Overkill with i7 4770k ?

1) The FX 8350 does NOT have 8 cores when it comes to gaming. Gaming is almost entirely reliant on floating-point calculations. The 4 modules that the FX 8350 have contain two "calculating cores," yes, but have to share a single FPU.

2) The reason people say to get the i5 is because no, it's not really the best you can get for gaming. It's pretty much the exact same as an i5. The only differences between the chips is that the i7 has 8MB of L3 cashe compared to the i5's 6MB (negligible difference), and hyperthreading. Hyperthreading only gives about 30% extra performance, and it only applies to double-precision calculations, NOT the floating point calculations that any halfway-decently-coded game will be using. That means that it performs pretty much exactly the same as the i5 - in fact, a lot of people get lower framerates but smoother gameplay with hyperthreading turned off.

Oh, and 3) ...a titan is a really dumb option, do not blow your money on it. The GTX 780ti performs BETTER than a titan at gaming, and the only things it's lacking are 3 more GB of VRAM and the compute performance.

The i7 4770k is not an ovekill even just for gaming.
And it is Better then the fx 8350 and normaly the any i5.
also a better gpu like gtx 690,gtx 780 ti helps in performance

Okay, I just wanted to make sure you knew the Titan was a bad option.

It depends on how much you're going to render. Quite honestly, I consider it to not be worth the $100 over an i5 unless you're doing the rendering for work or for school - i.e. times when you're on a deadline and a quick render means money... but then, I'm perfectly happy to wait a little bit longer for them, so that's up to you. As for everything else, you're perfectly fine with what you've got.

No, the i7 will not be more future proof for gaming. Period. Hyperthreading only applies to double-precision workloads, and games do not need and never should need that kind of precision. If a game is making enough of those calculations for hyperthreading to make a difference, I would consider its back end to be poorly coded.

That being said, you shouldn't have to upgrade either an i5 OR an i7 any time soon - if you overclock it a little bit, it will last you for a very long time to come.

I... what even is that post? That's nonsensical. (EDIT: He had a link to this page.)

No, an i7 is NOT normally better than any i5 by any means - they perform almost exactly the same for gaming.

And yes. A better GPU does help in performance, but the OP has a 770, and the issues are coming from a CPU bottleneck in this case, not general performance complaints.

Yep, you're going to be set for a long time.

You should update your hardware sig, though.

What about a socket 2011 i7?

Isn't what we're talking about in this thread.

it'll also perform worse for GAMING than a modern i5, except in the very very very few games that can take advantage of six cores.

Its only advantage is that it has a whole lot of PCIe bandwidth, and with PCIe 3.0, even that is mitigated quite a bit.

No I thought we were talking about the 4770k. And I believe me, you are WRONG about how the core i7, socket 2011 does for gaming. You see I have the 4930k and its a beast. Its out of 10 about 1.2 to 1.5 better then any i5.

...and your mistake right here is thinking that the i5 isn't a beast just because it's cheaper. You're wrong.

Clock for clock, Haswell is faster than Ivy Bridge E. Add on top of that the fact that the Xeons and extreme i7s have such awful thermals that they don't even come with a cooler, meaning you can get way better overclocks with an i5, and things start to become awfully clear.

Yes, an LGA 2011 chip will be very strong at gaming (despite not having PCIe 3.0 and being ridiculously expensive)... but no, it's not better than an i5. Sorry. There's no reason to try to tell people to waste their money - they're way better off keeping it, or spending it on a better graphics solution.

Ok, let me just clarify, I do not think the 2011 sockets are better because they cost more, I simply think they're better because I use them. And i5s and i3s. Not yet have I used the 4770k but from what I read on it, I don't know if I'll get it. But you guys are right though, its a fucking CPU that costed me a paycheck and I should've saved it for a graphics solution, but my way is that you have to have a CPU that'll support your graphics card and have plenty of fast memory.

But whatever you guys think.

No need to get pissy, dude.

I completely understand the confirmation bias. I have that issue too - no more so than when the 600 series was the current gen of graphics cards and I would go off on people for recommending the 680 when the 680 was only 5% better than the 670, for 25% more money.

That being said, the facts are stacked against you, bud... an i5 (while being way better than the effectively dual-core i3) is actually going to outperform a xeon or extreme i7 until games start to have 4-, 6-, and 8- core threading as default... and considering we're barely at dual-threading as a standard, that's going to be a while. :/

Im not pissy man, I'm just saying you guys are right. I should have taken that money and fixed up my i5 4670 system.

It's a lot of personal preference, man.

For example, if you do a lot of photoshop work, or video editing, or rendering, or other applications that are multithreaded well, your CPU is doing a LOT for you, and will trash an i5.

If you had just an i5-4670 that was locked and couldn't be overclocked, the upgrade was huge, because you can overclock now.
If you're running three or four graphics cards, your CPU will handle it better than an i5 would.
Also, your motherboard, being socket 2011, will probably last longer, which could be a long-term benefit.

There are plus-es and minus-es to every choice, is what I'm saying.

I know what you're saying.

Most of what you wrote is horribly misinformed and downright incorrect. Ergo I am unselecting it as the best answer.

1. All tasks, games or otherwise, are dependant on integral arithmetic. Integral arithmetic forms the backbone of discrete computing and has done so for over half a century. Floating point and vector arithmetic are afterthoughts that are bolted on as supplementary instruction sets, but the core of x86 is pure integral. 3D rendering is heavily dependant on floating point calculations, but this is entirely offloaded to the GPU.

2. Hyperthreading allows instructions from two separate threads to be issued to the backend execution pipes in an effort to reduce the total number of unused issue slots over time. This affects all types of unprivileged instructions, not just floating point, and certainly not just double-precision floating point. I don't know where you got that information from, but it is in no way related.

3. The occasional Hyperthreading stutter is a result of poor application coding that results from naive thread spawning. This is not caused by Hyperthreading itself, but by programmers not doing their homework.

1. The best answer is supposed to be given to the best answer, not the best uneducated guess.

2. When 3D APIs are used, rendering is performed by the GPU. The CPU is responsible for executing the API calls to setup the state necessary for the rendering operation to complete properly, but the heavy lifting is done by the GPU. This is why modern GPUs are little more than arrays of floating-point vector machines. The CPU sets up the state and then moves on. Games do make use of floating point arithmetic, but it's for logical reasons only, such as calculating damage, moving around in space, or through software driven physics engines such as Havok or the PhysX on the CPU. Even then, this is bad form and is just asking for performance issues.

The overwhelming majority of all calculations that the CPU performs will be logical operations performed on fixed point integer numbers. Equality tests, inequality tests, boolean algebra, shift and rotate, address generation for load and store operations, branch target computation, in addition to the traditional integer add/subtract/multiply/divide that are used like crazy. Even in a fairly CPU driven game, floating point operations will account for an extremely small fraction of the total code. If you've ever wondered why Intel's quad core microprocessors stomp AMD's octal core microprocessors it's because the Haswell microarchitecture can perform four primitive scalar integer operations per clock cycle along with up to 4 memory operations (Sandy/Ivy perform 3 and 3). The FX series microprocessors can perform only two. So, for an application with a traditionally fixed level of concurrency (gaming), the Haswell wins out every time by up to a 100% margin core for core and clock for clock, but when concurrency is increased they trade blows clock for clock because 8 cores performing two operations per cycle is the same as 4 cores performing four operations per cycle.

3. It most definitely is correct, but there's a reason why games traditionally don't benefit from it. Hyperthreading allows for more total work to be done within the same number of cycles, but in order for it to be an effective technology it cannot discriminate against running tasks. When two tasks are competing for resources as opposed to having exclusivity over the resources the total resource utilization will be higher but the completion time of both tasks will be pushed back. For example, if tasks A and B are run sequentially on the same microprocessor, task A will finish in 1 time unit, and task B will finish 1 time unit later for a total runtime of 2 time units. If they are both run concurrently on a hyperthreaded microprocessor, both tasks may finish after 1.7 total time units as a result of the greater resource utilization. Task B finishes earlier, but task A finishes later. Gaming is very sensitive to these real time variations, so having task A delayed by 0.7 time units may cripple performance if task B is not essential to the game's performance. Game developers have to be careful to work around this.

I got I7 and it's not very useful for gaming really. I mean, it's good sure, but I measured the CPU usage and such and came to conclusion that Hyper-Threading did nothing worth of a note for me.

Kinda wasted money really, but oh well, it's good for the e-peen purposes, plus there was a sale on those in my area, so f*ck it.

Well I told you my opinion, if I could get back in time and save 100$ by going I5, I would have done it, because I7 is nice and all, but it's pretty much a "pay 100$ and MAYBE you will get like up to 15% boost in something you barely use IF it supports the thing to begin with".

And honestly, looking at your current setup, I'm not even sure what kind of boost you expect to get anyway? Going from FX8350 to Haswell I5 is not really noticeable anyway, I mean I5 is better, but by what? 10%? What's the point?

GPU upgrade I can understand, although what you have is really decent as it is, but CPU makes no sense to me.

I7 should very well handle streaming better than an I5.