Rosetta@Home scoring UPDATE! (they heard me!)

[Insidious]

I am seeing wildly varying scoring in my Rosetta results. When I compare the awarded score and compare it to the completion time (for the same computer) it seems like this is WAY outside the realm of (normal variance)

You can see an example here

Look at the first and last results shown on the page :shocked:

Not sure how I feel about this...... any opinions (or perhaps a warm pat on the head for a points hound. )

-Sid

:beer:

 

[Smoke]

That last one was really unusual ... that's for sure.

Hey, don't complain about success is the way I feel about it.

Good job! :beer:

 

[natethegreat]

I got 1 word for ya Scot:

:evil:QMD:evil:

 

[Insidious]

Originally posted by: natethegreat
I got 1 word for ya Scot:

:evil:QMD:evil:

LOL

it is starting to look like maybe I am just getting too hard to please in my old age.

LOL

If I'm reading these right, I should just smile and enjoy my little pile

-Sid

 

[petrusbroder]

Hmm, interesting - but I'd be happy for all the points I'd get ... no idea why this is so ... :Q

 

[BlackMountainCow]

Now, for the "younger" folks in here, what's QMD? I only that one as a F@H core.

 

[natethegreat]

Originally posted by: BlackMountainCow
Now, for the "younger" folks in here, what's QMD? I only that one as a F@H core.

QMD's were a special F@H wu that was given a huge point advantage to crunch based on the ammount of resources it required - the catch was that it was only able to run on Intel chips because of a licensing issue:roll: The bonus was big enough that AMD rigs were not point competitive

 

[BlackMountainCow]

Ahhhh .... now I get it. Yes, in that respect, the Rosey problem is indeed a minor issue.

 

[Insidious]

indeed

Compared to the QMD fiasco.... this is nothing.

So are all of you seeing similar stuff on your results pages?

-Sid





/me sighs... I still miss F@H from time to time...

 

[GLeeM]

Originally posted by: Insidious
/me sighs... I still miss F@H from time to time...

We miss you more than from time to time

There are some of those great 364 points 14## series WUs out now on -advmethods.

:roll: C2Ds with 4MB cache chew through them in about a tenth the time my P4 takes :roll:

I doubt there will be any more QMDs.

 

[Insidious]

Originally posted by: GLeeM

Originally posted by: Insidious
/me sighs... I still miss F@H from time to time...

We miss you more than from time to time

There are some of those great 364 points 14## series WUs out now on -advmethods.

:roll: C2Ds with 4MB cache chew through them in about a tenth the time my P4 takes :roll:

I doubt there will be any more QMDs.

Thanks GleeM :beer:

one of these days........

 

[imaheadcase]

I don't know about anyone else, but i started this R@H just about a week ago but i get a LOT of ones stuck at %1 and never go higher, i have to get new stuff almost every day now manually.
Its not like its stuck, cause they go like 20+ hours before i notice sometimes. My old P4 sucks as it is, hate getting ones that waste the time

 

[Insidious]

OK, this sounds Good!

I saw this post on the (just released) newest Rosey version...

Message 30397 - Posted 31 Oct 2006 22:50:39 UTC - in response to Message ID 29979.
In Rosetta 5.3.6:

The workunits with varying bond geometry were slow in the previous release, and led to some very uneven crediting. We've put in features that significantly accelerate those workunits. There are also fixes that will allow us to test more accurate energy functions.

There are some fixes for graphics to make the native fold always show up. Finally, we're trying out a new output format that will enable us to send back results for designed proteins (e.g., for HIV vaccines) in compressed format.

Good news indeed :beer:

-Sid

 

[Spacehead]

Originally posted by: imaheadcase
I don't know about anyone else, but i started this R@H just about a week ago but i get a LOT of ones stuck at %1 and never go higher, i have to get new stuff almost every day now manually.
Its not like its stuck, cause they go like 20+ hours before i notice sometimes. My old P4 sucks as it is, hate getting ones that waste the time

Post that in one of BlackMountainCow's R@H stats thread & i'll bet he can help you out.