2 replies to this topic

[frenchhorn1]

Hey phasemage,

How are you folding so many WU's???

Anybody got any ideas.

I fold about 12/day with 2 ati 3850's. How do you fold hundreds with only 2 active processors?? -AND- Why are they getting so few points/WU???

-Frenchhorn1

[dnamechanic]

Hey phasemage, How are you folding so many WU's???

Anybody got any ideas.

Thanks for your comment, frenchhorn1.

I noticed this soon after PhaseMage started contributing. I was hoping he would post here, or register, so I could contact him.

It is strange. My only guess would be that he has a defective computer(s) that are failing as soon almost as soon as a work unit is downlaoded. When a failed work unit is returned (sometimes called EUE, for early unit end) , Stanford tries to give partial credit or in some cases no credit but often counts the work unit, with or without partial credit.

Yes, Stanford has him listed as using two computers.



It is possible to fold with more computers than Stanford can recognize (there are at least two ways of doing this). If he is using more than two computers, and one or more are immediately failing work units, that would help explain the phenomenon.

I fold about 12/day with 2 ati 3850's.

Yes and you are racking up the points! Thanks!

How do you fold hundreds with only 2 active processors?? -AND- Why are they getting so few points/WU???

There are no work units listed with so little points as PhaseMage is averaging per work unit. This is not to say his contribution is little. Phase Mage has made substantial contributions to the TLM team.

The lowest credit per work unit that Stanford has listed here is 15 points.

Phase Mage your contributions are appreciated, I hope you (PhaseMage) see this posting and will provide comments.

Edited by dnamechanic, 18 June 2008 - 03:26 AM.

[dnamechanic]

Canaca has come online for the TLM team. I suspect Canaca may have a representative that reads these forums, if not maybe someone here is in communication with them.

There is an anomaly in their work units versus points accumulated.

For example, so far this week:

- 6,128 points
- 4,364 work units

The anomaly is that there are no work units listed with a point value less than 15 points per WU. Canaca's average is less than about 2 point per work unit. There are a few causes of such, see speculation earlier in this thread regarding PhaseMage, who has similar statistics:

Thinking about Canaca being an ISP leads me to think they probably do not have a computer hardware/software problem. I notice that there are only two Canaca folding CPUs reported by Stanford, here.

Based on points reported (1,250) for one of today's work work units, at least one number-crunching machines is a PS3.

It is possible that Canaca is folding with many more CPUs than this and that the CPUs are folding duplicate work units. If this were the case Stanford could credit the completed WU count but they only award points for one folded WU no mattter how many times it may be completed and returned.

It may have been convenient for Canaca to clone the folding client and install it on many computers. When the client installation software is cloned, all computers can look the same to Stanford and they can keep assinging the same work unit until it is finally completed, then they start assigning a different WU. When this occurs all the cloned machines may be doing duplicate work. Stanford IDs computers based on a UserID in the registry and a Machine ID in the folding client.

There are workarounds for this ID situation that are not difficult.

If anyone can communicate with Canaca, forward this message to them.

If the issue is not an identity issue, then Canaca may have some other issue that probably needs to be addressed.

Anyhow, Canaca is doing well and helping out, and that is appreciated.