lots of questions. sorry. can someone please inform me about previous model ownership, task model number ending in _1. i read in another thread that it is advised to only acccept _0.

I very much doubt that there is ANY post advising not to accept a model with a name ending in anything other than zero. If there is, please post a link to it so that a moderator can have a look at it.

Models only exist on peoples computers. Before this, they are just a set of initial start conditions on the download server. These datasets can be issued up to a max of 5 times, but this doesn\'t necessarily happen. It depends on whether the researchers decide that they have enough in this area of values, whether the server has lots of \'new\' datasets left to issue, etc.

Some times a model will fail on an AMD processor, but not on an Intel, and vice versa. Sometimes it fails because the person\'s computer isn\'t sufficiently stable.

Just looking at the posted computer specs doesn\'t give the full answer. There\'s also whether or not it\'s overclocked, how much cooling is being applied to the processor chip, which involves how many case fans it has, and how dust free the processor heat sink is.
And then there is the matter of the rating of the power supply; does it provide the full voltages for continuous intense running, or is it a bare minimum unit, just meant for \'normal home use\'.

There\'s a lot of info listed for people\'s computers and models, but most is only meaningfull for people who have been running the project for months, and have seen how models behave over time. And it\'s handy for \'experienced crunchers\' to be able to see, for those times when people are having difficulty with their models.

please see.

http://bbc.cpdn.org/forum_thread.php?id=4828&nowrap=true#31019


don\'t know how to attach link sorry.

For making links, and lots of other nice things, look at help file on BBcode tags at the top left of the posting window.

Thanks for the reference.
Mo was talking about the possibility of a model crashing again after being re-issued, not advising against crunching one.
After all, that\'s why they ARE re-issued; to \'get a second opinion\'. And a 3rd, 4th, and 5th.

Hi all

I had been advising Peter on the BBC forum to keep network activity suspended apart from when he decides to allow the model to trickle. Doing this means that if your model crashes, you can restore your backup without the server knowing about the crash.

This isn\'t an essential procedure, but I do it myself and advise it for all models, whether new or reissued.

I have never advised anyone not to accept reissued models and have never seen anyone else giving this advice. I think with all due respect that you\'ve misinterpreted what was said.

What we do know is that a lot of reissued models were never even crunched first time round. For one reason or another, the first computer received a model surplus to requirements and the member aborted it.

When models crash, it\'s nearly always due to something wrong on the machine: it\'s unstable, it doesn\'t meet the minimum specs, or the way it\'s used causes a model crash.

We also know that in the past, models have been deliberately sent out to more than one computer to see whether the results (for example from different types of processor) are comparable.

If you have a model whose number ends in _1, I\'m sorry if I gave the mistaken impression that it isn\'t as good as an _0.

---

Cpdn news

sorry for the confusion. im new and dropped out of high schol. thanks for the advice.

No problem at all. Other members may also be wondering about the same thing, and looking back at what I said on the BBC forum, I can see how the misunderstanding arose.

On the forums you\'ll find a good few dropouts, and not just from high school! I\'d better not elaborate on that......

---

Cpdn news