New to folding, ready to get started, just a few questions please!

Status
Not open for further replies.
hyt3k9

hyt3k9

2[H]4U
Joined
Aug 26, 2006
Messages
2,240
Okay, I wanna start folding for team [H] #33.
I am familiar with the Wiki on DC computing.. and understand what it is all about, and I too, want to make a difference with my CPU cycles.
I have read tons of threads, including the "[H]orde|Central - Everything DC (FAQ's, Guides, Tips and Tricks)"
Lastly, I also have an older PC I am dedicating to this process.

Questions:

1) How does my hardware stack up? I would like to know mostly, would this thing be even worth it as far as getting the processes in by timeline, and PPD?

CPU: Celeron 2.8GHZ, Socket 478 - SL77T w/ D1 stepping - Stock HSF
Motherboard: Intel D865GVHZ mATX
Memory: 512MB (2x256MB) PC2700 Infineon, SDRAM
VGA: Onboard video, Intel Extreme Graphics 2
HDD: 40GB Seagate Barracuda/IDE 7200.9
Optical: CD/RW Lite-on generic
Case: Ultra Microfly SFF
PSU: Sparkle FSP-300W
[/B]OS: Windows XP Professional SP2

2) I plan to run Windows, since I am not at all familiar with Linux. I read a bunch of posts about people saying not to use the GUI programs, but I know nothing about other options at this point. I read that there are SMP and GPU versions of the client. I assume after reading the Wiki on the SMP that is for Dual core procs.. I would appreciate further information on this.

3) As far as the process goes when your interested in this.. is build your machine, install your OS and the DC client you want to run.. and then sign up for the [H] team.. I am alittle fuzzy about the last step in this process. I saw where it looked like you need to register for a profile page on the [H]ard for Humanity page. Other than that, I am not sure how to get started when I get my OS/DC client up and running. Could anyone clarify this a little for me?

Lastly, Thanks a ton in advance on this information I am requesting. I admire all of you that have been doing this for a long time, and to the [H] community that is currently participating.
 
Your machine will run the console (recommended version) fine.
http://folding.stanford.edu/download.html
Its the last one in the grey section. The Graphical version sometimes has issues and folds a bit slower. The SMP is for multi core (or processor) machines. The GPU version is for folding on ATI x1800 or better graphics cards (maybe lower cards but I'm not sure but somebody will correct me if I'm wrong) and the is also a version to fold on the PS3.
 
1. Its not the fastest around, but it will do just fine. You don't need to run the "timeless" units, that is for slower machines.

2. Yes, SMP is for multi-core (or dual processor) systems.

3. Download a client from that page, you want the no-nonsense client. After you install it, run the fah.exe program and it will configure it the first time it is run. enter your name and team 33 here (this is how you "join" team 33 from f@h's perspective). Tell it to run as a service so you don't have to run it manually. It will start running after setup, but you can close that and start the service (or reboot). When it finishes working on a unit, it will send it back to stanford along with your name and the team you entered.

You can use a monitoring program to check how it is progressing on a work unit (they can take hours or days to complete), or just look at the fahlog.txt file in the program folder.
 
1. You'll be looking at around 100-150 PpD.
If you run slower than that then somethings wrong.
You could run faster with a good set of protiens.
The timeless work-units are no longer run so thats not an option.

2. Run the console version.
Used EMIII to monitor the progress of the client.
It adds all the funtions of the graphical client without the bug.

3. When you first run the client, you will need to configure it to add your folding name and team number (33).
After that just let it run.
You will only show up on any of the stats page once you've turned in a work-unit.
Also Stanford at the moment is playing around with the stats server so it may take a day before you show up on the [H]ordes stats after that.

The registration at [H]orde Central just adds you to the [H]ordes private database not Stanfords.

Luck ........... :D
 
AWESOME guys!!! Thanks for the feedback.. I was hoping a for some detailed information and you didn't disappoint. :)
I will get this thing up and running tonight, and will add questions here, in this thread if I come across any setting it up. (Although it sounds very straight forward) ;)

EDIT: ""8 Boxen folding at +1,000 Points per Day per Boxen"" (Tigerbiten, what kind of specs do these machines have? What kind of hardware does it take to fold 1,000 PPD?) Just curious.. I have 2 other machines I could use.. concerns at this point are power consumption is all.
 
A stock E6300 should get you 1000 ppd running the SMP client. Overclock it and you'll get more.
 
hyt3k9 said:
EDIT: ""8 Boxen folding at +1,000 Points per Day per Boxen"" (Tigerbiten, what kind of specs do these machines have? What kind of hardware does it take to fold 1,000 PPD?) Just curious.. I have 2 other machines I could use.. concerns at this point are power consumption is all.
Click to expand...

The farm at full output was..........
Five X2's @2.2Ghz running the CPU client with X1950xtx'x in running the GPU client, good for ~875 PpD each.
A C2D @2.76Ghz with dual X1950xtx's running twin GPU clients, good for ~1,350 PpD.
A dual Opteron 250 @2.4Ghz running the Linux SMP client, good for ~900 PpD.
A dual Opteron 285 @2.6Ghz running the Linux SMP client, good for ~2,000 PpD.
So 5x875+1350+900+2,000= ~8,625 PpD.
Power useage is close to 2.5 KwH .......... :eek:

At the moment only half of the farm is running as I play around with the Windozes SMP client.
But it looks like I'd drop ~1,000 PpD running that on the X2's for a saveing of ~0.5 KwH.
So not sure which way to go.

Also looking at dual 5355's for a +4k PpD boxen .......... :D

Luck ............. :D
 
Tigerbiten said:
1. You'll be looking at around 100-150 PpD.
If you run slower than that then somethings wrong.
Click to expand...

I will check with this after it has been running a few days, to see what the average PPD are, and if less than what you mentioned, I will let you guys know.

Tigerbiten said:
2. Run the console version.
Used EMIII to monitor the progress of the client.
It adds all the funtions of the graphical client without the bug.
Click to expand...

Okay, in process of downloading it now. Basically, I have a fresh version of XP Pro installed/patched and only two programs going to be on this are the F@H text console and the EIII

Tigerbiten said:
3. When you first run the client, you will need to configure it to add your folding name and team number (33).
After that just let it run.
You will only show up on any of the stats page once you've turned in a work-unit.
Click to expand...

Okay thanks for the heads up, would have wondered why I wasn't showing up right away.

Tigerbiten said:
The registration at [H]orde Central just adds you to the [H]ordes private database not Stanfords.
Click to expand...

Do I have to register with Stanford as well? I thought you just downloaded whichever version of the client and plugged in your information when you installed it. I know you have to register separately with the [H]orde stat page, but didn't realize I had to register with Stanford as well... will look into this.


Thanks to all, for the warm welcome. I got my boxen built last night, and in process of getting the OS loaded and F@H client downloaded and installed. Here is a pic of her:

Picture.jpg


Anyone know how long I have to wait to link my F@H stats to my forum signature. I would "guess" this is after my first WU is processed, like Tigerbiten mentioned above. :)
 
hyt3k9 said:
Do I have to register with Stanford as well? I thought you just downloaded whichever version of the client and plugged in your information when you installed it. I know you have to register separately with the [H]orde stat page, but didn't realize I had to register with Stanford as well... will look into this.

Anyone know how long I have to wait to link my F@H stats to my forum signature. I would "guess" this is after my first WU is processed, like Tigerbiten mentioned above. :)
Click to expand...

When you install the client and first connect to Stanford, that is, in essence, your registration with them. No other input from you does Stanford need.

Stats can't be shown in the [H]ard Forum sigs because links are not allowed. [H]ard Folding only updates every 3 hours. So, 3 to 4 hours after you turn in your first WU will be when you will show up there and can find your page and get your badge. At the bottom of your user page at [H]ard Folding will be where you can find the java scripting or other coding to place your badge in forum posts.

 
[H]ugh_Freak;1030846120 said:
When you install the client and first connect to Stanford, that is, in essence, your registration with them. No other input from you does Stanford need
Click to expand...

Oh ok, good news. Thanks for the information!! ;)
 
Okay, got it up and running!! :D

Tigerbiten mentioned in an above post that I should be able to turn over 100-150 PPD on my little weakling boxen.
According to EMIII, I have the following statistics after being connected for about 3.5 hours now. (I know the numbers are an estimate based on current performance)

Points per hour: 18.5
Points per day: 444
Points per week: 3,107
Time per work unit: 21h 11m 40s

Is something wrong? Tigerbiten estimated 4x less than what this machine appears it can do. Or are these numbers in the EMIII bogus?

EDIT: After my first frame was processed, I see new stats on my computer as follows:

PPH: 4
PPD: 97
PPW: 682
Time per work unit: 96h 30m 00s

Whoa! What a difference..
 
I just finished my first frame, and EMIII reported an error. It didn't give me ANY details to go on, but after I right-clicked the small error bar, it cleared the error and pushed on. A couple questions:

1) If I hadn't cleared the error, would it have stopped folding all together until the error was cleared?

2) Does anyone know why I may have got an error to begin with?
 
EMIII just reads from the folding logs, and it doesn't affect folding (except for the few cpu cycles it uses). I have never seen an error in EMIII. Keep us posted.

Welcome to the [H]orde.



 
EM will sometimes say there is an error condition when processing is just starting up. It isn't a problem unless it stays that way for more than a couple minutes.
 
Sometimes EMIII thinks your first savepoint is the end of the first frame.
That could well be why it gave you a high PpD for the first frame and an error when the second frame did not finish whan it expected.
Always wait untill you finish the second or third frame before working out your PpD.

If you look in your FAHlog.txt file, you'll find all the frame times and true errors shown there.

Luck ............. :D
 
Well, I decided to chuck the celeron, and I am using my workstation (AM2 3800+) and gaming rig (E6600) until I get my e6300 c2d setup. (I am building the e6300 system specifically for F@H)

I have noticed one weird thing though...

EMIII on the AMD machine doesn't report any frame activity or anything... no stats at all!

Does the beta SMP console and EMIII work together? Maybe that is the problem


Also...

Should I OC the AMD in the meantime to produce better PPD? I mean, it is running @ it's stock 2ghz and is currently @ 29% of a 500,000 task (Started last night 4-1-07 at around 23:45 EST and is now 4-2-07 20:42 EST. Does this seem right?
 
With the OCing just be aware that a 100% stable (prime95 etc) OC may still crash on F@H. I had my 6400 OCed to 3.4 100% stable with no issues at all and then had F@H crap out about 6 WU in a row.. Machine still showed as stable but just wasn't folding stable. I just put it back to stock and let it fly. I was only saving about 4 min a frame so in the long run I am getting more points and science done with a non OC than I was with a monster OC being as I kept getting EUE..Kicking the OC down to 3.0 or 3.2 only gave me a min or 2 advantage and with as often as I am out of town (3-4 days a week) it didn't sem prudent to have it crapping out and trying to explain to my wife how to reset the bios on my PC :)
 
I think I will just leave all stock speeds for now, until I get my 6300. I ran across this F@H monitor program, and I like it. So far.. I think I am setup, both boxes running SMP consoles and folding ftw! Now if I can just get my first work unit :rolleyes:

screenshot_folding.jpg


Again, not too sure how accurate those PPD predictions are.. but that is pretty sad (499 PPD) from my AMD rig, when a PS3 can do 900 PPD.
OMG I can't believe the addictive nature of folding. I can already see that you can never have enough freakin' GHZ.
 
Alright guys, I am up and running.. and no problems with anything up to this point..
I processed my first work unit.. but have a quick question about it, here is my log:

[02:20:50] Completed 1000000 out of 1000000 steps (100 percent)
[02:20:50] Writing final coordinates.
[02:20:51] Past main M.D. loop
[02:20:51] Will end MPI now
[02:21:51]
[02:21:51] Finished Work Unit:
[02:21:51] - Reading up to 804816 from "work/wudata_01.arc": Read 804816
[02:21:51] - Reading up to 393900 from "work/wudata_01.xtc": Read 393900
[02:21:51] goefile size: 0
[02:21:51] logfile size: 627704
[02:21:51] Leaving Run
[02:21:51] - Writing 3090052 bytes of core data to disk...
[02:21:51] ... Done.
[02:21:51] - Failed to delete work/wudata_01.sas
[02:21:51] - Failed to delete work/wudata_01.goe
[02:21:51] Warning: check for stray files
[02:21:51] - Shutting down core
[02:21:51]
[02:21:51] Folding@home Core Shutdown: FINISHED_UNIT
[02:21:51]
[02:21:51] Folding@home Core Shutdown: FINISHED_UNIT
[02:23:59] CoreStatus = 64 (100)
[02:23:59] Sending work to server


[02:23:59] + Attempting to send results
[02:24:40] + Results successfully sent
[02:24:40] Thank you for your contribution to Folding@Home.
[02:24:40] + Starting local stats count at 1
[02:26:45] - Preparing to get new work unit...
[02:26:45] + Attempting to get work packet
[02:26:45] - Connecting to assignment server
[02:26:45] - Successful: assigned to (xxx.xx.xx.xx).
[02:26:45] + News From Folding@Home: Welcome to Folding@Home
[02:26:45] Loaded queue successfully.
[02:26:57] + Closed connections
[02:26:57]
[02:26:57] + Processing work unit
[02:26:57] Core required: FahCore_a1.exe
[02:26:57] Core found.
[02:26:57] Working on Unit 02 [April 4 02:26:57]
[02:26:57] + Working ...
[02:26:57]
[02:26:57] *------------------------------*
[02:26:57] Folding@Home Gromacs SMP Core
[02:26:57] Version 1.74 (March 10, 2007)
[02:26:57]
[02:26:57] Preparing to commence simulation
[02:26:57] - Ensuring status. Please wait.
[02:27:02] - Starting from initial work packet
[02:27:02]
[02:27:02] Project: 2610 (Run 1, Clone 322, Gen 7)
[02:27:02]
[02:27:02] Assembly optimizations on if available.
[02:27:02] Entering M.D.
^^I know that when it says "entering MD" that it is getting some information about the next work unit to process... so the above log should show all the detail from the completion of one wu to the start of a new one.

I just wanted to make sure I got credit for my first work unit, and it appeared that it transfered okay

Lastly, is there a place to go to learn more about the names of the proteins that we get? I know when you pull down a new work unit, they have names.. and have heard names referred to in the past (ie. Supervillians?) what does all this mean?
 
Everything looks good as far as the transfer, same as my SMP machine always looks. Wait a few hours and check stanford or hardfolding and see if you're on there. If not, still don't worry, I've had a few of them take WAY more than three hours to show up for some reason. Never sure why it happens, but it does sometimes.

As for learning about proteins, go to stanford's folding@home page, click on "stats" on the left, then on the right of the next page there's two things: Project summaries and Project descriptions, both can give you a little info. It gives you mostly biological info though, I've never seen one go into "minor" details as they probly see them, such as explaining names, etc.
 
waltherone said:
Everything looks good as far as the transfer, same as my SMP machine always looks. Wait a few hours and check stanford or hardfolding and see if you're on there. If not, still don't worry, I've had a few of them take WAY more than three hours to show up for some reason. Never sure why it happens, but it does sometimes.
Click to expand...

Okay, sounds good! ;)

waltherone said:
As for learning about proteins, go to stanford's folding@home page, click on "stats" on the left, then on the right of the next page there's two things: Project summaries and Project descriptions, both can give you a little info. It gives you mostly biological info though, I've never seen one go into "minor" details as they probly see them, such as explaining names, etc.
Click to expand...

Alright, was just curious about this. I already read about them on the Stanford webpage, didn't give a whole lot of details.. I will try Wikipedia! :D
 
Status
Not open for further replies.
Back
Top