• 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.

OpenSSI or openMosix

openMosix sounds cooler than OpenSSI?

I've got nothing here.
 
Tormond said:
QFT

I also am in the dark

Click to expand...
QFTx2

what are these things? FAH add-ons? Operating Systems? might try PMing unhappy_mage to post in here (so we can rival the Vid. Card. Forum) since he is the resident expert on all things non-Windows...
 
They are linux add ons that allow you to share cpu power between computers. Sadly they wont help F@H. I was looking at the Open Mosix desciption on this site Open Mosix How to.

This whole Mosix thing is pretty cool but I don't know enough about linux to change over 2+ of my computers to it.

-MN Scout
 
I've heard more about OpenMosix than OpenSSI; I guess it's either more popular or more promoted. Both of these are good for Distributed Computing in general, but not for FAH or UD. They make a bunch of computers act like one large N-processor machine. FAH only allows 8 instances per machine, so even if you had a 20-processor cluster, you could only run 8 copies of FAH.

 
It be interesting to see what inpact the transfer over the network had on folding at home. Say you had the node that you are working on and then you boot up 1 F@H record all the times and what not, and then boot up the second while the first is running and see how big a hit it has on the first, and then shutdown the first and see what the second runs at. That way you could get a bunch of data. I'm actually sort of interested in this Mosix thing maybe it'll push me to learn some linux stuff.

-MN Scout
 
The second FAH can run on a cluster, mage is gonna be busy helping me with CusterKnoppix or what ever.

I call first dibs on mage!

Now that's a sad thought....
 
Of course you can run the second F@H instance on the cluster, but I was wondering if there'd be a hit to the speed of them because something would be going over the network. So I guess the question would it be better to have 2 individual computers folding by themselves, or have an open mosix cluster where everything run off the one node and then gets distributed.

-Mn Scout
 
With any cluster of any type, to my knowledge, you will take a preformance hit.

Efficiency will not be 100%.

That being said, I wonder how quickly a 40 gzh cluster can do a QMD?
 
If each machine is 1 ghz with gigabit connections(to remove network latency) then a 40ghz cluster would fold at the same speed as 40 normal 1 ghz systems. All because F@H isn't multithreaded.

-MN Scout
ps. I don't have any experience about this, but everything that I have read is saying what I just said. Now if you are doing povray or ogg vorbis, or Seti then you can get a nice big performance increase.
 
well, the FAH website says that the network of Home PCs (i.e. us) is faster than any supercomputer precisely because of the hits taken in networking...

it's in the FAQ...
 
Now if they got around to making the cores multithreaded than all would be good. Or if they got the GPU version out the door. They hinted that that was almost done a few months back.

-MN Scout
 
MN Scout said:
Now if they got around to making the cores multithreaded than all would be good. Or if they got the GPU version out the door. They hinted that that was almost done a few months back.

-MN Scout
Click to expand...
the GPU version would finally give me a reason for having a Radeon 9800PRO (and an excuse to OC it :p)
 
a) Folding a protein over a cluster is going to have HUGE impacts on performance. The whole idea of a cluster is to have a FAST interconnect - gigabit barely qualifies. Trying to get anything done on a 100 mbit connection would leave the processors sitting idle most of the time. MPI (message passing interfaces) are predicated on the idea that you can send a message between nodes for practically free. Since this isn't the case with a slow network like ethernet, it leaves the nodes sitting around with nothing to do.

b) The GPUgromacs project is focused on nVidia due to their different implementation of floating point numbers (as I understand it). The 9800 won't help.

 
unhappy_mage said:
b) The GPUgromacs project is focused on nVidia due to their different implementation of floating point numbers (as I understand it). The 9800 won't help.

Click to expand...
*EGADS* no! what you speak of must be false!

all my dreams..the extra ppD...no!!!!!
 
Oh no, I thought the 9800 were the only good ok ones on the ATI side that would work. Darn it my card is useless and worth nothing now. I dont' forsee a gpu purchase in my future. That makes me sad.

-MN Scout
 
Sorry if this forum cat is dedicated to F@H :D I just saw Distributed Computing and felt it was the closest to the subject matter at hand. I was necessarily going to be running F@H on it
 
OverlordQ said:
Sorry if this forum cat is dedicated to F@H :D I just saw Distributed Computing and felt it was the closest to the subject matter at hand. I was necessarily going to be running F@H on it
Click to expand...
don't worry, you can post non-FAH stuff on here...we're just very FAH centered :eek:
 
You must log in or register to reply here.
Back
Top