Well, I've been waiting for the Ivy Bridge i7-3770 processors to rebuild my desktop as a power-efficient, always-on folding rig... and now I learn that -bigadv will only accept 16 core CPUs!
Fair enough, but now all my research months ago is out-of-date and I'm here again to ask for input on my new machine.
I'm definitely an advanced user, but not nearly hardcore enough to read many posts of the Most Powerful Folders sticky without wanting to claw my eyes out.
My desktop will be on 24/7 mainly because it will be receiving CrashPlan backups constantly from a dozen or so computers. I want to run FAH at the same time so I'm getting the most use out of the electricity I'm using. I'm willing to pay extra to make it an excellent folding machine as a donation to science but I'm not concerned about bragging rights.
If a few of you are willing to guide me, I'll put in the extra cash to make this new machine more useful to FAH.
DRIVE: The only thing that matters from my current machine is that I have a SSD, Crucial M4-c128m4ssd2. I assume that's worth keeping. I'll also be running about six 2TB drives for data storage, but that won't affect the folding.
CASE: My Thermaltake Xaser III should be fine unless y'all somehow talk me into liquid cooling or modern GPUs.
GPU: I'm currently satisfied with my middle-aged motherboard: ASRock Fatal1ty Z68 Professional Gen3*. It treats me just fine with L4D2, Fallout 3 and the like. If you talk me into GPU-based folding that would obviously need to change.
RAM: I'll buy however much helps with the folding, with appropriate specs. Back in November that seemed to be 4gigs of mid-grade ram.
So I see three options...
1) Get an Ivy Bridge i7 3770 with appropriate motherboard and do CPU-based folding. That's what I had been heading for.
2) Get a more modest CPU (Sandy Bridge i5?) and invest instead in GPU-based folding. I'd do this only if I'd be producing more PPD/watt.
3) Build a 16-core monster or better and do CPU folding. Here too I'd want to produce more PPD/watt and hopefully more PPD/cost as well.
I'd likely keep the same hardware for 5 years or so, so I'd be interested in best guesses as to how long a proposed high-end machine (option 3) would continue to be eligible for -bigadv. But it's not like being demoted to regular units is the end of the world: it would still crank out a lot of PPD.
Anyone care to point me in the best direction for this research investment?




