Dual core options

[Rick]

Is it more productive to run one session of FaH and have both CPU's running on it (large work unit) or simply two regular sessions?

 

[SmokeRngs]

F@H isn't multithreaded. You have to run two instances from different directories to get it to use both cores.

 

[marty9876]

FaH is not multi threaded/don't run on 2 CPU's anyways.

If you have a dual core, run 2 instances (at lease one has to be console version, I suggest both be console) and be sure to use the -local switch.

Different WU's combos (config's) produce different amount of points.

Max points is 1 QMD (-advmethods -big packets) and 1 -big packets I think.

Depending on how you use the machine you can very things up a bit (no QMD's 2x- big packets or 1 QMD one timeless ect)

Good luck!

 

[Rick]

Qmd?

 

[marty9876]

Qmd?

big point boost, at the cost of large ram usage (200-300 meg ram) and floods the memory bus. Overall they cause me no troubles in games ect, but in the workplace/some cases they can be nightmares.

Basically set the client for big packets, and use -adv methods flag, and have >504 meg available ram 90% chance you will get one.

Big points boost...

 

[Rick]

Just saw that there's a "new to me" core 5.04 that lets you set a memory limit for the big units.

 

[marty9876]

the neat part of 5.04 beta is you can set the -adv methods flag with out doing a reg edit.

The memory limit option is really a "soft" limit (what ever that means...)

Basically the client checks total avalable amount of ram. IF > 256 won't server assign big packets IF > 504 won't assign QMD's.

This is not a limit to how much FaH will use in total. Specifically it's designed for multi CPU systems which want to run big backets and -adv methods BUT do not want QMD's.

IE set to 384 meg ram, get big packets but not QMD's on a system with 2 gig ram for example. Problem is people running with -big -adv flags with 4 instances were getting 4 QMD's for example, big mess.

 

[gnewbury]

ASSUMing you are running it on what's in your sig -
If it's a fairly dedicated box for folding and minor stuff -
Run 3 instances, with one being a timeless tinker. Set the two regular at low prority and the tt at idle.
If you want to steal some cycles for yourself run two regular at idle.
ALWAYS have a tt or two ready for when the net or servers are down.

For now forget about QMD.

From the FAQ:

What about AMD support? We are currently using the Intel Fortran compiler and libraries. This software intentially runs slowly on AMD CPUs (SSE2 is not supported). While there have been hacks around this issue, such modifications violate Intel's EULA. At FAH, we follow software EULAs and are thus bound to this limitation imposed by Intel.
That's not fair to AMD chips! Yes, it isn't fair and AMD is taking actions against this. We hope that things will change, but our hands are tied here. We could give these WUs to AMD machines, but they would have a much worse points per day (less than half of what one gets on Intel hardware) and would not perform well in general.

And the forum
according to Vijay:

There is a misunderstanding/miscommunication of the issue here. It's the Intel library (LAPACK/BLAS) which is limited to SSE2 support on Intel CPUs, not the compiler.

http://www.intel.com/cd/software/products/asmo-na/eng/perflib/mkl/219823.htm

I used to have a little saying: "NT don't do USB"
I'll add to that "Intel don't do AMD, and AMD don't do QMD"

 

[marty9876]

Opps, I've got sigs turned off... Missed his dual being an AMD