How do I stop this deadline crap?

Caporegime
Joined
12 Mar 2004
Posts
29,962
Location
England
For some reason on the smp core, I keep getting "deadline passed" resulting in all that time, electricity, uranium, fossil fuels and valuble results being wasted. I never set it to accept projects with deadlines yet It gives me about 40 hours to do a wu. Do they think everyone has supercomputers that they leave on 24/7 or something?
 
Energize said:
For some reason on the smp core, I keep getting "deadline passed" resulting in all that time, electricity, uranium, fossil fuels and valuble results being wasted. I never set it to accept projects with deadlines yet It gives me about 40 hours to do a wu. Do they think everyone has supercomputers that they leave on 24/7 or something?

In a word yes it's designed for quad core pc's - the project deadline are 3-4 day so if your not running at lease a fast duel processor on at lease 18h a day your miss the deadlines.
It's a beta client and they need the results fast. As with many here, the windows smp beta client is not only giving poor ppd but people are missing deadlines. Especially if they use their machine for gaming or it's not on 24/7. It's all in the offical FAQ here
 
Energize said:
For some reason on the smp core, I keep getting "deadline passed" resulting in all that time, electricity, uranium, fossil fuels and valuble results being wasted. I never set it to accept projects with deadlines yet It gives me about 40 hours to do a wu. Do they think everyone has supercomputers that they leave on 24/7 or something?


How are you running SMP on a single core 3500+? :/
 
I'm using it on a 1.66 core duo. Which is left on for about 16 hours a day. Well that was a complete waste of time, I guess I will have to go back to running multiple versions of the program, with all the fossil fuel used, it's more likely someone will have died from lung cancer than anyone been saved, clever. :rolleyes:
 
Last edited:
1.66 Core Duo needs to be 24/7. I have one and it only just meets the deadlines most of the time (sometimes misses them). Core 2 Duo might get away with being less than 24/7, but not the Core Duo.
 
Energize said:
I'm using it on a 1.66 core duo. Which is left on for about 16 hours a day. Well that was a complete waste of time, I guess I will have to go back to running multiple versions of the program, with all the fossil fuel used, it's more likely someone will have died from lung cancer than anyone been saved, clever. :rolleyes:

To be honest if you were always missing deadlines you should have stopped. All the information is there, if you didn't read it then it's your own fault. :)
 
You shouldnt have to read loads of info when you try to help people. I still don't understand why they would simply throw away results instead of just adding them at a later time resulting in greater accuracy.
 
Energize said:
You shouldnt have to read loads of info when you try to help people. I still don't understand why they would simply throw away results instead of just adding them at a later time resulting in greater accuracy.

Deadlines are there to prevent people duplicating work for no reason.

Also remember that this is a beta client, so you have to expect to do more reading and watch the client very closely etc.
 
They do have some redundancy in the WUs, but as said it is a BETA client and they need results back ASAP so they can see how well it is going. It far better for the science getting a few similar WUs back quickly than lots of similar WUs back days or weeks later.

When the client comes out of BETA the dealines will probably be increased. At the moment you just have to accept that a very short deadline is the price you pay for being a BETA tester :)
 
I must admit I'm not overly happy with the direction things seem to be going at the moment - the client and servers are capable of recognising if you have a dual or quad core machine yet some of these WUs are even having trouble completing when a fairly high-end machine is on 24/7

The WUs meant for quad-core machines have even shorter deadlines so I don't think it's entirely a case of the work not really being meant for these machines

/semi-rant :p


I'm back on the GPU client anyway - at least if I have to go away for a couple of days I can come back and finish the WU well within the deadline :cool:
 
Berserker said:
1.66 Core Duo needs to be 24/7. I have one and it only just meets the deadlines most of the time (sometimes misses them). Core 2 Duo might get away with being less than 24/7, but not the Core Duo.

Clock speed plays a large part, T2300 might struggle to get to the deadline but a T2600 does it ok. :)
 
SiriusB said:
They do have some redundancy in the WUs, but as said it is a BETA client and they need results back ASAP so they can see how well it is going. It far better for the science getting a few similar WUs back quickly than lots of similar WUs back days or weeks later.

When the client comes out of BETA the dealines will probably be increased. At the moment you just have to accept that a very short deadline is the price you pay for being a BETA tester :)

I disagree, the winsmp client is definately designed for quad and I cannot see them encouraging dual core machines to get onboard as it is the science that is driving this. They need the results very fast for the clients to be of value for their modelling, highend duals and quads will be the future for the science as speed is of the essence, reading the papers at f@h confirms this, I think even highend duals will slowly fade in the not to distant future (for the science that is) as quads and Octs take over. Its the speed which is at the cutting edge not the fodder for the single cores, although this will always be valuable. PS3 with its cells and the gpu clients will take over as they are complimentary to the science due to their architecture (again speed related) so I'm afraid we are going to have to get our check books out for the cause. :)
 
lay-z-boy said:
Clock speed plays a large part, T2300 might struggle to get to the deadline but a T2600 does it ok. :)
Indeed, for the WinSMP client clock speed seems more important than multiplier. I have an E6300 which is very mildly overclocked to just over 2GHz but it crunches through the WUs faster than a stock E6400. Most odd.....
 
I've done a couple of the p2651s on a Core Duo 1.66, and I didn't have a problem at all.

[04:33:53] Completed 65000 out of 500000 steps (13 percent)
..
[05:06:18] Completed 70000 out of 500000 steps (14 percent)
..
[05:38:09] Completed 75000 out of 500000 steps (15 percent)
..
[06:07:56] Completed 80000 out of 500000 steps (16 percent)

So it took about half an hour to do a frame of a 4-day WU, making 50 hours to complete a WU which had a 96-hour deadline - so it could do them fairly comfortably? I dunno if the other WUs are different.

I think you've kinda missed the point of the SMP client.
 
Last edited:
Snapshot said:
Indeed, for the WinSMP client clock speed seems more important than multiplier. I have an E6300 which is very mildly overclocked to just over 2GHz but it crunches through the WUs faster than a stock E6400. Most odd.....

Not really the architecture of the E6300 is better for folding as the 6400 is a slightly poorer relation for the market (in folding terms that is).

Btw, expect to see me removed from your conquest list once the stats settle down dear boy. ;)
 
Mattus said:
I've done a couple of the p2651s on a Core Duo 1.66, and I didn't have a problem at all.
Actually, you're right - I was thinking back to the early days of the OSX SMP client. Deadlines have been relaxed a little since then.
 
Back
Top Bottom