• 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.

Min specs for SMP

S

SeaMonkey

Lurker
Joined
May 3, 2007
Messages
73
Hello Everyone,

Most of the machines folding for me are probably at least 4 years old. I have a few dual 1GHZ p3s sitting around that I'm about to begin folding with so they are at least doing something productive other than heating my server room.

Based on the speed, does anyone have an opinion if the SMP client is worth it, or should I just run two instances of the standard client? I've got 4 machines that fit this description, so any suggestions on the best way to use them would be appreciated.

Thanks
 
SMP client will not be able to finish in time on those..... A dual 2.8 netburst xeon only makes it sometimes so I'm only running standard clients on that. And..... an early congrats on hitting that 1 million point milestone!! Thats a big one. :D

 
I'm running smp clients on 3.4GHz HT P4s (not dual core) and they make deadlines just fine. Sure they only average ~650PPD but it is better than the single client.
 
russnuck said:
I'm running smp clients on 3.4GHz HT P4s (not dual core) and they make deadlines just fine. Sure they only average ~650PPD but it is better than the single client.
Click to expand...

Do you make the preferred deadline? If not your workunit, as I understand it, is reissued, forcing two machines to do the work and thus not benefiting the cause as much. Making pref deadlines really is critical.

 
they make the preferred deadlines with about 8 hours to spare :D

I wouldn't be running it that way if they didn't meet the preferred deadline.
 
Thanks. You pretty much confirmed what I had already expected. I'll just run multiple instances of the standard client. Most of my other machines are just getting work done under the Perf deadline. Don't want to waste cycles on dupe efforts.

Thanks
 
Damn Seamonkey your the the top name on my threat list, 4-7-08 overtake. I have some new hardware that just came online so lets see if i can push that back just a little bit.



 
Sunin said:
Do you make the preferred deadline? If not your workunit, as I understand it, is reissued, forcing two machines to do the work and thus not benefiting the cause as much. Making pref deadlines really is critical.

Click to expand...

For the moment I'm going to retract my comments on preferred deadlines. The FAQ seems unclear but 7im seems very clear.

http://foldingforum.org/viewtopic.php?f=16&t=1654

If 7im is correct I may have to rethink my entire position on folding. Getting the points and not getting credit for the science just to make people feel better is not somthing I am digesting very well. While I'm having no issues with with preferred deadlines I have a serious issue with the duality of a second deadline.

Probably more on this later.........:rolleyes:


 
If that's true about the preferred deadlines then I'm going to be going back to all my older machines and make sure they are getting these things in on time. I might have one or two vmware instances that are not making the preferred deadline cause I was just getting more points by stacking my one big box. The thing has 8 cpu cores (4 dual xenons), so I've been running 4 guests folding. Might be time to pull one of those guests offline.

Since I'm loading these older machines with the standard client, I'm going to have to throw a lot of junk at Connundrum1 to keep up with any new hardware. If they only had a solaris client, I'd be in good shape. I have several old sun v440s and v490s that are pretty much doing nothing right now.

Thanks for the info. I learned something today I didn't know about deadlines. :) Love the points, but my conscience about what's best for the cause is going to win out on this one.
 
SeaMonkey said:
If that's true about the preferred deadlines then I'm going to be going back to all my older machines and make sure they are getting these things in on time. I might have one or two vmware instances that are not making the preferred deadline cause I was just getting more points by stacking my one big box. The thing has 8 cpu cores (4 dual xenons), so I've been running 4 guests folding. Might be time to pull one of those guests offline.

Since I'm loading these older machines with the standard client, I'm going to have to throw a lot of junk at Connundrum1 to keep up with any new hardware. If they only had a solaris client, I'd be in good shape. I have several old sun v440s and v490s that are pretty much doing nothing right now.

Thanks for the info. I learned something today I didn't know about deadlines. :) Love the points, but my conscience about what's best for the cause is going to win out on this one.
Click to expand...

In a nutshell here is my problem. I always turn work in way before the preferred deadlines BUT what about the folk who miss it, accomplish no science and yet get full credit for the work?

That’s a bit like the second place horse in a horse race paying the same amount as the winner.

It throws off any valid team competition. It wastes money and resources. There should be one simple deadline, period. Can’t make it fix your issues or fold something smaller.

I’m just sorry I gave you the wrong information man. I do apologize.




 
Wow, that's extremely stupid. There is absolutely no reason to have two deadlines if the first one is actually the cut off. Having a 'Preferred Deadline' would obviously infer that they want the WU back in that amount of time, but don't need it back by then. If the Preferred Deadline is when they need the WU back then it isn't preferred at all, but is really the Final Deadline.

 
I think I might have missed something here. In BillR's link above there was a question about when a WU was turned in right before the deadline (an hour). If the WU is turned in just prior to the deadline are Stanford's servers going to realize that the WU was turned in on time (even if not by much) and not send out the same WU to someone else. So basically does anyone know how much "lag" time there is...?

 
Stanford says that it can take a couple hours for the stats DB to update, but the WU is registered when the client completes. I always understood it to be that the work assignment DB was separate from the stats DB.

That is why I wouldn't use the SMP client on any machine that wont beat the preferred deadline by atleast 6-8 hours. On the 3.4GHz HT cpus the SMP WU takes just a bit longer than the normal client WU but the science and point value is much greater. 6-8 hours is 10% margin of time error, and it would suck if their system does not allow for something that large, I think we need clarification from Stanford.
 
russnuck said:
Stanford says that it can take a couple hours for the stats DB to update, but the WU is registered when the client completes. I always understood it to be that the work assignment DB was separate from the stats DB.

That is why I wouldn't use the SMP client on any machine that wont beat the preferred deadline by atleast 6-8 hours. On the 3.4GHz HT cpus the SMP WU takes just a bit longer than the normal client WU but the science and point value is much greater. 6-8 hours is 10% margin of time error, and it would suck if their system does not allow for something that large, I think we need clarification from Stanford.
Click to expand...

I have a hard time believing the a 3.4 GHz HT P4 is making the deadlines. My dual 2.8 Xeons with HT won't make it all the time and Pentium Ds usually have to be over 3.0 to make all the deadlines. You're saying they take just a tad longer then a normal workunit? A P4 at 3.4 should complete normal WU's in less then a day.
 
I can post my PPDmon stuff when I get back into the office tomorrow. But last time I checked it was the following.

Right now I have 2 x 3.4GHz HT systems running (I'd have to get the project # later)

Scheduled Completion / Preferred Deadline / Final Deadline

3/28/08 3:45 AM / 3/28/08 10:45 AM / 3/29/08 10:45 AM

Sorry, I was looking at the wrong system for the normal client times, that is my very old 2.4 non HT system.
 
You must log in or register to reply here.
Back
Top