined by the server and independent of the claimed credit, the latter is one of our lesser worries.

BM

It is just that S5R3 always had claim == grant, so apparently you're saying you don't have the parameters set to where the claim is up to where the server will grant...

If something had validated already, that would be seen. As it stands, I can't find a validated task...thus I brought it up with caveats...

Brian Silvers
Brian Silvers
Joined: 26 Aug 05
Posts: 772
Credit: 282700
RAC: 0

RE: RE: Please have some

Message 83430 in response to message 83428

Quote:
Quote:

Please have some faith & patience, and stay tuned.

CU
Bikeman

I've been through a transistion here before and this is my pet project. Faith & Patience in this project is what has kept me here amoung a lot of other good things such as the fellow crunchers. :-)

I don't make hasty decisions and I'll wait to see what happens. Considering what it is like on other projects why would I want to get thrown to the dogs..

Same here, as far as transitions. S5R2 was very bad, but Bernd (and presumably Akos) turned things around. The servers are up and running more than other projects, and urgent issues don't seem to languish like they do elsewhere (:cough: Cosmology :cough:)

Stranger7777
Stranger7777
Joined: 17 Mar 05
Posts: 436
Credit: 417525914
RAC: 33706

We are here mainly to crunch

We are here mainly to crunch the data and find out the way whether GWs exist or not >:[
And I think it is not a matter how much credit do you get for it. To reach more credits I myself turn on more machines and use power apps (if possible) just because I want to work only on E@H. The other way (not for me) is to choose a suitable projects that grants more credits per hour. Any other possibilities are on the project organizers opinion.

So, now it is better to ask the question: how long our new run will be and how much data does it contain comparing to for e.g. S5R3?

P.S. Please, don't perceive it too seriously. I'm not angry and didn't even thought to say something bad :)

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.