Missed the WD Black 1TB hard drive - if you have a 256GB or bigger SSD, no need for a Black drive. Any reliable HDD will do. Preferably with 1TB per platter. It will save $20-$25.
Steves suggestion on 4x4gb ram is also cheaper/better than 2x8gb ram, as you spread the amount of ram over more channels. And yes, don't buy 'good ram', because they're all the fuckin same. Buy cheap ram, I guarantee it's the same.
Earlier in the thread, S2M (or more like A2M) stated that there will be video editing, so 4770/4770K over 4570/4570K, but since there's little difference in the unlocked price (K) then might as well get it.
For RAM, since Haswell has an integrated memory controller, 2 DIMMS put less stress on it than 4 DIMMS, and it is preferred when overclocking, especially near max multiplier. For RAM frequency, CL and true latency, good frequency ram with good true latency will improve minimum frame rates in demanding multiplayer games (e.g. BF4), limited benefit elsewhere, but video editing and number crunching (benchmarking
) also benefit.
For the record, my rig is a 4770K overclocked from 3.5GHz (3.9GHz turbo) CPU / 3.9GHZ Cache to 4.5GHz CPU / 4.4GHz Cache with DDR3 2400CL10 RAM.
With DDR4 coming quickly, it is unlikely that S2M will buy upgrade this DDR3 RAM, so I advise to get the best frequency / true latency within his budget. Here's an idea of RAM true latency, recommended matched frequency and CL are in bold. Let your wallet do the talking.
RAM Frequency ^ CAS Latency (CL) = True Latency in nanoseconds (ns)
800 ^ 4 = 10ns
1066 ^ 5 = 9.381ns
1333 ^ 7 = 10.503ns
1600 ^ 8 = 10ns
1600 ^ 9 = 11.250ns
1600 ^ 10 = 12.500ns
1866 ^ 9 = =9.646ns
1866 ^ 10 = 10.718ns
2133 ^ 10 = 9.376ns
2133 ^ 11 = 10.314ns
2133 ^ 12 = 11.252ns
2400 ^ 10 = 8.333ns
2400 ^ 11 = 9.167ns
2400 ^ 12 = 10ns
As you can see, if you double or triple the frequency and double or triple the CAS Latency, the true latency stays the same. (800-4, 1600-8, 2400-12 all = 10ns)
Nanoseconds doesn't sound like a lot, but think of it this way, 1600C10 is 25% more latent than 1600C8 for every single memory function. Frequency without consideration of latency isn't the answer to everything, especially if you're not at 100% RAM utilization.
Math for the calc:
((CAS Latency / Frequency) * 1000 ) *2
((4/800)*1000)*2 = 10ns
the *1000 converts it to nanoseconds
the *2 is because DDR = double data rate
