Sighting of the legendary Monster Workunits (630 cr)

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686138814
RAC: 555564

RE: RE: Based on the

Message 69470 in response to message 69469

Quote:
Quote:

Based on the client's estimate, yes, the WU should never have been sent. That's granted. But the question still is whether this estimate is realistic or not. To me it looks too pessimistic.

CU

BRM

The part your overlooking here is the estimates from the scheduler log are not client side generated other than they are based on the BM's and other metrics the host has last reported. They are server side calculated, although I'm not 100% sure at the moment (brain fade) about who actually calculates the RDCF.

Anyway, we can compute RDCF ourselves :-)

ETAcpu = Nflop /BMflops * RDCF

where NFlops is a hypothetical floating point operations equivalent of the computational effort for this WU, estimated by the WU-generator

So, for the 400 MHZ box in question, we have a Benchmark of 300 MFlops

ETAcpu ~ 24 d ~ 2 * 10^6 sec
BMFlops ~ 300 *10^6 ops / sec
Nflop ~ 5.5 * 10^14 (taken from one of my own monster results, see the last argument for the client in client_state.xml for the result in question)

==>RDCF would be close to 1 :-(.

Oh-oh, this estimation seems to be realitic after all for that box ...

CU

BRM

Alinator
Alinator
Joined: 8 May 05
Posts: 927
Credit: 9352143
RAC: 0

LOL... The prosecution

LOL...

The prosecution rests! ;-)

Although the bright side is that S5R2 has brought the issue to a point where I think it deserves another close look at by the project team, once they get a firm handle on functional problems with the new app and validator.

To step up on the soap box for a bit, this is why abandoning the project during this BETA TEST PHASE is really shooting yourself in the foot in the end, regardless of what projects you run.

Taking hosts off the project solely to chase cobblestones makes it so that the project doesn't get tested against every combination it will see in practice. Therefore subtle problems can go undetected and unfixed.

I don't like running a host for 2 weeks for no credit anymore than the next guy does. But if it means it can help refine EAH in particular, and the BOINC framework in general, then that time was not 'wasted' ultimately.

Alinator

Donald A. Tevault
Donald A. Tevault
Joined: 17 Feb 06
Posts: 439
Credit: 73516529
RAC: 0

My AMD 3500+ machine, running

My AMD 3500+ machine, running WindowsXP, has just finished the first "monster" that I've received. It was worth 656.09 Cobblestones, and took 42.35 hours. It looks like the next one that I've received is about the same size.

I hope that my wingmen are up to the task!

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3522
Credit: 686138814
RAC: 555564

RE: I hope that my wingmen

Message 69473 in response to message 69472

Quote:

I hope that my wingmen are up to the task!

Oh-Oh...wingman is a PowerMac with ca 300 MFlops only...I wouldn't hold my breath :-(

Alinator
Alinator
Joined: 8 May 05
Posts: 927
Credit: 9352143
RAC: 0

Yep, I was thinking the same

Yep, I was thinking the same thing. Most likely he'll take a pretty heavy hosing from the massive skypoint flak and go down in flames. :-)

Donald's still flying solo on the new one.

Alinator

Donald A. Tevault
Donald A. Tevault
Joined: 17 Feb 06
Posts: 439
Credit: 73516529
RAC: 0

RE: RE: I hope that my

Message 69475 in response to message 69473

Quote:
Quote:

I hope that my wingmen are up to the task!

Oh-Oh...wingman is a PowerMac with ca 300 MFlops only...I wouldn't hold my breath :-(

He might make it. I have a G3 PowerMac that's slower than this guy's, and it's made deadline on a couple of 525-Cobblestone units.

Alinator
Alinator
Joined: 8 May 05
Posts: 927
Credit: 9352143
RAC: 0

Maybe, if he was running full

Maybe, if he was running full time. However, he's got a DQ of 70 which indicates the last result was a failure, and there's no scoring activity shown on BOINCStats within the last sixty days.

Pretty grim looking prospects at this point, but hope springs eternal. ;-)

Alinator

MattDavis
MattDavis
Joined: 18 Jan 05
Posts: 68
Credit: 10565341
RAC: 0
Alinator
Alinator
Joined: 8 May 05
Posts: 927
Credit: 9352143
RAC: 0

Hmmm... There's some

Hmmm...

There's some interesting observations to be had by recursively looking through your wingman, and his wingmen and so on. ;-)

Alinator

Odysseus
Odysseus
Joined: 17 Dec 05
Posts: 372
Credit: 19556274
RAC: 3250

RE: RE: Oh-Oh...wingman

Message 69479 in response to message 69475

Quote:
Quote:
Oh-Oh...wingman is a PowerMac with ca 300 MFlops only...I wouldn't hold my breath :-(

He might make it. I have a G3 PowerMac that's slower than this guy's, and it's made deadline on a couple of 525-Cobblestone units.


I have a couple of G4/400 Macs, benchmarking at about 300 MFlop/s & 600 MIPS, each attached to this project and at least one other; both are returning three- to four-hundred-CS results on time—but admittedly without a lot to spare, and the systems run BOINC day & night. These tasks take around 100 to 140 hours of CPU-time; extrapolating (given that we get about 330 h of clock-time) suggests that dedicated hosts in this performance bracket should be capable of getting through a “monster WU� within the deadlines.

I’ll take this opportunity to register another plea for the deadlines to be increased. I understand that the science app can’t be changed at the drop of a hat, but do the same ‘conservative’ practices apply to the settings on the servers?

Comment viewing options

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