I'm never so sure if contributions stay with a country if you change flag. I know swapping team makes credit stay with the teams you supported [done it myself and it shows on the MC page], but was that too implemented for the geographical orientation? Probably it would be easier to look back on a small nation as Belarus in hind-cast view to see if there's permanency to this.

/ot end of thread hijack :D

Fuzzydice 555 - If you could, that would be awesome. I am not going to be able to start the build for 5-6 weeks due to being out of the country, hence I have the time to buy parts right now.

KLiK - Pretty much as ASI said. I don't need to populate every slot and I am just buying what I need. Ideally, I would like 4*4GB to get dual channel for each processor, but 2*8GB is way easier to find and cheaper. Thankfully, for something like this, it doesn't affect the performance.

This is entirely false. WCG factors in core/thread count, clock speed, AND architecture. I've done tests on this. My 4790K is currently overclocked to 4.6GHz--a 15% overclock from its stock frequency of 4.0GHz. I ran WCG for a while at stock before overclocking, and hey, what do you know. The overclock saw a roughly 15% performance improvement in time donated, points generated, and results returned. The same is true of my Q6600, as I briefly overclocked it to 3.0GHz from its stock 2.4GHz frequency. I saw a ~20% performance improvement in all areas. (I then turned that one back down to stock speeds because the temps were outside of my comfort zone and the chip is pretty old as it is).

I will concede that I don't currently own any AMD hardware, so I can't test any, but I doubt it's a whole lot different.

Either way, the point is that core count is only one side of the story, and the clock speed and architecture of the CPU certainly makes worlds more difference than any amount of RAM in any channel configuration will for WCG.

Yeah, I was hoping KLiK was joking about the core count...

2011v3 xeons have the memory controller on the CPU, so the northbridge has nothing to do with memory anymore.

The only way to test if single/dual/quad channel matters is if you run the same rig for a week in single channel, then 1 week in quad channel.

@vcd683s: I've created a doc with all the useful commands/etc for installing WCG on Ubuntu Server:

https://docs.google.com/document/d/1XYNJVryuq...dQp-j4zc/edit?usp=sharing

This is an old, old thread, but I thought I'd mention that my dual Xeon V4 (2x 20c/40t) @ 2.3 GHz is currently being outperformed by a dual Xeon V2 (2x 12c/24t) @ 3.1 GHz. The difference: the V4 system has 2 sticks of memory, the V2 system has 12 sticks of memory.

As an example, the mbw benchmark on linux gives (while both systems run WCG @ 100% load, btw):

V4: Method: MEMCPY Elapsed: 1.72534 MiB: 1000.00000 Copy: 579.596 MiB/s
V2: Method: MEMCPY Elapsed: 0.26779 MiB: 1000.00000 Copy: 3734.214 MiB/s