CurrentStruc 0 21 124 106 1 4 7.374 -1404.343 -399.594 -319.077 14287885.000 1.250 2.300 764.752 -------------------------HHHH-------------------------------------------------------------------
Printable View
CurrentStruc 0 21 124 106 1 4 7.374 -1404.343 -399.594 -319.077 14287885.000 1.250 2.300 764.752 -------------------------HHHH-------------------------------------------------------------------
Howard;
I've done a bit of elementary stats work on our top 10 results. A few things come out of this:
1) The best rmsd reported on the website in the top 10 list differs from that posted within the text details of the generation. I assume from this that you're running your own algorithm on the the best ones and coming up with your own #. Since we're spending so much time on a single structure sometimes, why aren't we using the same algorithm?
2) # 9 on the list is seriously whacked. The top 10 list says 175 gens have been completed but the text details only says 66.
3) 7 out of the top 10 have data for beyond gen 100. Of these 7 only 2 found their best results near the end of their efforts. On average, the clients found their best result at 58% of their total effort. If you remove the two that did best towards the end from the equation it drops to 44% effort.
My proposal is as follows:
1) Make the # of generations dynamic based on the results of the gen.
2) Always fold out to 100 gens.
3) take the best result from the 100 and keep going another 20%. If within that 20% you do not find a better fold, abandon the set and start all over again.
4) If you do find a better one within the 20%, go out another 20%. This way we will only waste about 20% of our effort.
By my calculations the two sets in the top 10 that found their best result towards the end of their work so far would still have been included. But the others would have given up sooner which we now know to be a good thing.
This is similar to a suggestion I made a while back, but now I have some stats to back up my position.
ms
CurrentStruc 0 36 124 127 1 13 6.924 -90.197 1460.041 418.031 19083954.000 2.500 4.800 25174.842 -------------------HHHH-------------------------------------------------------------------------
This one seems to have run into a brick wall. On the 127th generation, the laxness graphs in dfGUI are 50%, 68%, and 100%.
Hey Angus, do you happen to notice the response of the computer becomes jerky at this point?
Mine seemed to be...
And what is wrong with #9 on the list ?! :mad:Quote:
Originally posted by Brian the Roman
Howard;
I've done a bit of elementary stats work on our top 10 results. A few things come out of this:
2) # 9 on the list is seriously whacked. The top 10 list says 175 gens have been completed but the text details only says 66.
G
Georgina;
as I said before (and you quoted), #9 on the list says it has completed 171 generations on the top 10 list. If you click on the 'View Details' link for it , however, all the stats are only reflected to generation 72. These two should match pretty closely, the only difference being the updates being out of synch for an hour or so. It's been like it is for over a day so that can't be it.
Human Factors
With the old client, a participant looking at the display can see structures being built -- and he knows that the increment he sees will be added to his "contribution value". To me this accumulating credit provides continuous positive feedback.
With the beta, the participant knows that completing a generation will add to his "contribution value", but he does not necessarily know by how much he will be credited, nor does he see that value CONTINUOUSLY accumulating. [Instead, some generations take a LONG time to complete, leading to a sense of "contribution value" NOT accumulating.] Seems to me that the beta manages to convey the message "Don't expect continuous encouragement".
mikus
Perhaps we should randomly print out subliminal messages such as 'GOOD JOB' or 'KEEP UP THE GOOD WORK'. :haddock:Quote:
Originally posted by Mikus
Human Factors
With the old client, a participant looking at the display can see structures being built -- and he knows that the increment he sees will be added to his "contribution value". To me this accumulating credit provides continuous positive feedback.
With the beta, the participant knows that completing a generation will add to his "contribution value", but he does not necessarily know by how much he will be credited, nor does he see that value CONTINUOUSLY accumulating. [Instead, some generations take a LONG time to complete, leading to a sense of "contribution value" NOT accumulating.] Seems to me that the beta manages to convey the message "Don't expect continuous encouragement".
mikus
CurrentStruc 0 51 124 239 1 35 6.518 -2691.339 -994.826 -1919.212 189732912.000 1.600 3.000 2034.258
It's interesting to see the client actually dropping after generation 200.. and it's been falling between gen 211-241. (will have to view the movie to see the hills and valleys..) Alas, it probably won't get down to the top ten scores that are already on the board. *sniff*
Howard, got the following error:
========================[ Apr 12, 2003 6:48 PM ]========================
ERROR: [010.003] {taskapi.c, line 1217} [ReadServerResponse] Timeout waiting for response, got 13002 chars.
My client was not crunching after this error.
Well, I have decided to give the Beta a try, but have no clue as to how ot get it up and running. downloading the Windows client from the link on the first page gives me a folder with 4 files. Foldtrajlite. readme native and protein. This does not seem right..as I am pretty dumb, which results in my grumpiness, could someone point me in the right direction :(
You need to first, get a current version of the non-beta client. Unpack that to a directory somewhere. Then, unpack the beta zipfile to the same place, overwriting foldtrajlite.exe, protein.trj, and native.val, and adding readme.txt.
Don't bother starting up foldit.bat until you've updated the other files.
Hmm, that is odd. Can you descrbe what you/it was doing just prior to the error? Was it trying to download something? (it appears so, the message says it got 13002 bytes of the download before Timing out...)Quote:
Originally posted by m0ti
Howard, got the following error:
========================[ Apr 12, 2003 6:48 PM ]========================
ERROR: [010.003] {taskapi.c, line 1217} [ReadServerResponse] Timeout waiting for response, got 13002 chars.
My client was not crunching after this error.
Sorry, I don't know what it was doing. Came by the computer and noticed that it wasn't folding, so I started it up again.
I just thought to check the error log this morning.
I am seeing a strange thing with the stats. Right now the Stats page shows that my best RMSD is 5.955425 which I believe was found by the first beta6 client I started (which has since finished all 250 generations).
My third beta6 client which is now on generation 182 shows an RMSD of 5.392 in my GUI as the best it has seen in the progress.txt file. (I should modify dfGUI so it also displays the generation # that the best so far was found in).
At least 2 or 3 generations have been completed and uploaded since I first noticed the 5.392 value in the GUI but there is nothing in the stats indicating this. Any idea what might have happened?
Jeff.
:thumbs:Quote:
Originally posted by Digital Parasite
I should modify dfGUI so it also displays the generation # that the best so far was found in.
Jeff.
Beta 6 up & running, thanks all :cheers:
Howard: Would it be possible to have that info in the progress.txt file? Ie. the best structure currently found by client. That would also make the job much easier for Jeff (dfGUI).Quote:
Originally posted by Digital Parasite
My third beta6 client which is now on generation 182 shows an RMSD of 5.392 in my GUI as the best it has seen in the progress.txt file. (I should modify dfGUI so it also displays the generation # that the best so far was found in).
Well, the Beta is running well, on my most unreliable PC, so well done guys. And DFGUI runs well too, thnx Jeff. As OCworkbench has suffered a bit of a collapse, I may transfer my dual 2400 MP over to the Beta too, will have a think about it overnight. Our Team will be back with a vengence once this goes Final Release :p
Apparently your best uploaded struc is 5.95. Where do you get the 5.39 value from? Is it using the same handle? Are you sure its not the old beta running (beta 5 or less)? (check file dates)Quote:
Originally posted by Digital Parasite
I am seeing a strange thing with the stats. Right now the Stats page shows that my best RMSD is 5.955425 which I believe was found by the first beta6 client I started (which has since finished all 250 generations).
My third beta6 client which is now on generation 182 shows an RMSD of 5.392 in my GUI as the best it has seen in the progress.txt file. (I should modify dfGUI so it also displays the generation # that the best so far was found in).
At least 2 or 3 generations have been completed and uploaded since I first noticed the 5.392 value in the GUI but there is nothing in the stats indicating this. Any idea what might have happened?
Jeff.
Crap! The date on my foldtrajlite.exe file is March 18th. I downloaded the beta .zip from the DF web site and saved it to my DF directory.Quote:
Originally posted by Brian the Fist
Apparently your best uploaded struc is 5.95. Where do you get the 5.39 value from? Is it using the same handle? Are you sure its not the old beta running (beta 5 or less)? (check file dates)
I just tried again and I am getting the same .zip file dated Mar. 18th so I think my IE cached the download from before so when I came back and it said it had finished downloading it must have just given me the old cached .zip file.
Damn, waste of a week!
Jeff.
Can you fix this to autoupdate when there is a newer beta please? :)Quote:
Originally posted by Brian the Fist
Are you sure its not the old beta running (beta 5 or less)? (check file dates)
We can beta test that part to. :D
Instrumentation
Particularly with the beta client, you have an opportunity to put in "reporting routines" which would record the significant values at "waypoints" in the clent's decision-making processes. Then, at the expense of adding more server-side data analysis, you could accumulate specific "how well does the client function" information.
Given that your overall purpose is to validate the approach you are taking to 'protein folding', what I'm suggesting is taking maximum advantage of the data-gathering that the client could do. The more self-description the client preserves (via the server), the better your ability to analyze how close the client comes to being able to do what you hoped.
mikus
How bad is it when the laxness levels get real high? I just came home and clicked on dfGUI and the laxness levels were at 74%, 100%, and 100%.
Just to let everyone know, there is no real maximum value that I'm aware of for the laxness values. I chose values that Howard suggested would be considered very lax and picked those as an artificial 100% mark for the bar graph. If it says 100% that means it is at least the value I selected or higher.
As for if it is bad or not, I will have to let Howard answer that one.
Jeff.
Do it! ;) And see how different the rates are on the two cpus, when they start running into various amounts of trouble folding tight folds.. :)Quote:
Originally posted by Grumpy
Well, the Beta is running well, on my most unreliable PC, so well done guys. ...I may transfer my dual 2400 MP over to the Beta too, :p
There's a claim that those horrible generations that take forever :bang: lead to great low RMS scores..
if so, I wish my Beta machines got a set of those horrible generations.. ;)
*-BennyRop
Well, the Dual 2400 MP is Folding. Will see how they travel over a day or two. See if they can crash the Tyan...ruddy sound cards on SMP machines :swear:
xp1800/1.5 GHz takes right at 8 days to finish a protein with the beta. Thanks for joining! :)
8 D.A.Y.S. :shocked:
Hey everyone, I had just a little bit of time to work on dfGUI so I added a couple of features.
You can get the latest beta here:
http://gilchrist.ca/jeff/dfGUI/dfGUIv22beta.zip
v2.2beta3 (Apr. 15, 2003)
- Benchmark output to file info now shows current generation
- Best energy so far now shows which generation # it was found in.
Enjoy,
Jeff.
:spank: Just after I posted links to your files. Time to edit. Doh, same filename...Kool
Hmmm, CPU2 on the Dual 2400MP has almost beaten my 1800XP RMS.. been running for 2 hours & the 1800XP a day and 1/2 :rolleyes:
Update: Hope it is OK Jeff to post the Image on the OCworkbench Folding Forums :cheers:
Well, for the curious, my Dual 2400MP has been Folding 2 Clients for 11 hours 40 minutes.
Client 1: Generation 36 RMS 7.423
Client 2 : Generation 34 RMS 5.828
The 1800Xp is Generation74 RMS 7.616
Running time is 1:22:21:34
:|party|:
Howard,Quote:
Originally posted by Brian the Fist
So anyhow that will be the NEXT test, after this one (its never good to change too many things at once after all). We'll let this one go for 10 days or so again first to see how it does, and we find that 'fixing' the helices is not needed.
Just a reminder that for some of us, a long 3 day weekend is coming. If the next test is posted after Thursday afternoon, I may not be able to change my PCs until Monday.
Thank You.
G
a DFGUI for linux question:
any idea what the following might mean?
[derek@Scoofy12 dfGUI-2.2beta2]$ dfGUI
(dfGUI:20533): GLib-GObject-WARNING **: invalid cast from `GtkWindow' to `GtkDialog'
(dfGUI:20533): Gtk-CRITICAL **: file gtkdialog.c: line 943 (gtk_dialog_run): assertion `GTK_IS_DIALOG (dialog)' failed
Segmentation fault
Amazing what the beta client can do in a little time if you get a lucky pick in generation 0. :)Quote:
Originally posted by Grumpy
my Dual 2400MP has been Folding 2 Clients for 11 hours 40 minutes.
Client 2 : Generation 34 RMS 5.828
Bah, up to Gen 54 and it has not improved, had to hit bottom eventually..knew it was too good to be true. Even if it has bottomed, it gives the Powers That Be another nice Protein to have a look at ;)
As long as you crunch for Team Stir Fry - no problem :jester:Quote:
Originally posted by Grumpy
Update: Hope it is OK Jeff to post the Image on the OCworkbench Folding Forums :cheers:
Argh ! Poachers, I will never forgive you for eating Bambi :spank:
Hmm, it looks like the problem is in the on_dfGUIform_realize function, though that's just a guess.Quote:
Originally posted by Scoofy12
a DFGUI for linux question:
any idea what the following might mean?
[derek@Scoofy12 dfGUI-2.2beta2]$ dfGUI
(dfGUI:20533): GLib-GObject-WARNING **: invalid cast from `GtkWindow' to `GtkDialog'
(dfGUI:20533): Gtk-CRITICAL **: file gtkdialog.c: line 943 (gtk_dialog_run): assertion `GTK_IS_DIALOG (dialog)' failed
Segmentation fault
Oh, hang on, I bet I see the bug. It's hitting this when trying to display the config window the first time. It's using gtk_dialog_run() to do that, because the config window's modal and I had a thinko when writing that bit of code -- I forgot that gtk_widget_show(), if the widget is "modal", runs it modally. Anyway, the actual problem is that gtk_dialog_run needs a pointer to a GtkDialog, but the config window "object" isn't a dialog. It's just a window.
Anyway, there are other issues with it too. Fixed packages are up on the site.
I'm still new to the world of beta testing (and DF in general), so forgive me for askin stupid questions here :-p
I downloaded beta 6 last night, set it running, and lost my net connection so it started buffering files. Net was restored this morning, but DF still won't upload files. Tried deleting filelist and *.log.bz2 files to start over, but DF still isn't uploading... it just keeps buffering. Any ideas?
Also, where is foldtrajlite.exe supposed to be located? I was gonna try out the dfGUI beta, but I can't seem to locate the file it's looking for....