• 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.

SMP Deadlines?

FLECOM

Modder(ator) & [H]ardest Folder Evar
Staff member
2FA
Joined
Jun 27, 2001
Messages
15,828
I am sure this has been asked but I searched and couldent find it... about how long are the SMP deadlines? I installed it on my main machine (Dual Opteron 1.6Ghz)

its taking about 2.5~ days to finish a WU... about 30 mins per % on P2610 is this ok?
 
Yanked from the beta thread a few pages back (Thanks Killer):

Current projects are:
P2610 - 1523 points, deadline 4 days
P2651 - 1760 points, deadline 4 days
P2652 - 1148 points, deadline 3 days

Looks like your fine.
 
FYI FLECOM on your production boxes depending on load deadlines could be close also. Generally my P4D @ 3.0 finishes with ~ 50% of the time remaining

I'm actually really surprised at those frame times on a dual opty @ 1.6, I would have thought higher.

Bear in mind those deadlines listed above are final deadlines and not preferred. If your not hitting the preferred deadlines it's really all for not and the only thing anyone is getting is the points at that time. If preferred is missed, the WU is assigned to another donor and 2x donors end up crunching the same WU.

Netburst Xeons sub 2.0 GHz have troubles hitting the deadlines at times from reports.
 
well no i dont have it on any of my production machines, i loaded it on some machines at home that dont do much anyway, my workstation, webserver, and exchange server...

i cant put it on any work machines until its out of beta
 
Netburst Xeons sub 2.0 GHz have troubles hitting the deadlines at times from reports.

I have troubles with deadlines on netburst dual Xeons at 2.8GHz. Not every work unit, but enough to make me switch back to normal clients.
 
I have troubles with deadlines on netburst dual Xeons at 2.8GHz. Not every work unit, but enough to make me switch back to normal clients.

Hell yeah, I wouldn't run it on a netburst Xeon unless it's a Nocona or up. Anything less will end up not making the deadlines on some of the WUs.
 
how could a 1.6ghz opteron get them done and a 2.8 xeon cant? :confused:
 
how could a 1.6ghz opteron get them done and a 2.8 xeon cant? :confused:

Because the opteron is much much more efficient per clock cycle. Netburst was intel's attempt to max out the cpu core speeds and pipeline size. They went back to their P3 designs when it became obvious that netburst was tapped out. A 1.6GHz opty must be getting close to the limit, though.
 
how could a 1.6ghz opteron get them done and a 2.8 xeon cant? :confused:

Really what might happen is a netburst rig w/ HT on looks like a 4-way box. There are 4-way SMP WU's and 2-way SMP WU's.

Running a 4-way WU on a 2 way box (with 2 HT "cores") ends up taking most of the allotted deadlines for this 4-way WU since...well.... the deadline is based on a 4-way box not a 2 way.

The deadlines are a little over the top in SMP land, in my opinion.

I know you don't have this on any production boxes, just an FYI to look into down the road. Suck to reconfig a few hundred boxes and 1. not get any points or 2. end up hurting the project more overall than helping (2x donors doing the work).

AMD is more efficent but a 2.8 Xeon not making the dealines is dumb as hell...in my opinion.
 
I think it's suffice to say, the beta SMP client needs to learn the difference between HT and another core. Who knows if it'll get fixed...
 
I think it will get fixed. It appears they are moving towards SMP as the primary (CPU) client. For OSX Intel, SMP is even the only client. Stanford has stated it will have a configurable number of threads at some point in the future. Hopefully soon...

As an aside, do you think they would do any better with 2 threads? I kind of doubt it, they need to increase the deadline a bit, 24 hours more would probably be enough. Is there a reason you can't disable HT? Due to the large number of boxes and the manpower needed to change bios settings?
 
Really what might happen is a netburst rig w/ HT on looks like a 4-way box. There are 4-way SMP WU's and 2-way SMP WU's.

Running a 4-way WU on a 2 way box (with 2 HT "cores") ends up taking most of the allotted deadlines for this 4-way WU since...well.... the deadline is based on a 4-way box not a 2 way.

The deadlines are a little over the top in SMP land, in my opinion.

I know you don't have this on any production boxes, just an FYI to look into down the road. Suck to reconfig a few hundred boxes and 1. not get any points or 2. end up hurting the project more overall than helping (2x donors doing the work).

AMD is more efficent but a 2.8 Xeon not making the dealines is dumb as hell...in my opinion.

well my borgs are dual core 3ghz p4's if that cant meet the deadline then the people at stanford are smoking some good stuff when they set those deadlines
 
My 2.8 PentiumD did the first WU with 50% of time to spare. The next WU it downloaded was taking 61 minutes a frame. Then the power went out and it restarted the WU from scratch (that was at 97%)..Needless to say that machine went right on back to the standard client.. Have some 935s (3.2Ghz) boxes coming in this week I will retry with but the SMP client is still WAY to buggy for production
 
I never realyl see any issues with the WIn SMP outside of install issues. Well, I tend not to pay much attention to these things also whihc could explain a few things... :)
 
I had an error on my last WU that said it hadn't written a checkpoint in 3 hours, never seen that one before. But other than that its ran ok for me. I agree the deadlines might be a little steep but I think its like that so that they only get the fastest machines working on these. If they dropped them too much people might try to run higher clocked P4's and Athlons with it just for the points, kinda defeating the purpose of it. Course, if they get it to where you can specify the number of threads, or have it autodetect, then they could force some boxen, ie single core, to only run non SMP WU's. Then they could lengthen the deadline for some of the slower SMP boxen. my 2 cents.

 
i loaded it on some machines at home that dont do much anyway, my workstation, webserver, and exchange server...

i cant put it on any work machines until its out of beta
:eek:
I was an Exchange Administrator for years...for a 20,000 account user base. The concept of an Exchange Server as a "box that doesn't do much" is not one that I ever experienced...LOL
 
:eek:
I was an Exchange Administrator for years...for a 20,000 account user base. The concept of an Exchange Server as a "box that doesn't do much" is not one that I ever experienced...LOL

LOL

well, its a box with 3 accounts , mine, myselfs, and i's ;) :D
 
LOL

well, its a box with 3 accounts , mine, myselfs, and i's ;) :D

Woah.... who gets the most mail out of the three accounts? :)

202276
 
all of them lately, they get hammered with spam....

but mostly me... :D
 
Back
Top