Tinker?

[wainwra]

Hello. I'm new to Folding@Home (and to spcr too, actually; I'm supposed to be planning my next PC, but I kinda got distracted).

I have installed the text-only software on my IBM Thinkpad (R50p, 1.7Ghz Intel Pentium M) and it's busy churning through the first WU (400 frames - will take another couple of days).

Thing is, I can't seem to stop it from using the TINKER method, even though (according to SiSoft Sandra), this machine supports SSE (and SSE2).

I am starting the program using

Code: Select all

"C:\Program Files\Folding@Home\FAH502-Console.exe" -advmethods -verbosity 9 -forceasm

but I still get that TINKER line appearing:

Code: Select all

Launch directory: C:\Program Files\Folding@Home
Executable: C:\Program Files\Folding@Home\FAH502-Console.exe
Arguments: -advmethods -verbosity 9 -forceasm

Warning:
 By using the -forceasm flag, you are overriding
 safeguards in the program. If you did not intend to
 do this, please restart the program without -forceasm.
 If work units are not completing fully (and particularly
 if your machine is overclocked), then please discontinue
 use of the flag.

[07:22:06] - Ask before connecting: No
[07:22:06] - User name: wainwra (Team 31574)
[07:22:06] - User ID: 23390DC173278039
[07:22:06] - Machine ID: 1
[07:22:06]
[07:22:07] Loaded queue successfully.
[07:22:07] + Benchmarking ...
[07:22:09] The benchmark result is 5564
[07:22:09]
[07:22:09] + Processing work unit
[07:22:09] Core required: FahCore_65.exe
[07:22:09] Core found.
[07:22:09] Working on Unit 01 [September 29 07:22:09]
[07:22:09] + Working ...
[07:22:09] - Calling 'FahCore_65.exe -dir work/ -suffix 01 -checkpoint 15 -force
asm -verbose -lifeline 4040 -version 502'

[07:22:09] - Autosending finished units...
[07:22:09] Trying to send all finished work units
[07:22:09] + No unsent completed units remaining.
[07:22:09] - Autosend completed
[07:22:10] Folding@Home Client Core Version 2.53 (June 29, 2004)
[07:22:10]
[07:22:10] Proj: work/wudata_01
[07:22:10] Done: 22928 -> 142981 (decompressed 623.6 percent)
[07:22:10] nsteps: 5000000 dt: 2.000000 dt_dump: 250.000000 temperature: 298.000
000
[07:22:10] xyzfile:
[07:22:10] "   393  p1110_L939_K12M_nat_min1
[07:22:10]      1  N   -105.535073   75.113680  114.225911..."
[07:22:10] keyfile:
[07:22:10] "parameters  ./proj1110.prm
[07:22:10] NOVERSION
[07:22:10] ARCHIVE
[07:22:10]
[07:22:10] cutoff        16.0
[07:22:10] taper         12..."
[07:22:10]
[07:22:10] Hashes matched on file work/wudata_01.dyn
[07:22:10] ARC file integrity verified
[07:22:10] Restarting from checkpointed files.
[07:22:10]
[07:22:10] Protein: p1110_L939_K12M_nat_min1
[07:22:10] - Run: 165 (Clone 1, Gen 6)
[07:22:10] - Frames Completed: 82, Remaining: 318
[07:22:10] - Dynamic steps required: 3975000
[07:22:10]
[07:22:10] Writing local files:
[07:22:10]
[07:22:10] parameters          work/wudata_01.prm
[07:22:10] - Writing "work/wudata_01.key": (overwrite) successful.
[07:22:10] - Writing "work/wudata_01.xyz": (overwrite) successful.
[07:22:10] - Writing "work/wudata_01.prm": (overwrite) successful.
[07:22:10] - Writing "work/wudata_01.key": (append) successful.
[07:22:10]
[07:22:10] PROJECT="work/wudata_01", NSTEPS=3975000, DT=2.0000, DTDUMP=25.000000
, TEMP=298.00
[07:22:11] TINKER: Software Tools for Molecular Design

(and so on). What am I doing wrong?

[wainwra]

Hmmm... looks like I could use some help with my signature, too!

[NeilBlanchard]

Hello & welcome to SPCR Folds!!

I think that the work units are the "luck of the draw" -- if your machine is fast on some units and slow on others, this may be taken into account in the future. But at the beginning, it is sort of first come, first serve.

I have an AMD K6/3 400mHz (as one of six machines) and this one timed out on about 3 or 4 units -- and now it gets units that it can finish "on time".

[wainwra]

Thank you Neil!

I wasn't trying to make a point about having a slow machine, and I didn't mean to imply that FAH wasn't working - it's just fine. I've now finished 130 frames.

I was asking about something I read in this sticky post in this forum:

Tinker - Generally slower and usually fewer points, becasue this older core does not support the SSE instructions found on modern WU's. If you have a relatively modern cpu, you want to try to avoid Tinkers if you can, by using the -advmethods command-line flag, explained elsewhere in this guide. For some older processors (those which do not have SSE), Tinkers are OK.

Gromacs - Generally faster and usually more points than Tinkers, because this newer core is optimized for processors with SSE instructions.

DGromacs - Double-precision Gromacs. Uses SSE2 instructions found on Pentium 4's, and recent generation Athlon processors.

So my question is, shouldn't I expect to be using Gromacs or DGromacs instead of Tinker?

[Lord_Anubis]

I think you need too have big WU's enabled.

[wainwra]

Great! I'm all for big WUs, I think.

er... how?

[Lord_Anubis]

2 ways that i know of

1. edit the client.cfg (in your folding@home folder). Using notepad or anyother text editor and look for bigpackets=no and change the no to yes.

or

2. if you are using the graphical version, right click on the icon in your system tray (by the clock). click configure and click on the connection tab. You will see a box unchecked (Allow receipt of....) and just check the box.


Either way it's preatty easy.

[wainwra]

Ah yes.. Okay, so I edited the client.cfg file, and found that it ALREADY had bigpackets=yes. Here:

Code: Select all

[settings]
username=wainwra
team=31574
asknet=no
bigpackets=yes
machineid=1

[http]
active=no
host=localhost
port=8080
usereg=no
proxy_name=
proxy_passwd=

[graphics]
subtitle=Distributed Computing    -    http://folding.stanford.edu
drawgap=125
drawtitle=1
saver_mode_enabled=0
saver_runs_core=0
saver_idle_minutes=5
drawlogos=1
logourl=
drawmode=1
logopath=C:\Program Files\Folding@Home\EM\Skins\spcr_1\teamlogo.gif

[core]
priority=0
cpuusage=100
disableassembly=no
checkpoint=15
ignoredeadlines=no

[power]
battery=yes

So - can anyone tell me while it's still using TINKER?

[StarfishChris]

That's because you're already working on a unit. When it's finished you will get another one.

[wainwra]

Well, yesterday when I got back to my machine, it had finally finished its first packet. So I checked the following...

1) was I now using Gromacs?
2) what was my score
3) would I finally start appearing on the list of users in the team listing ?

Well... No, zero and no.

Worse. I seem to have restarted the same packet (165,1,6). I wouldn't mind so much if it weren't for the enormous size (400 frames, will finish Tuesday lunchtime).

What is going on?

[NeilBlanchard]

Hello,

1) It sounds like you are still using the Tinker core , from the fact that you are on the same project.
2 & 3) Is your Folding name the same as your screen name here: wainwra? It should show up soon enough...those are big databases, so they may take a day or two. Also, did you get the SPCR team number (31574) right in your client config?

As for the same project -- you just got another "chunk" of the the project. There are possibly thousands of these chunks, and they are probable running on hundreds of computers; and the Folding servers "piece" the results together from all the clients that are running them.

Patience is the best thing -- though I remember the anticipation of waiting for my first work unit to show up...it will show up, and there will be plenty more to keep us all busy for the foreseeable future.

[wainwra]

Update: yes, my username and team number are correct. However, I think I've found the cause of the problem.

My machine had been restarted several times (4 I think) over the first 3 days. Each time FAH was able to continue successfully - apart from the last time.

I've checked the FAHlog.txt file, and found this crucial chunk:

Code: Select all

[13:52:39] Hashes matched on file work/wudata_01.dyn
[13:52:39] Header on frame 1 differs from expected header
[13:52:39] Got:      A50456-60001-433ACF29-36B4BC85:                             0001
[13:52:39] Expected: A50456-60001-433ACF29-36B4BC85:                             0383
[13:52:39] Starting from initial work packet

Ug. Sigh. And Other Expressions.

I think this explains most of the problem (though not why I'm still using Tinker). I'll post again in a couple of days, presuming I get no other issues.

(Oh, can anyone tell me what is wrong with my signature?)

[peteamer]

The answer to your signature questions.

[NeilBlanchard]

Hello,

If you are folding for a particular project -- then you have to use the Tinker core. Other projects use the GROMACS core, and you can't really do anything to change it.

Are you overclocking?

And to get a working sig, you can use this: Sig Helper

[wainwra]

Thanks Pete, Neil!

It seems that project just required Tinker, whatever. I've since loaded other projects (big problems; I'll start a new topic) and they have been using Gromacs.

Serious kudos to dedogs, too for the awesome sig helper!