The ***OFFICIAL*** F@H Harassment and discussion thread!!!

[sc00p]

Ah ha ha, I crack myself up.

Yes you crackhead Can you see mine I got smps now... and I got even older reserves than Sparky's

[GAM]

Well, I still have my old P4s8x P4c 2.53GHz 2x256MB RAM sitting in a box. When I build my new server cube case (on the drawing board and scheduled for construction soon), this can go back into 'Server's' current case.


I will have to look into this SMP thing everyone is doing too. Ah... too many things to do.

[Sparky]

Well, I still have my old P4s8x P4c 2.53GHz 2x256MB RAM sitting in a box. When I build my new server cube case (on the drawing board and scheduled for construction soon), this can go back into 'Server's' current case.


I will have to look into this SMP thing everyone is doing too. Ah... too many things to do.

no......no no no, you don't want the SMP client. It burns the occasional WU.
....also will give you a big boost in PPD, making it harder for me to stay ahead...

[sc00p]

Well, I still have my old P4s8x P4c 2.53GHz 2x256MB RAM sitting in a box.

Ooh that's hightech compared to my athlons I got here getting dusted

When I build my new server cube case (on the drawing board and scheduled for construction soon), this can go back into 'Server's' current case.

So you got cube on the way too I just got last saturday finished my own SteelCube See Case-section tomorrow. I'll try make a thread about it. But no proper work log coming cause I did that case in hurry.

I will have to look into this SMP thing everyone is doing too. Ah... too many things to do.

Project 2608 takes ~16min/frame with my deadstock 6420 => 1385 points coming (2 hours away)

[GAM]

@Sparky, okay, I'm convinced.... TO DO IT ASAP!!!

That means in the next week or so as I have so much on at the moment.

btw, the occassional WU's been getting burned anyway at the moment. A couple of weeks ago I waited over 24hrs for a measly 292pts, it got about 60% or something and EUE'd. I got credited a measly 3pts! Phewy.

@Scoop, well... given that my P4e@3.0 1GB, only puts out about 200ppd, I didn't think the extra power, heat and noise was worth it. But I will definately look into it, especially as the weather is getting colder now

I'll keep an eye out for you case. Mine will be more of a 'fatboy' than a proper cube. Dimensions are similar to a mid-tower but closer to 400mm wide. It also has some funky, probably total waste of time, design changes. For instance, the mobo will be rotated, facing inwards with I/O connectors on top and all fans on the floor pushing up, no exhaust.

[GAM]

EXTRA EXTRA, I've just made 1,858ppd - that's a GAM Record

[Sparky]

@Sparky, okay, I'm convinced.... TO DO IT ASAP!!!

That wasn't the effect I was looking for

EXTRA EXTRA, I've just made 1,858ppd - that's a GAM Record

Yeah and now I have 3 months at my current PPD before you crunch me. Going to have to do something about this... Are you hogging all the 149x projects?!

[sc00p]

@Sparky, okay, I'm convinced.... TO DO IT ASAP!!!

Yes man, let's take him down

@Scoop, well... given that my P4e@3.0 1GB, only puts out about 200ppd, I didn't think the extra power, heat and noise was worth it. But I will definately look into it, especially as the weather is getting colder now

I had to say my old puny rigs probably stay off... last summer was horror, my room temp stayed always around 30c and electricity bill was nice

I'll keep an eye out for you case. Mine will be more of a 'fatboy' than a proper cube. Dimensions are similar to a mid-tower but closer to 400mm wide. It also has some funky, probably total waste of time, design changes. For instance, the mobo will be rotated, facing inwards with I/O connectors on top and all fans on the floor pushing up, no exhaust.

Had to warn: I haven't had too much time to put for better grade of finishing but it's for LANs so do I care

BTW: folding now project 2610 Have no clue ~ credit it gives -> going oced 11min/frame

maybe better to go sleep some... it's 3am here LOL

[Sparky]

Yes man, let's take him down

Sounds like I just may have to pull away from WCG, you guys sure seem bent on taking me over Although, I think you need to buy more systems sc00p, get yourself up to the red Just, don't buy too many and try and pass me

[sc00p]

Sounds like I just may have to pull away from WCG, you guys sure seem bent on taking me over Although, I think you need to buy more systems sc00p, get yourself up to the red Just, don't buy too many and try and pass me

Are you going to have me red My face is definately going to be red colored from the heat if I put all the rigs powered again Not sure about ppd

[Sparky]

Are you going to have me red My face is definately going to be red colored from the heat if I put all the rigs powered again Not sure about ppd

Hey I know all about the heat part, my room must have been a good 28+ degrees C yesterday! It was sweltering in here, all the systems kept running though. My CPU which usually runs about 51-52C was hitting 57C. That poor old athlon XP in fold6 was sweltering at 69C.

For you though, just turn the AC up a little

[sc00p]

Hey I know all about the heat part, my room must have been a good 28+ degrees C yesterday! It was sweltering in here, all the systems kept running though. My CPU which usually runs about 51-52C was hitting 57C. That poor old athlon XP in fold6 was sweltering at 69C.

For you though, just turn the AC up a little

Oh you are so hot
That's the way to crunch

Hey mate... had to go to sleep now... later...

[Sparky]

Well there was no way I was shutting down the crunchers just because of a little heat As long as they didn't overheat and shut down on their own they kept running. I'm a hard taskmaster for my crunchers

[mk5652]

Another question I'm afraid

I abandoned SMP, because of a couple of things, one being the need to have a password on the user account, and decided to have a go with two instances on the normal console client running instead.

It's working fine, but they seem to be running out of sync.

[13:13:29] Folding@Home Gromacs Core
[13:13:29] Version 1.90 (March 8, 2006)
[13:13:29]
[13:13:29] Preparing to commence simulation
[13:13:29] - Assembly optimizations manually forced on.
[13:13:29] - Not checking prior termination.
[13:13:29] - Expanded 292446 -> 1461493 (decompressed 499.7 percent)
[13:13:30]
[13:13:30] Project: 3040 (Run 2, Clone 897, Gen 11)
[13:13:30]
[13:13:31] Assembly optimizations on if available.
[13:13:31] Entering M.D.
[13:13:51] (Starting from checkpoint)
[13:13:51] Protein: p3040_supervillin-03
[13:13:51]
[13:13:51] Writing local files
[13:13:51] Completed 473196 out of 5000000 steps (9)
[13:13:51] Extra SSE boost OK.

/snip

[16:01:18] Completed 700000 out of 5000000 steps (14)

[13:13:29] Folding@Home Gromacs Core
[13:13:29] Version 1.90 (March 8, 2006)
[13:13:29]
[13:13:29] Preparing to commence simulation
[13:13:29] - Assembly optimizations manually forced on.
[13:13:29] - Not checking prior termination.
[13:13:29] - Expanded 290480 -> 1461493 (decompressed 503.1 percent)
[13:13:29]
[13:13:29] Project: 3040 (Run 0, Clone 496, Gen 11)
[13:13:29]
[13:13:31] Entering M.D.
[13:13:51] (Starting from checkpoint)
[13:13:51] Protein: p3040_supervillin-03
[13:13:51]
[13:13:51] Writing local files
[13:13:51] Completed 175843 out of 5000000 steps (4)

/snip

[15:50:31] Completed 250000 out of 5000000 steps (5)

5% vs. 14%, both were started at exactly the same time.

The shortcuts are as follows:

"C:\folding@home 1\FAH504-Console.exe" -local -forceasm -verbosity 9
"C:\folding@home 2\FAH504-Console.exe" -local -forceasm -verbosity 9

Is this normal behaviour, or have I done something wrong?

[Sparky]

That is normal. The normal CPU client each works on its own WU, and some go faster than others. In your case, they both have the same WU, but depending on your computer usage, one core may be getting used for other windows tasks as well as F@H while the other is left mostly to F@H. Nothing abnormal about it

[mk5652]

Okay, thanks. I was wondering mainly because they are both doing the same project; 3040, just different Run/Clone/Gen's. Also the system is mostly running idle at the moment.

I originally thought you'd be able to assign a physical core to each client, but it doesn't seem to work like this. Unless Windows automatically assigns each instance of the client to it's own core.

[sc00p]

I originally thought you'd be able to assign a physical core to each client, but it doesn't seem to work like this. Unless Windows automatically assigns each instance of the client to it's own core.

What Now if you follow guides... link which you can find in Sparky's sig too ... if you install two instances of f@h for a dual core processor you have to manually set physical core =machine ID. See fourth part at Sparky's guide ...not sure why these guides aren't stickied yet
Right now you're probably using HT which is not recommended

[sc00p]

On the other hand, Scoop, my old friend...
Currently 1.1 months till 'Poocs' day, but I'd imagine it will end up more like 2

How is it going You're Oracle ... Mr Spooc is going to the outer space...

[mk5652]

if you install two instances of f@h for a dual core processor you have to manually set physical core =machine ID.

Yeah, I'd already done this, the first instance has a machine ID of 1, the second has a machine ID of 2. Both cores are going at 100%, As I mentioned, as it's the same project, I just thought there would be more of a balance in the progress. They are now at 7% and 19%. Perhaps the different run/clone/gen makes more of a difference then I had anticipated.

The weird thing is, if I close one instance, both cores drop to 50%, so the wokrkload is being shared between them, rather then one core being utilized 100%, and the other "sleeping". This might be supposed to happen with only one instance running, I don't know, which is why I'm confused.

[sc00p]

Yeah, I'd already done this, the first instance has a machine ID of 1, the second has a machine ID of 2. Both cores are going at 100%, As I mentioned, as it's the same project, I just thought there would be more of a balance in the progress. They are now at 7% and 19%. Perhaps the different run/clone/gen makes more of a difference then I had anticipated.

The weird thing is, if I close one instance, both cores drop to 50%, so the wokrkload is being shared between them, rather then one core being utilized 100%, and the other "sleeping". This might be supposed to happen with only one instance running, I don't know, which is why I'm confused.

My bad... I misunderstood, that you haven't done this... so you've.

BTW: each wu is different... even they are from the same project. It's how they make science... simulations of the same "happening" with some minor variations, I think

Now comes to my mind... how many services of f@h(-cores) you see at Task Manager?

[mk5652]

BTW: each wu is different... even they are from the same project. It's how they make science... simulations of the same "happening" with some minor variations, I think

Ah, I see.

Well, I have 2 x FAH504-Console.exe, and 2 x FahCore_78.exe

Which I'm guessing would be right.

I've just checked the CPU time, and one instance of FahCore_78.exe has clocked up 25 minutes, where as the other is at 1 hour 54 minutes. hmm.

Up until a short while ago, their priority was set to idle, I've since chanced it to low as I'm not using the machine for anything else at the moment. Maybe that was it then, one core was busy with other processes. I don't know what could use up that much CPU time though, as the system hasn't been used all afternoon.

[GAM]

@mk5652
Not going to quote everything... too messy. But I agree with Scoop, check your task manager as I have had the same thing happen before, i.e., two services running, yet also a 'console client' running i.e., 3 instances on one machine. This might explain the 'significant' difference between the two.

For reassurrance, one of my cores usually completes faster than the other for reasons specified above, one core gets devoted to more of the 'housework' while the other is left to fold more devotedly. Still, not such a significant difference as you're indicating above.


@Scoop, sorry mate, I don't quite follow you.
But, my amazing PPD is starting to fall for natural reasons, although I did make the mistake of leaving some memory hungry apps open last night which significantly affected the ppd on one core--min 600ppd down to avg 266ppd!?!?

On another note, I just received my replacement HZ sticks (RMA'd a week ago) and can now migrate the new 2x1GB HK's to 'HTPC'. This should give HTPC a very significant boost on PPD as it is only using 1GB at the moment and for some reason I only specified 500MB.

[sc00p]

@Scoop, sorry mate, I don't quite follow you.

Have you checked this Whatta a spike

[GAM]

That is one INCREDIBLE spike!

So what's the story?

[sc00p]

I don't know what could use up that much CPU time though, as the system hasn't been used all afternoon.

Spywares and viruses LOL

BTW: I was afraid that you've leaved some leftovers of the SMP client to your machine. I guess you are fine right now... I'm sorry had to go sleep soon
...if not lets think about this tomorrow or with someone else.