Wiggle on ED puzzle is still too slow (even on LOW)

Case number:954892-996746
Topic:Developer Preview
Opened by:Timo van der Laan
Status:Open
Type:DevPrev Issue
Opened on:Sunday, January 26, 2014 - 23:56
Last modified:Friday, June 27, 2014 - 00:46

Today I did a timing test on the ED puzzle. My previous timing tests were on 663.
One wiggle(6) on LOW takes about 3 minutes and that is about 2 1/2 times the duration I measured on 663.

My conclusion: even on LOW, big ED puzzles will be very, very slow. Too slow for non top end computing power. For a normal kind of puzzle the speed is playable.

(Sun, 01/26/2014 - 23:56  |  8 comments)


Joined: 05/19/2009
Groups: Contenders

A bump because the client still is too slow.

Joined: 05/19/2009
Groups: Contenders

The current client is so excruciatingly slow that I am contemplating to stop folding. No joy.

beta_helix's picture
User offline. Last seen 1 day 6 hours ago. Offline
Joined: 05/09/2008
Groups: None

BP, just checking that you are only referring to ED, right?

Joined: 05/19/2009
Groups: Contenders

Thanks for checking beta. No, in general the client is just very sluggish compared to what it used to be. I have checked and found the filters to be slowing things down, especially in 919.

beta_helix's picture
User offline. Last seen 1 day 6 hours ago. Offline
Joined: 05/09/2008
Groups: None

Darn filters!
It's a pity because those filters have been very useful results-wise: http://fold.it/portal/node/998024

But the CASP puzzles haven't suffered this sluggishness?

spmm's picture
User offline. Last seen 49 weeks 1 day ago. Offline
Joined: 08/05/2010
Groups: Void Crushers

imo the filters also seem to create 'holes' in the score where the protein is actually gaining slightly but then the filter drops by 10 or 15 points for a short space of time, and scripts just reset back to the highest score and try again. Some scripts 'disable slow filters' but that doesn't seem to help either and turning filters off and on again continually doesn't seem to be all that productive.

The filters seem to be very jerky especially when you are in later stages of the game and the filter score is bouncing between 510, 520 and 540 on a single local wiggle, so the script can't do its job.

Joined: 05/19/2009
Groups: Contenders

depends on the size, bigger ones tend to be slower. The filters are apparently the bottleneck.

frood66's picture
User offline. Last seen 6 hours 40 min ago. Offline
Joined: 09/20/2011
Groups: Marvin's bunch

u don't want to know about my speed here - see feedback elsewhere on the speed test. Getting close to call now...carry on? Or find something else to do.

I'm weirdly fascinated to know why the game aspect of foldit has been knowingly destroyed for so many players.

Sitemap

Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, RosettaCommons