Folding@Home: public prerelease of version 5 client

[GLeeM]

This public prerelease expires on August 30, 2004. You may want to wait for the final release. (Probably out soon)

Some of the new features of the console: during config you can choose: Request work units without deadlines (no-pref/no/yes) - no more confusing GaH option.

A built-in run as service configuration: Launch automatically at machine startup, installing this as a service (yes/no)
quote: "service accepts flags now, wherever you input them. Also, it interacts with the Windows Services administrative tool properly (pause, stop, etc.), in addition to Sc but not Net." Whatever all that means.

And large WUs. Config asks: Allow receipt of work assignments and return of work results greater than 5MB in size (such work units may have large memory demands) (no/yes) [no]?
There probably is ram minimum and connection speed minimum.

Also: Pause if battery power is being used (useful for laptops) (no/yes)

Read all about it here.

 

[amdxborg]

Thanx for the info GLeeM!

 

[raynman68]

Thanks for the info GLeeM. :beer::beer::beer::beer:

I am already running my rigs as a service, but it will be nice to not have to configure machines as a service as I add more.

Some of the other features sound intersting as well including the OpenBSD support.

Robert
And the men who hold high places
Must be the ones who start

 

[ProviaFan]

Danke, mein Herr.

 

[MDE]

Nice to see the new features, especially the pause on battery power for the ol' lapburner.

 

[lobadobadingdong]

any way to benchmark it to see if it runs any faster than the old client?

 

[ProviaFan]

Originally posted by: lobadobadingdong
any way to benchmark it to see if it runs any faster than the old client?

I would doubt that there would be any speed difference, as long as both use the same WU and same core.

 

[Gravity]

:thumbsup:

 

[GLeeM]

Originally posted by: lobadobadingdong
any way to benchmark it to see if it runs any faster than the old client?

The core is hand coded assembly and Stanford says it can't be made faster. The core is what does the calculations. The client does the other stuff.

They will probably adjust the "simulated time per iteration" in some large WUs to get more info per computer time. This will result in a larger percent of early ends for these WUs, but an early end is one of the possible outcomes and still gives them valuable data. (the warning will be reworded)

 

[InlineFive]

Originally posted by: MDE
Nice to see the new features, especially the pause on battery power for the ol' lapburner.

Implemented, no doubt, for all the Banias and Dothan people.