Why the short due date with long workunits?

ErichZann
ErichZann
Joined: 11 Feb 05
Posts: 120
Credit: 81,582
RAC: 0

RE: because the other side

Message 66976 in response to message 66957

Quote:
because the other side of the coin is people whining that they're going to have to wait two whole weeks to get credit for a WU because the idiot they're parterned with isn't actually returning any results.

So what? Where is the problem of waiting 2 weeks for credits? Can you buy anything with them so you need them instantly?
Important is that you get some, it doesnt really matter how long it takes.

You should better think about the scientific part of the project where its more important many people participate, rather than people like you whining for fast credits...

Donald A. Tevault
Donald A. Tevault
Joined: 17 Feb 06
Posts: 439
Credit: 73,516,529
RAC: 0

RE: The problem is that CPU

Message 66977 in response to message 66975

Quote:
The problem is that CPU performance varies more and more (meaning the difference between the slowest and the fastest hosts is growing simply gigantic). If the WUs are long, slow hosts can't cope. If the WUs are short, the fast dual, quad or "whatever" cores will kill off the server. I think what we need is what we had during the S5R1 science run: Decide whether a host counts as "slow" or "fast" and send only the shorter kind of WU to slow ones. It's the best solution I can come up with atm and I do hope the project staff is going to implement it asap (yes, I know they're up to their necks in work already but I have no idea how the crunchers should get along without that feature... honestly... not with 200 MHz Pentiums as well as hyperthreaded Quad Xeons out there)

I agree, but I also believe that deadlines could be lengthened a tad bit.

With the SZTAKI project, some units are long, and some are extremely short. They all still have deadlines of about a month or so.

Annika
Annika
Joined: 8 Aug 06
Posts: 720
Credit: 494,410
RAC: 0

I'm not against longer

I'm not against longer deadlines. But as they can't be extended infinitely, for scientific reasons as well as to keep the crunchers happy, I'd still say the WUs should be distributed based on host performance.

Winterknight
Winterknight
Joined: 4 Jun 05
Posts: 482
Credit: 201,251,344
RAC: 121,304

The main arguement for

The main arguement for extending deadlines, at least until the app is optimised, is too leep the workunit database as small as possible. Allowing part time and older computers to return a unit on time, without the need to replicate another copy and extend the deadline another 2 weeks.

Half of the units in my pending tray are in the re-issue state, because of no replies.

And taking a random look at WU's, there seemed to be a lot of;
no replies - therefore not completed in time
aborted units - user intervention, so they either realised it wouldn't finish in time or don't like long units
comms errors - could this be modem users not willing/or able to maintain long connections
got msg 4 - ???
error 99 - ???

Of the work units I looked at 25 to 30 percent of WU's had three or more replications. This is not good for the servers, or the the project in general because there will always be those who chase credits and want them asap.

So if there is any chance of keeping users who will carry on crunching for Einstein 'for better or worse' until improvements are made, and not jump ship, the deadlines have to be improved.

The option to have initial replication of three still remains if the scientist want validated results sooner, or they think keeping the 'instanteneous gratification' crowd is important.

tullio
tullio
Joined: 22 Jan 05
Posts: 2,081
Credit: 47,433,957
RAC: 10,800

One advantage of a low

One advantage of a low frequency CPU like my 400 MHz PII Deschutes is that it consumes little power, produces little heat, and can be left running 24/7. It takes me about one week to complete a S5R2 Einstein WU, that is about one half of the time allowed by deadlines. This leaves me time to run other apps, like SETI@home and QMC@home. SETI deadlines are easier to meet, while QMC@home is giving heavy CPU loads with deadlines even stricter than Einstein's. But QMC quorums are just made of one, so even a late result is accepted. Compute errors are very rare on my CPU, which must have a very good FPU.
Tullio
I am running SuSE Linux 10.1.

Pav Lucistnik
Pav Lucistnik
Joined: 7 Mar 06
Posts: 136
Credit: 853,388
RAC: 0

Pentium II on 400 MHz have

Pentium II on 400 MHz have thermal envelope of 24W. That's comparable with modern Pentium M class processors, clocked up to 2.1 MHz. MT line of AMD Turions do 25 watts. Need to say the modern chip will do much more work per watt than your oldie. You can even opt for two cores Turion clocked 2000 MHz which will radiate 35 watts.

And if you want a desktop processor, not a notebook one, there's choice too. The energy efficient line of Core 2 Duos are rated 31W.

Now the power consumption is entirely different beast, on which I can't speculate lacking the data.

Bikeman (Heinz-Bernd Eggenstein)
Bikeman (Heinz-...
Moderator
Joined: 28 Aug 06
Posts: 3,516
Credit: 474,398,895
RAC: 113,037

RE: Pentium II on 400 MHz

Message 66982 in response to message 66981

Quote:

Pentium II on 400 MHz have thermal envelope of 24W. That's comparable with modern Pentium M class processors, clocked up to 2.1 MHz. MT line of AMD Turions do 25 watts. Need to say the modern chip will do much more work per watt than your oldie. You can even opt for two cores Turion clocked 2000 MHz which will radiate 35 watts.

And if you want a desktop processor, not a notebook one, there's choice too. The energy efficient line of Core 2 Duos are rated 31W.

Now the power consumption is entirely different beast, on which I can't speculate lacking the data.

If you figure in the energy consumption by all the other components that go with a crunching PC (hard disk, motherboard, graphics card in idle mode ... ), the old boxes are even less competitive.

Maybe just for the fun of it one could compute

average credits
---------------
k watt * h

as a (miles per gallon) equivalent ;-) and to help judge whether it's worthwhile and economically & ecologically sound to use this or that box for crunching

E.g. my
Dual Pentium III 866 , 2 HDs, consumes measured 115 watt under full load.

Approx 11 credits / hour (combined) ==>

ca 96 credits per kilo watt hour

Contrast this with my Mac mini:

Core Duo 1,83 GHz, consumes ca 41 watts (measured) under full load

ca 773 credits / kilo watt hour

If car manufacturers could increase fuel efficiency that fast we all would soon use syringes to refill at the gas station ;-).

So while I'm kind of emotionally attached to my old crunchers, it's not really reasonable to have them run just for crunching IMHO. If they have some function otherwise (router/webserver/fileserver...), that's a different story, of course.

CU

BRM

MAGIC Quantum Mechanic
MAGIC Quantum M...
Joined: 18 Jan 05
Posts: 1,374
Credit: 465,881,704
RAC: 140,746

They find out pretty fast

They find out pretty fast when certain machines are able to run these high data wu's and these are made to run on our modern processors.

Now where I live electricity is much cheaper than gasoline but to run these wu's (or even the smaller ones) I just run the dual or the amd's or p4's that are made in the last couple years so they are actually able to do this.

But still,I tend to drive my old cars and 4X4 that I have had for around 30 years regardless of the price of gasoline........I just drive them when I need to.

(ok some times in the summer I need to drive them just to go on a cruise and listen to music in my paradigm on wheels )

Urban
Urban
Joined: 20 Feb 05
Posts: 7
Credit: 49,304,802
RAC: 15,557

RE: One day after is pure

Message 66984 in response to message 66963

Quote:
One day after is pure luck, there you get the credits.
Four or five days after you get nothing.

This is for me the Point to say for now good by to E@H until the due time is changed!

Dr. Atanu Maulik
Dr. Atanu Maulik
Joined: 5 Jan 06
Posts: 16
Credit: 5,308,855
RAC: 0

It is not just about

It is not just about processing speeds, it is also about how many hours per day you allow your computers to run. For example I use my home computer (Pentium IV 3 Ghz) for 2 hrs. a day on average. From the part of the world I come neiher electricity nor computers are cheap, so it is just not possible to keep computers running 24/7. There are a lot of people like me and they will be forced out of this project, which is very unfortunate.

Comment viewing options

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