• 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.

Folding with 8 core xeons?

L

lsudvm

n00b
Joined
Oct 15, 2010
Messages
60
I saw some 8 core ES Xeon 1.86 ghz 18mb cache for sale. How are these for folding, say in a quad setup?
 
What versions are these? Xeons usually have UP, 2P, and 4P parts. Part of AMD's value is being able to use the same CPU across all three use cases.
 
Quad 8-core Xeons perform about as well as quad 12-core Opterons of the same clock speed, but the motherboard is so ungodly expensive ($1800) that it's not really worth it for pure folding.
I still have these 8-core 2.13 GHz Extra Special ones chillin' on my shelf...impulse-bought for $50 each and the board is nowhere to be seen.
 
Been looking at them myself, but as other have said the board is stupid money and is amssive to boot. Also IIRC correctly its not as good at folding as opteron, Now the thing is this may have been with older chips - i'm not 100% but i think there is more than one cpu arch available for 1567
 
Nathan_P said:
Been looking at them myself, but as other have said the board is stupid money and is amssive to boot. Also IIRC correctly its not as good at folding as opteron, Now the thing is this may have been with older chips - i'm not 100% but i think there is more than one cpu arch available for 1567
Click to expand...

westmere-ex are the only chips on 1567, it was a dead end test socket for quad channel DDR3 basically. there was the 6 core/12 thread, 8 core/16 thread, 10 core/20 thread chips.
 
sirmonkey1985 said:
westmere-ex are the only chips on 1567, it was a dead end test socket for quad channel DDR3 basically. there was the 6 core/12 thread, 8 core/16 thread, 10 core/20 thread chips.
Click to expand...

Not quite, there were some older 45nm Nehalem cpu's as well that went to 8c/16t, the westmere's go upto 10c/20t
 
So basically we can expect linear performance to our SR2's (at stock obviously)? Maybe a tad faster due to quad chan memory, but it's pretty much the same arch right?
 
Vaulter98c said:
So basically we can expect linear performance to our SR2's (at stock obviously)? Maybe a tad faster due to quad chan memory, but it's pretty much the same arch right?
Click to expand...

Not sure on the performance, there were a couple of threads on here a while ago and the PPD was shite, given the performance of xeon at lower thread counts it was expected that 40c/80t would at least be comparable to 48c Magny cours if not better due to better design in the xeon's arch, but PPD was much worse.

Yes, basically the same arch at lower clocks, more cores and a shit-ton more cache
 
We really do not know. There is not enough data to say that it is shit. We had one sucess story, and one less than spectacular story. Two machines are not enough to determine the performance of the entire platform. The price of the boards prohibits many of us from experimenting with these CPU's.
 
Best info i've found from elsewhere so far is a 32c machine maxing out at 300k PPD, this was a while ago and it doesn't give the OS, whether HT is on, or the WU it was chewing on. I'll do some more googling tomorrow and see if i can get any better infor
 
Here is some info:
http://hardforum.com/showthread.php?t=1633736&highlight=n31l

and here:
http://hardforum.com/showthread.php?t=1606387&highlight=n31l

n31l was running an 8P of 2.26Ghz 8 cores, so 64 cores 128 threads @ 2.26ghz

He ran them in 2 separate clients.

1,324,344 points for 2 x 6904
So that is 662,172 points per 6904

Unkown upload speed ( did not dig for it).

That works out to 00:20:43 TPF average with the upload factored in.
For one client that would be about 00:10:22 TPF or 1.3M PPD on 6904

He did not have access to the internet on the machine so I believe he had to manually upload them by transfering them to his laptop and uploading them later. I do believe the TPF will be much lower, and consequently the PPD higher if it had internet access.

We would really need more info to get the true PPD of these machines, but just by looking at that, these machines are fast.
 
Last edited:
You must log in or register to reply here.
Back
Top