• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

What if .....

E

EvilAlchemist

2[H]4U
Joined
Jan 11, 2008
Messages
2,732
What if all the [H] members stopped using SMP and went back to the standard client ..

I almost bet we could wipe out the server of work ....

Since I hate these 2665 so much .. been thinking about it

 
EvilAlchemist said:
What if all the [H] members stopped using SMP and went back to the standard client ..

I almost bet we could wipe out the server of work ....

Since I hate these 2665 so much .. been thinking about it

Click to expand...

A tip (disclaimer : Not liked by Stanford at all) is to disable advmethods on the smp clients and you will surely get mostly 2605/2653, which is my steady diet.

 
Xilikon said:
A tip (disclaimer : Not liked by Stanford at all) is to disable advmethods on the smp clients and you will surely get mostly 2605/2653, which is my steady diet.
Click to expand...

All mine are set to advmethods = no bigWu=yes ..

All I get is 2665 ... I know linux clients are still getting 2605 .. but us win users are out of luck
 
EvilAlchemist said:
All mine are set to advmethods = no bigWu=yes ..

All I get is 2665 ... I know linux clients are still getting 2605 .. but us win users are out of luck
Click to expand...

Ouch... Didn't experiment with setting advmethods to no on work boxes (all winsmp). Even with that, I did get 2-3 2665 once in a while on my linux boxens (probably out of 2605/2653). Guess the 2653 is depleted now and 2605 cannot be crunched by windows boxes.

 
EvilAlchemist said:
All mine are set to advmethods = no bigWu=yes ..

All I get is 2665 ... I know linux clients are still getting 2605 .. but us win users are out of luck
Click to expand...

I haven't used the -advmethods flag in like forever plus it's turned off in the client. Makes no difference in windows at least.

Might be more interesting to delcare next week the [H]Green week (green for saving some money, screw the global crap) and turn everything off for a week....maybe even two:rolleyes:;)

 
EvilAlchemist said:
All mine are set to advmethods = no bigWu=yes ..

All I get is 2665 ... I know linux clients are still getting 2605 .. but us win users are out of luck
Click to expand...

I don't know if it will help with the 2665's but bumping down the reported memory stopped getting the 2619's.


216
 
I tried all the flag combinations in Windows and it didn't make any noticeable difference what WU I received. It seems that Stanford servers ignore the configuration settings in Windows clients and randomly dispatch WUs to users. The last 24 hrs I have been receiving 30xx WUs with one 2653 and hardly any 2665s (thankfully). The last WU I received a few minutes ago I don't recall working on before (not recently), a 3051 worth 1440pts.

 
I'm running the Linux client inside my VM's without the -advmethods flag and getting a nice steady diet of p2605's.

This slight drop in PpD per protien is normal.
You get a good run of quick protiens which have a nice PpD bonun.
Then you get hit by a run of chewwy ones.
Its why I used play around with the -advmethods flag and packet size while running the standard client.
The standard client range was 45-100 PpDpGhz or +100% before Stanford altered the bonus, which is bigger then the SMP clients range of 200-335 PpDpGhz or around +66%.

Stanford benches at 200 PpDpGhz with the SMP client so as long as I fold at that speed I'm happy.
But I'm even happier to fold at 330 PpDpGhz .............. :p

Luck ................. :D
 
BillR said:
I haven't used the -advmethods flag in like forever plus it's turned off in the client. Makes no difference in windows at least.

Might be more interesting to delcare next week the [H]Green week (green for saving some money, screw the global crap) and turn everything off for a week....maybe even two:rolleyes:;)

Click to expand...

Same here...I stopped using it a while back and continue to get the 2665's on most all my SMP windows boxen.

As far as being green...regretfully I shut down practically all my boxes for a few weeks yeaterday. The heat from my downstairs bunker is making it real difficult for the kids, especially since all the heat is rising straight into one of their bedrooms, and the heat wave outside hasn't helped either. I know it'll be a letdown for science, but science (or me) ain't going to get any help at all if the family isn't happy. :eek:

 
We're all folding for science. I think the issue here, as has been said in the 2665 thread, is that the currently released SMP software has issues with efficiency scaling past 2 cores and consistent point structure in terms of the amount of points given vs the amount of time that a SMP client spends on a WU. The points recently have been inconsistent.

I don't know about everyone else, but I find it convenient to check the health of my folding clients via fahmon and look at the ppd values. I can usually tell if there is trouble because the ppd is lower than expected. PPD is a convenient measure of cpu/client productivity for me.

0.png
 
HighYield said:
We're all folding for science. I think the issue here, as has been said in the 2665 thread, is that the currently released SMP software has issues with efficiency scaling past 2 cores and consistent point structure in terms of the amount of points given vs the amount of time that a SMP client spends on a WU. The points recently have been inconsistent.
Click to expand...

This is exactly what I am saying.
I am not mad at F@H, just wanting to run the best client to help the project.
Since the SMP is not scaling very well, it might be better for some to switch over.

After I got to thinking about it, I realized that everyone with quad cores, would likely run 4 clients ... and that times the number of ppl would be a lot of WU's going out ..

Just on my setup alone ... 9 boxen = 16 SMP WU's at a given time
If that was standard client = 9 boxen = 32 WU's at a given time.

Now, take that number and multiply how many folders we have .. tons of WU's

 
Hey! And that would give me a chance to catch up to you all, seeing as how 99% of my production comes from the console client:p

Nah... joking aside, I see what your saying here, but I think the SMP will likely settle down before too long.....bear in mind I try to be an optimist, so who knows... but we can hope.

 
well i thought the point was quad cores. i have dual cores, but i didn't know how many were putting quads into folding service.
 
APOLLO said:
On this team definitely, nine for me if you include my dual dual-core.

Click to expand...

ditto 9 quads for me and soon as A2 Core hits possibly 2 more... I've already begun purchasing bits! :)

 
So Sunin, if you don't mind me asking, are you married? Do you keep all of these boxen at home? I'm just curious because when I was still married about 6 years ago my wife hated my folding. I had built a cabinet and put bare boards in it to fold. I'm not saying that is why i got divorced, but she hated it. My current girlfriend that I am living with doesn't mind me putting it on our systems, but doesn't want extra boxes cluttering up the place. So I ask just to see if this is just coincidence between my two women or if it is something that is a problem amongst a lot of folders.
 
My wife actually encourages it. I have 7 boxen in my living room (four fold 24/7), and piles of parts scattered all over the place:)

It's my hobby and something I'm very passionate about, and she understands that it's important to me.......of course, my supporting her equally (if not more so) hobby of photography doesn't hurt anything soooo......;)

I think the big thing is making sure that your significant other really understands the project...and I know it can be hard to really explain it to other people sometimes, but I think that's the key.

 
Crazy Larry said:
Pointing you to the cheap IP-35E's didn't hurt ;):D
Click to expand...

Yeah $50 buck Abit IP-35's are [H]ard to pass up! I keep my eye out for deals.. trying not to do to much purchase wise as I have a vaca coming up!

 
OhSoCheesy said:
So Sunin, if you don't mind me asking, are you married? Do you keep all of these boxen at home? I'm just curious because when I was still married about 6 years ago my wife hated my folding. I had built a cabinet and put bare boards in it to fold. I'm not saying that is why i got divorced, but she hated it. My current girlfriend that I am living with doesn't mind me putting it on our systems, but doesn't want extra boxes cluttering up the place. So I ask just to see if this is just coincidence between my two women or if it is something that is a problem amongst a lot of folders.
Click to expand...

Single, but I have a live in GF that is totally fine with it.. in fact while I am away for work she actually monitors the boxes... Admittedly that is not a ton of work anymore... but still nice to know if something does go wrong she can tend to them. She actually helped build two of the boxes... kinda used them to teach her how to build a computer.

Sorry but its not taking up that much room I have it in a walk in closet and it takes up about half the closet... its a bit snug but overall its liveable. And truthfully someone that is that unsupportive of such a great cause makes me question the person... Now I do understand clutter... but if you do it right its not horribly intrusive!

 
EvilAlchemist said:
Here is a thread from the foldingforum
http://foldingforum.org/viewtopic.php?f=4&t=2343#p24464

It kinda backs up what I am saying. Granted it is a crap-shoot on the WU, but PPD can be on par or sometimes better the SMP with standard clients ... all depends
Click to expand...
I already moved two of my systems to the standard client, which were previously working on SMP WUs. Those systems are sometimes producing better PPD than before but as you say it's a crap shoot. However, with the recent spate of 2665s and near disappearance of the 2653, the standard client is resulting in superior PPD. Furthermore, these machines are are now always meeting the preferred deadlines whereas before that was also a crap shoot...

And, for those that observe such minutia, my WUPW has also increased substantially because these units are completed so quickly. So far with these test systems, it's a favorable situation but I don't know if that can be said across all platforms/architectures... YMMV

 
I decided to test on my main boxen to see what kinds of PPD would be produced.

Q6600 - 3.4Ghz - 2GB - XP

SMP Numbers
SMP 1 = 1694.46 PPD
SMP 2 = 1691.01 PPD
TOTAL = 3385.47 PPD for Boxen

Standard Client Numbers
Client 1 = 750.23 PPD
Client 2 = 838.17 PPD
Client 3 = 819.12 PPD
Client 4 = 792.13 PPD
TOTAL = 3199.65 PPD for Boxen

So SMP has a slight advantage over standard Client with the first WU's I received.
I will let this test run for 7 days and post more numbers.

 
EvilAlchemist said:
I decided to test on my main boxen to see what kinds of PPD would be produced.

Q6600 - 3.4Ghz - 2GB - XP

SMP Numbers
SMP 1 = 1694.46 PPD
SMP 2 = 1691.01 PPD
TOTAL = 3385.47 PPD for Boxen

Standard Client Numbers
Client 1 = 750.23 PPD
Client 2 = 838.17 PPD
Client 3 = 819.12 PPD
Client 4 = 792.13 PPD
TOTAL = 3199.65 PPD for Boxen

So SMP has a slight advantage over standard Client with the first WU's I received.
I will let this test run for 7 days and post more numbers.

Click to expand...

That is much much closer than I thought it be. :eek:

 
EvilAlchemist said:
So SMP has a slight advantage over standard Client with the first WU's I received.
I will let this test run for 7 days and post more numbers.
Click to expand...

Should be interesting results. Staying tuned....

Maybe we should start a pool (poll).... :)
I'm in for classic client will catch and pass SMP.

0.png
 
Now Granted, I don't have many EUE's on SMP ...

But if you work on a SMP WU for 20 hours and EUE all is lost.
If you have 4 clients running and get an EUE only 1/4 of production is lost.

And all of this is just preliminary data so far.

It will be very interesting to see what happen later in the test.

Small Update:
Since I am doing this on my main system - look what happened when i walked away for a little while.

Standard Client Numbers
Client 1 = 755.79 PPD
Client 2 = 901.03 PPD
Client 3 = 872.97 PPD
Client 4 = 805.29 PPD
TOTAL = 3335.08 PPD for Boxen

I am going to also do this test on one of my dedicated boxen, so we can get some additional figures.

 
Very interesting Evil, please keep us posted. :D

In all honesty I didn't think the "straight" (non SMP) units would come as close as they did to the SMP'ers. Maybe nomad8u is "right on the money" and the SMP'ers will be caught (less than 200 pernts?) :eek:

 
EvilAlchemist said:
Now Granted, I don't have many EUE's on SMP ...

But if you work on a SMP WU for 20 hours and EUE all is lost.
If you have 4 clients running and get an EUE only 1/4 of production is lost.
Click to expand...
This is one of the biggest advantages running the console client. Since the number of installed clients is double or more, the amount of potential wasted time due to an EUE is greatly reduced. Moreover, the console client tends to be much more stable than the SMP client, possibly allowing for higher OC. Mind you, I wouldn't try increasing frequency in the middle of a heat wave that seems to be gripping many parts of the continent.

I also would like to point out following more than one week of personal observation, that on some systems running additional clients above the total number of cores could yield a small boost in PPD. This does not work with every system, and the boost is nowhere near running a second SMP client over one, but it's something worth to try for further tinkering.

Lastly, the type of bonus WU greatly affects the overall PPD of a system. Some WUs give a much better PPD than others even if they're all bonus Ambers. Generally, the lower the point value, the higher the resultant PPD probably due to much faster processing. Like I stated earlier, the results are sometimes better, sometimes worse than SMP, but even during situations where the PPD is better, it's only marginally better.

Keep in mind that Stanford stated the bonus Amber units will run till the end of August, or thereabouts. The favorable situation with the console client will not last very long and most users that have switched will probably opt to run the SMP client again despite some really nasty WUs.
 
Before I went to work today - base numbers from Dedicated Test Rig

Dedicated Test Rig - Taken @ 3%
Standard Client Numbers
Client 1 = 772.28 PPD
Client 2 = 821.79 PPD
Client 3 = 824.47 PPD
Client 4 = 813.83 PPD
TOTAL = 3232.37 PPD for Boxen

Will post PPD from boxen again tomorrow morning!


 
I'm thinking of making the switch on my main boxen, because the 2X SMP on windows is quirky and it would be nice to get a WU in every update from my farm... even if it is a small one ;)

 
Kendrak said:
I'm thinking of making the switch on my main boxen, because the 2X SMP on windows is quirky and it would be nice to get a WU in every update from my farm... even if it is a small one ;)

Click to expand...

Yea, when exiting one SMP it tends to crash the other. How about if you are running 4 standard console clients on a quad. If you close the clients down, do you risk losing your work if you don't back up first like for dual SMPs on a quad?

0.png
 
HighYield said:
Yea, when exiting one SMP it tends to crash the other. How about if you are running 4 standard console clients on a quad. If you close the clients down, do you risk losing your work if you don't back up first like for dual SMPs on a quad?
Click to expand...
No, multiple console clients don't run the risk of crashing the others when you shut one down. BTW, there are ways to avoid that from happening with the SMP client.
 
APOLLO said:
No, multiple console clients don't run the risk of crashing the others when you shut one down. BTW, there are ways to avoid that from happening with the SMP client.
Click to expand...

yeh, mash the power key :p

 
So Sunin, how do you keep the temps down if you keep it in a closet? Doesn't it need better airflow than that? Her only problem is that she wants things "neat and clean". She doesn't mind what I do and keeps an eye on the laptops and tells me if there is a heat or instability problem. She just wants the house clean. I guess I could buy cases for each box, but IMO it takes up too much space. I used to have 6 boards in a cabinet that I left open and took half the back off so it could breath. I was considering it again until we are moving into my house in a few months. The only place for the cabinet is in my bedroom, so I would need to be able to close it. I didn't want to have to turn it off and restart it every night. So what kinds of temps are you getting in your closet?
 
You must log in or register to reply here.
Back
Top