I posted a few hours back about some benchmarks that I got comparing a 2700k and a 3930k, and I've been looking through more and more benches since. To make a long story short, I'm thinking of upgrading. This is more of a future-proofing measure than anything seeing as IB-E will be coming Q3 of next year and it uses the same socket as the 3930k. For the price tag, though, I just need to be sure. I use the machine for just about everything in some capacity or another. I do game and mod games, and I have a 7970 for that. I do some F@H and BOINC, but not constantly. I also run Linux and do a lot of compiles from source. I'm planning on working on going deeper into the programming realm, and would use the machine for compiling there as well. I don't edit video. I don't run Photoshop. I don't render or do any 3D animation. I've heard very mixed things about the benefits of the 3930k, and I know my uses of the computer fall somewhere in a gray area(or seem to for me). So my question is; is it worth the upgrade? I'm also open to other CPU suggestions as long as the price isn't totally out of control. Thanks.
What about large compiles? Gentoo regularly requires compiles of large packages, and I plan to do some Android work in the future that would require kernel compiles. Would the 3930k make a significant difference there?
In my opinion its not worth it.If something is done in 3 minutes with the 2700K it will be done in 2 by the 3930K.Thats is a 50% increase but... i dont know if you're not making money out of it i dont see why you'd upgrade.
Ok, one last question; what about Haswell? I know information is scarce, but is there a chance it can overtake the SB-E platform? Future proofing for easier upgrades was a concern of mine. The idea of putting IB-E CPU into the same socket I had was a big selling point for the 3930k... even the 3820, actually. I'd hate to upgrade to the top of the line when the "average" Haswell CPUs come out and top it. That is more or less what happened to th 980X right?
So Haswell doesn't seem that promising? I know the Haswell-EP(server CPU, I know) looks awesome with a supposed 14 cores and DDR4 support, but the regular Haswell CPUs are going to be about as lackluster as Ivy? As far as futureproofing, I know it's impossible, but I was searching for the best path to updatability without having to update my whole system every few years. That's what looked most attractive about socket 2011.
Its not worth it. Im not 100% sure on the compilers you use but most of them aren't threaded to take advantage of 6 cores. You also don't have anything else to take advantage of the cpu. The things you are currently doing are already very light for an 2700k, you should have just gone with an i5 as hyperthreading is not useful in compiling code or gaming. BOINC and F@H are the only 2 thing that you do to actually use your cpu right now.
It looks like you just want to spend money on something you don't need. Ivy bridge extremes will probably not even be useful to what you are doing so the upgrade path is pointless.
GCC actually does use the threads if it's configured to do so, and I would assume that it would also use the additional ones with the 3930k if they were available. I have actually seen kernel compile benches of the 3930k compared to the 2700k, and the 3930k only has about a 10-15 second advantage. I do understand your point, though. Nothing I do really does use the 2700k save the compiles and BOINC. I guess that's why Tom's "real world" test of the 3930k came out only a 15% increase over the 2700k and 3770k as opposed to the theoretical 50%.
3930K has 12 Mbyte cache and 3820 has 10 Mbytes cache compared with 2700k that has 8 MByte cache you don’t really get high differences unless you have really hard data to feed them but have in mind that they are running full powered at 130 watts because of this. I7 2007K is working at 95 W flat and is easier to cool it..
Go for i7 3820 in LGA 2011 ivy motherboards if you thinking to upgrade NOW at list you would not change motherboard again for the upcoming new ‘wonder’ Intel chips.
Compiles also depend just as much on disk being fast.
If you want to speed up your compiles, you should first look at being on SSD.
Also, I assume you have more than enough memory to shake a stick at.
If you haven't crossed those bridges yet, do those first.
Then overclock (you bought a k processor. Might as well use it, right); then if things are still too slow, then maybe think about cpu upgrade or reconsider why your programs are so sucky that it takes that much cpu power to build...
Consider it like this, the SB-e chips are production chips, in that realm they well and truely beat a 1155 chip, but gaming is a quarter mile sprint and the 39XX chips have slower acceleration, in gaming the 39XX perform the same or worse than a higher end 1155 chip at a 3x the premium rate.
You have absolutely no need for a X79 platform unless you like to blow money.