Welcome to the MacNN Forums.

enola · Mar 18, 2003, 08:43 PM

I see this in the log file:

[02:38:12] Protein: Des2/pdb1wel.44.spa

And most of the log entries look like this which looks really different than the previous protein I was working on:

[02:00:48] [SPG] 10.0
[02:05:37] [SPG] 20.0

I don't see this protein in the "Project Summary" page. The previous one I had, P628_vsd_nat, was there. I'm just curious. I wanted to see how many points this one was worth.

Also, when do I show up in our stats pages?

reader50 · Mar 18, 2003, 09:00 PM

That is a Genome protein.

Watch the crunch core that is required for each protein:

Core_65 - Folding (tinker)
Core_78 - Folding (gromacs)
Core_c9 - Genome
Core_ca - Genome

Crunching genome will never cause you to show in our Folding stats.

Once you return a Folding unit on team #16, the project server will take a certain amount of time to credit the unit. Usually this happens almost right away, but I've seen it take a day before it shows up.

When the server has credited your account, it will show on the next Folding stats run, which is every three hours. Our Folding stats try every 30 minutes for project updates.

Scotttheking · Mar 18, 2003, 09:20 PM

Originally posted by reader50:

Crunching genome will never cause you to show in our Folding stats.

Not true.

GAH and FAH work will both be credited to the team. Don't worry about what you get, it will all count.

enola · Mar 18, 2003, 09:21 PM

In the advanced settings I noticed there's a client type setting with options for no preference, folding, and genome. Mine is set for no preference.

Should I specify folding from now on? Should I just quit the GAH work unit and make it go get a FAH work unit?

enola · Mar 18, 2003, 09:28 PM

Originally posted by Scotttheking:
Don't worry about what you get, it will all count.

You just posted right before I did! Okay, will let it keep crunching the GAH unit. Thanks.

reader50 · Mar 18, 2003, 10:12 PM

According to posts I've read on PC boards, you specify two team numbers. One for Folding, one for Genome. Each type of work gets credited to the proper team.

If no Genome team is specified, I'm told that Genome credit does not go to the Folding team. It would go to wherever non-team Genome credits go. I'd assume a default team #0 like in Folding, but it might not get credited anywhere.

I'd suggest setting prefs to Folding only.

Scotttheking · Mar 18, 2003, 10:50 PM

Originally posted by reader50:
According to posts I've read on PC boards, you specify two team numbers. One for Folding, one for Genome. Each type of work gets credited to the proper team.

If no Genome team is specified, I'm told that Genome credit does not go to the Folding team. It would go to wherever non-team Genome credits go. I'd assume a default team #0 like in Folding, but it might not get credited anywhere.

I'd suggest setting prefs to Folding only.

From the project forums:

Absolutely! As long as your client is set up in the configuration with a F@H team number, all G@H wu's you do will get credit on your F@H team stats as the generic G@H project 799.

reader50 · Mar 18, 2003, 11:04 PM

That quote came from here. Unfortunately, the person who said that is a volunteer site mod, not a member of the Pande Group.

I've read on PC boards about the combi client "eating" the credit for Genome units, if no Genome team number was specified. Be careful, and let us know if the credit actually posts.

Shaktai · 11:52 PM

Originally posted by reader50:
That quote came from here. Unfortunately, the person who said that is a volunteer site mod, not a member of the Pande Group.

I've read on PC boards about the combi client "eating" the credit for Genome units, if no Genome team number was specified. Be careful, and let us know if the credit actually posts.

Using the CLI client, I don't recall an option to list two team numbers. However I do have 12 genome units that have contributed under 799. Near as I can tell, they have all contributed points and generally I find that genome units score slightly higher then tinker cores, but lower then gromacs. All of my genome have used the ca core, not the c9.

Enola, the settings I use with the CLI is FAH preferred with the -advmethods flag on the CLI. Hope that helps. Using the FAH preferred will get you more Folding units and the -advmethods will get you gromacs when they are available with an occasional advanced methods genome. If you have a SIMD enabled chip (SSE or 3Dnow) you will get great performance off of gromacs cores. Now if they would just release it for Macs too. (Altivec should do very well at Gromacs)

Edit: Looks like you last uploaded at 2003-03-18 21:00:47 and have posted a 624 and a 628. Don't see you in the team stats yet, but you should be soon.

enola · Mar 19, 2003, 01:22 AM

Originally posted by Shaktai:
If you have a SIMD enabled chip (SSE or 3Dnow) you will get great performance off of gromacs cores.

I have a 2.5Ghz P4 (win2k) and a 1.7Ghz P4 (NT). Are they SSE enabled? I'm using the gui client with those two. Would it be a matter of adding "-advmethod" to the shortcut? Oh, and just to round things out I've got one CPU on the dual Mac running FAH. The other one is back on dfold.

The next protein the Mac client picked up was p636_villin_c0_ext, a 48.3 pointer and is taking about 15 minutes/frame. The 624 I had previously was worth 19.7 points at about 11.5 minutes/frame.

What determines what kind of protien you get? I'm still trying to figure out how this project works. I'm finding it takes a little getting used to waiting days for WU to complete.

Shaktai · 02:47 AM

Originally posted by enola:
What determines what kind of protien you get? I'm still trying to figure out how this project works. I'm finding it takes a little getting used to waiting days for WU to complete.

Yes the P4's are SSE enabled. I think with the GUI, there is an "advmethods checkbox in the setup or preferences that will turn it on. I just use the console on my Athlons and start it up using FAH3Console -advmethods. I think you can do that in the shortcut too.

Proteins vary widely. You not only want to look at the time per frame, but the number of frames which typically vary from 100 to 400. The points awarded are based upon a "Benchmark machine" which I think is a 500mhz Celeron. That is what determines the deadline for each protein to be returned as well as the number of points. However performance can vary somewhat even among the same protein groups. The only thing that determines what proteins you get is luck, your preference setting, the -advmethods flag, your platform and oh yes did I mention luck.

Yes this project is different, and it requires a bit more patience. Nothing like getting a 73 pointer on my iBook for a protein that takes "days". Of course it is always nice when it finishes. More importantly though the project is showing some very real scientific progress.

*** Looks like you finally showed up in the team stats.

DAlex · Mar 19, 2003, 03:07 PM

Hi enola, good to see you on the F@H side of the fence - you got any huge surprises lined up? A whole heap of brothers ready to sweep Scott off the top of the pile

DAlex

enola · Mar 19, 2003, 08:18 PM

Originally posted by DAlex:
Hi enola, good to see you on the F@H side of the fence - you got any huge surprises lined up? A whole heap of brothers ready to sweep Scott off the top of the pile

No, no surprises although I did ask him today if he wouldn't mind running it on his desktop at work or at home. When I went home this past Christmas my brother made a point of introducing me to the guy that administers that big server we hijacked for a while for dfold. The guy's first words to me were "Oh, you're THAT sister." We had a good laugh over that.

Hey, Shaktai, one of my PC's picked up a gromacs core today.