E6300, two clients, different speeds?

Soldato
Joined
9 Jan 2003
Posts
21,087
Location
Cornwall
ok, I've got my mums rig running (E6300, 1gig ram, DS3 etc etc etc)
loaded up two FAH cores as per the sticcy, I've landed myself two 44 pointers (typical) but ones going to take (from now) 7 hours while the other is going to take 11 hours
this is what FAHMon says atleast
is this normal?
 
With regards to folding, I don't know.

But, I did see a little experiment done a while ago comparing superpi times for each core of a certain processor (I forget the details). Over a pair of 32M runs started at the same time, one core beat the other quite considerably. Though 7:10 hours is quite a difference.

Edit: also, get yourself on DR more mate, would be nice to see you in chat some time :)
 
joeyjojo said:
With regards to folding, I don't know.

But, I did see a little experiment done a while ago comparing superpi times for each core of a certain processor (I forget the details). Over a pair of 32M runs started at the same time, one core beat the other quite considerably. Though 7:10 hours is quite a difference.

Edit: also, get yourself on DR more mate, would be nice to see you in chat some time :)
I try to but never seem to have more than a few mins at any one time!
 
Even within the same project their are differences within each specific module (just check your benchmarks to see this) but 3h is a big differance on a small pointer WU.
More likely your OS/services is useing more resourses from on core. Any additional programmes/services running (clearing out your system tray is a good start :D )
Also if your running norton, it's a tad cpu cycle hungry on one core.
 
What about their Run, Clone and Gen numbers? These may affect how long a WU takes, even if they are the same project.

Also, FaHMon bases it's ETA calculations on the time it took the last few frames to complete, now if one core was slowed down for a little while with something intensive like a game or whatever it could take a little while for FaHMon to catch up and have the ETA level out.

SiriusB
 
It's probably just system services being more biased on one core than the other. Or a program which is a bit more cpu intensive is cycling on one core and the other is more "free".
 
no norton, Nod all the way!

Project : 773
Core : Gromacs
Frames : 100
Credit : 44


-- Folding 1 --

Min. Time / Frame : 4mn 48s - 132.00 ppd
Avg. Time / Frame : 4mn 48s - 132.00 ppd


-- Folding 2 --

Min. Time / Frame : 3mn 13s - 196.97 ppd
Avg. Time / Frame : 3mn 58s - 159.73 ppd



just got another 773 so I'll know if it is the same in 2h
 
oceaness said:
Perhaps SSE hasn't kicked in on one of them?

Worth a check.
it looks like system likes one core more than the other, I noticed that while FAH was not on that one core gets more work than the other.
 
Interesting.

My g/f's X2 balances out the work exactly to the nearest decimal place.

I'll find out how this new Allendale balances it out when it's finished.

could have course be that the WU's are quite different.

What are the temperatures like for each core when loaded?
 
Back
Top Bottom