Tinker plague returning?

haysdb · Post by **haysdb** » Thu Jan 29, 2004 11:13 am

Two of my systems just finished WU's and picked up Tinkers. I hope this isn't the beginning of another huge influx.

David

Zyzzyx · Post by **Zyzzyx** » Thu Jan 29, 2004 11:32 am

So what if it is?

Everyone should be getting approximately the same ratio of Tinker/Gromac work units, so it shouldn't matter.

ColdFlame · Post by **ColdFlame** » Thu Jan 29, 2004 4:14 pm

It does matter to me personally. *My* machines are folding at 1/2 speed now because of Tinkers.

There is certainly a point that "Tinkers needs to be done too" but I just like to see my machines folding at 100% of their capabilities.

Also, there is a way (trust me

) to avoid Tinkers if you really want to, and I'm pretty sure people out there do that to get more points. One day I'll figure that out myself

Zyzzyx · Post by **Zyzzyx** » Thu Jan 29, 2004 8:44 pm

Heck... I think I would rather have Tinkers than Gromacs on my systems that don't have (or can't use) SSE. Seems to outproduce by a guesstimated 25%.

haysdb · Post by **haysdb** » Thu Jan 29, 2004 11:12 pm

Zyzzyx wrote:Heck... I think I would rather have Tinkers than Gromacs on my systems that don't have (or can't use) SSE. Seems to outproduce by a guesstimated 25%.

The point has been made in threads at Folding-Community.org that the assignment of work could be more sophisticated, so that work units could be assigned "more appropriately." In the grande scheme of things, more work would be accomplished that way. But if Stanford needs me to work Tinkers, I'll work Tinkers. I still feel entitled to whine about it though.

As it turns out, I just got those two. As long as I have no more than 2 at any one time, I won't piss and moan TOO much.

David

tragus · Post by **tragus** » Sun Feb 01, 2004 4:00 pm

Well, I seem to be getting a bit more than my share of Tinkers. I currently have the highest number of WUs (~8600) on the SPCR team -- nearly 2X the second highest. I do notice that my folding points/day/week tends to go down significantly during my Tinker storms.

Ah well, my bit for science.

dasman · Post by **dasman** » Mon Feb 02, 2004 4:21 pm

9 boxes, 8 tinkers picked up since Feb 1 -- suddenly slow going

Zyzzyx · Post by **Zyzzyx** » Mon Feb 02, 2004 4:56 pm

9 systems.
9 Gromacs.

<shrug>

Still wish I could get a Tinker preference for my non-SSE AMD systems.

ArtCubed · Post by **ArtCubed** » Mon Feb 02, 2004 8:04 pm

Dumb question:
What is a tinker/gromac? How would you identify if you had one? My systems have been folding MUCH slower now, compared to the first 3 WU's I got.

EDIT: It could just be because I got badassly-huge project files the second time around. Both of them are the 55-day projects...at their current rate, it looks like it'll be another day, maybe two. Whatever, more points for me!

dasman · Post by **dasman** » Mon Feb 02, 2004 10:29 pm

What is a tinker/gromac? How would you identify if you had one?

Different proteins use different "core" executables to do the work -- Tinker (older and slower) and gromac (faster for most).

If you look at your log file, it will tell you what you're working on:

[04:48:04] PROJECT="work/wudata_03", NSTEPS=512500, DT=2.0000, DTDUMP=25.000000, TEMP=298.00
[04:48:05] TINKER: Software Tools for Molecular Design
[04:48:05] Version 3.8 October 2000
[04:48:05] Copyright (c) Jay William Ponder 1990-2000

or

[02:04:31] *------------------------------*
[02:04:31] Folding@home Gromacs Core
[02:04:31] Version 1.55 (December 22, 2003)
[02:04:31]
[02:04:31] Preparing to commence simulation

Or, you can download Electron Microscope -- lets you see what's happening a little easier...

Dave

dasman · Post by **dasman** » Mon Feb 02, 2004 10:31 pm

Artcubed,

Just curious, how do you get negative

PPW showing up in your sig???

Dave

haysdb · Post by **haysdb** » Mon Feb 02, 2004 11:34 pm

dasman wrote:Artcubed,

Just curious, how do you get negative PPW showing up in your sig???

Dave

Another LiquidNinja's glitch?

David

dasman · Post by **dasman** » Tue Feb 03, 2004 6:12 am

I guess the ninja's fixed Artcubed's sig, he's on the + side now

ArtCubed · Post by **ArtCubed** » Tue Feb 03, 2004 9:06 pm

dasman wrote:Artcubed,

Just curious, how do you get negative PPW showing up in your sig???

Dave

I only had 2 WU, and I submitted them way before I put the sig thing on...it went back to positive when I submitted my 3rd WU yesterday.

BTW, the 1 PC I have connected is getting Tinkers, and my Macs are getting Gromacs...

haysdb · Post by **haysdb** » Tue Feb 03, 2004 9:43 pm

ArtCubed,

Make sure you have -advmethods specified when F@H starts up.

If you do, then take a look at something for me. In the F@H log for your PC, look for a line like the following:

Code: Select all

[09:42:49] Unit 7 finished with 98 percent of time to deadline remaining.
[09:42:49] Updated performance fraction: 0.982543

What is the "performance fraction?" I read at folding-community.org where this number is now being used by the Assignment Servers and if the number is below .8, will assign Tinkers.

David

ArtCubed · Post by **ArtCubed** » Tue Feb 03, 2004 10:11 pm

I just looked...didn't see anything like that.

But I just passed it the argument now, when it's halfway through a WU...I'll tell you when it finishes if it comes up.

haysdb · Post by **haysdb** » Tue Feb 03, 2004 11:22 pm

You might also have to specify '-verbosity 9' since this might be something that doesn't display (isn't logged) by default.

David

ArtCubed · Post by **ArtCubed** » Wed Feb 04, 2004 10:13 am

The only new thing that came up was "Benchmark result is 5348"
My PowerMac G4 dual 1.25 came up with "Benchmark result is 1321" BUT has finished 2 WU's, yet the P4 has yet to finish a single one....what the heck is that all about? Is the tinker core really 10 times slower??

wgragg · Post by **wgragg** » Wed Feb 04, 2004 12:04 pm

I'm not sure about 10 times, but when some of my PIII's got Tinkers, what used to take one day or less took over 4. My point production for the week really dropped. It would be nice if Stanford could assign the WU's based on the capability of the processors rather than the somewhat random way it is done now. By the same token, they all need to be done, so it works out in the end.

haysdb · Post by **haysdb** » Wed Feb 04, 2004 1:23 pm

That Benchmark score isn't consistant across OS's. My Linux machines report numbers in the 8 and 9 hundreds, whereas equivalent WinXP machines report numbers in the 6 thousands.

David

ArtCubed · Post by **ArtCubed** » Wed Feb 04, 2004 1:33 pm

Just booted up the Athlon 1800+ - and it picked up a Tinker...ugh. If it weren't for my G4, I'd be doing 2 WU a week at the current rate...

monkiman · Post by **monkiman** » Thu Feb 05, 2004 11:25 am

Projects 693-697

My goodness these are some vicious tinkers. From Stanford "in an attempt to fold the largest, most complex, slowest folding target to date."

No doubt, I've only completed 140 of the 400 frames in 2 days (work computer that is not on 24/7)

ArtCubed · Post by **ArtCubed** » Thu Feb 05, 2004 2:15 pm

Yup...finally finished the 693 today...got a Gromac in it's place thankfully.

dasman · Post by **dasman** » Thu Feb 05, 2004 4:11 pm

(7) 694's so far this week, luckily I have all gromacs right now

Dave

monkiman · Post by **monkiman** » Fri Feb 06, 2004 10:13 am

greeaaat..... now both my machines are stuck on 694's

haysdb · Post by **haysdb** » Fri Feb 06, 2004 8:28 pm

monkiman wrote:Projects 693-697

My goodness these are some vicious tinkers. From Stanford "in an attempt to fold the largest, most complex, slowest folding target to date."

No doubt, I've only completed 140 of the 400 frames in 2 days (work computer that is not on 24/7)

No kidding!

p697 - 56 hours on a 2000+ Palomino

p693 - 42 hours on a 2600+ T'bred

David

ColdFlame · Post by **ColdFlame** » Fri Feb 06, 2004 8:58 pm

When I see a Tinker I shoot it with del /s work

haysdb · Post by **haysdb** » Fri Feb 06, 2004 10:21 pm

ColdFlame wrote:When I see a Tinker I shoot it with del /s work

Yeah, that's one way to do it. I won't say I have never done it, because I have, but most of the time I just "grin and bare it."

David

ColdFlame · Post by **ColdFlame** » Sat Feb 07, 2004 12:44 am

Actually my success ratio with these del commands was about 50% because if you attempt to get another WU right after you downloaded it from the server, you're gonna get it again

Also lately I've been so buy, I've no idea what my PCs are folding

haysdb · Post by **haysdb** » Sat Feb 07, 2004 4:25 pm

I am coming into the grips of the Tinker plague. 4 of 10 cpu's are now working Tinkers. As expected, my PPW according to LogStats has dropped like a stone, from the mid 5000's to the low 4000's.

What's interesting is that I don't remember the last time one of my P4/WinXP machines got a Tinker. For whatever reason, only my Athlon/Linus machines get Tinkers. Could just be "the luck of the draw," but I really don't think so. With 4 P4 clients, surely ONE of them would have gotten a Tinker in the last couple of weeks.

David