Intel i5 3570k vs i7 3770k vs AMD 8350 (or the one from 9k series soon) ?

1, from those you listed? none.
You might want to get 8350 since you've already got a mobo for that, but I think it's better to wait a while.
Also, the 9-series have stirred some sort of a mixed emotions in both war camps - do not trust anyone who says it's good or bad till the reviews/benchmarks are out.

ad games - gaming performance depends heavily on the GPU. Go SLI if you feel like it - will get you more power than buying new processor (or a CPU/mobo in case of an Intel)...

ad workmule - if you really need to cut/encode videos and play around in photoshop, get as much RAM as you can. Also a big SSD drive to use as a cache - because you can run out of space really fast and those WD Greens are only good as data storage.

2, nope, about 10% better on average plus it got massive boost from AVX2 and FMA in apps that can use them. Also quicksync video is superb for encoding...

FX-63XX for gaming but if your going to do high end editing the cores and threads are what you want, FX-8350 or you can chance overclocking an FX-8320.

If you want to look at a lot of CPU game benchmarks you can click on the following link:

http://www.tomshardware.com/forum/id-1698820/intel-amd-gaming.html

I included many CPU benchmarks from Anandtech all from the same review. I also post many CPU benchmarks from TechSpot of current games including Metro: Last Light and spanning back to May 2011 with Dirt 3 I believe.

Perhaps that can help you out with your CPU choice. Since the Anandtech review is focused on CPU performance they have naturally chosen CPU intensive games for their review.

8350 is a step up from your 8150 and actually more on a par with the OCed 2500K, a 3570K or 4670K will both run circles around the 8350...Could look at the i7s but unless you have a need for Hyper-threading, that's a waste of about $100 that could go to video card or more or faster DRAM

True, many features (additional USB, SATA etc) are available through AMD and have been since about Oct of last year, but then too, Intels precious Chipset release was about 6-7 monthe before that (AMD and Intel go back and forth about every 6 months adding to features to one up the other), still Intel is the overall stronger (3570K or 4670K) CPU, even the much, much older SB 2500K runs on a par with the 8350 in all but video rendering

The SB architecture is a bit slower in a few things actually...Crysis 3, encryption, compression/decompression, QTbench, and rendering. In video encoding, Intel is generally better on the first pass, and the 8350 is generally better on the second pass.

In games, that depends entirely on what you play...some games favor AMD slightly and others favor Intel slightly.

Well, a single 680 is enough for now /or you can sell it and get a single 780 if you feel like it/.
For games - I wouldn't mind any Unreal Engine based games, they scale pretty good with any card. As for Tera - I got a bit dated GTX570 and still get decent 60 FPS /sometimes my FPS drop a lot, but that's mostly attributed to my slow hard drive and/or lags/.

As for i5 vs i7 - well, Intel names their processors after their relative performance to each other on a 1-9 scale with an 4 core i5 being the reference. i7 is the same processor with HT enabled. HT gives the processor additional 4 logical cores /so it acts as 8 core processor/. Practically speaking you get about 30-40% boost per core, hence the i7 name.

Btw AMD uses the same trick - the execution is way different, but since they switched to modular architecture, they 'cheat' on the cores count. The FX8350 uses only 4 modules each equipped with 2 crippled "cores". And that's the reason they are beaten by any Intel processor at the same frequency when it comes to single-threaded apps. Also that's the the reason they shine at heavily threaded apps.

As for the games - you are mostly /95+%/ correct. Most games can use 4 cores max, so the HT doesn't matter much. It helps a lot with Vegas, PS, and other stuff you use and you get some boost in some games that can use all cores...

I'm not all that big a fan of reviews and Benchmarks, because in my mind all they do is provide a low level baseline of the potential of a component for use in the real world - but I found this review which was pretty much dead on to my own testing

http://benchmarkreviews.com/index.php?option=com_content&task=view&id=962&Itemid=63&limit=1&limitstart=2

it compared the 8350 to a 2500K on a Z77 mobo, sad part he does't OC the 2500K to it's potential...regardless when I tested the 8350 against my 2500K the Intel stood tall, and 3570K really shined...but end result the 8350 was tops (by a little bit in video rendering) but that was about it...Haswell should be about 10-12% even better...so to me (and in particular to my clients), it's no contest, I haven't even had a client want to talk about AMD since shortly after the release of the 8150 Dozer which was such a flop (I actually had high hopes for that, thought AMD was going to get back in the game, but is was so much more hype than substance, I actually bought one of the first 8150s released and built a rig, tested it, and the sold it at a loss, it had been a waste of my time

Well, obviously, that was the first generation of completely uncharted territory. As much as I would like to think PD should have been BD, and SR should have been PD, they needed to "rush" and get things out quickly. I think without a firm grasp of what the architecture would handle, they overpromised and under delivered. Though, with a good idea of how the architecture behaves, they're fine tuning and adjusting. With 4th gen intel's being a real world 4-6% gain and only about 1% in gaming...I see the door wide open for AMD's steamroller, which is a claimed 30% improvement, to come roaring back. At which point, intel's process woes will be even more exposed by the fact that their improvement generation over generation has been lackluster in their attempt to keep their nodes ever shrinking.

If intel would stop trying to go smaller and smaller without spending enough time at a given node to get it right, they might actually make something that is a dramatic improvement. It has taken AMD 2 generations of batches of 32nm HKMG SOI to get PD "dialed in". I cannot see how intel could expect to go any faster, even with their R&D budget. They keep laying eggs right now with each new release...(and I don't mean golden eggs either).

Ultimately, AMD has an opportunity in front of them right now. If Kaveri comes out and performs where it's advertised, that bodes extremely well for steamroller cores on FX series. I, personally, think we'll see great leaps forward in the next 12 months from AMD...will they catch intel "sleeping"? Who is to say...though I cannot expect that the die shrink and expected plethora of BGA formats will bode well for DT users at this stage in the game. Intel may lose a large % of DT market share if broadwell is as bad as hasfail has shown to be at this point. I would love to see AMD gain enough market share to represent a large percentage of the minority. I wouldn't expect something like a double in market share...but if they increased market share by 10+% that would put them around 40+% of DT market, and Intel would be scrambling.

No doubt in my mind AMD is going to breakthrough....which will be good for them, and I'll prob try the next line also, but we also have to keep in mind, that Intel is already holding back, Haswell is not, as the military advertise ALL IT CAN BE, and I seriously doubt Broadwell will be either, from what I've heard from Intel folks I know, Broadwell will (they hope) be a stepping stone into the DDR4 life cycle, and we may see like with the vanishing socket 775, mobos that run Broadwell taking the high end DDR3 and the newer DDR4, then they will loose the Skylake era on us starting with Skylake (at 14nm, same as Broadwell) and followed by Skymont)

I honestly doubt intel is holding back. Here's why:

I studied EE and computer hardware for 2 years before switching to game design, so keep in mind, this isn't insider knowledge, but it's a good speculation...

1.) Each process shrink past a certain point will see diminishing returns in performance with each node smaller. As you shrink transistors, if you don't improve the materials, they can handle less voltage and have lower frequency caps because of the limitations of the materials being used to manufacture them.

2.) Once we begin encroaching upon 10nm and smaller, the transistors will begin to get so small that we likely will only see them being useful for ULV tasks. The only way that would get any better would be one of 2 ways, you either have to: (A) Change materials dramatically to increase performance and reduce resistance to change the outcome, or (B) Dramatically change your process and inherent design to improve efficiency.

3.) Tri-gate FinFET on bulk wafers is nearing the end of it's capability to provide gains at smaller processes. If Intel goes to SOI, their production costs would skyrocket because of wafer cost. Though this would provide better thermal properties and improved performance, the question for them is, will the cost be worth the gains? To this point it hasn't been, I suspect it won't be in their eyes for some time to come yet, as it would price them completely out of the market at the same margins on their chips.

4.) A new process is extremely difficult to fab and produce...because of Intel's commitment to Tri-gate, they're locked into this process until it runs into the ground. I don't see them committing to a new design until at least past 10nm. This means the results they are returning will continue to be the same scale increases or marginally lower per generation because of diminishing returns on improvements.

5.) AMD has a long way to go before they have to make the same decision as intel, and they're competitive now. Ultimately, the playing field will eventually level completely because of the limits of materials and processes and architectures. When that happens, AMD will be at an advantage with their current architecture direction. Though, this is likely at least 3-5 years out...it's not far off barring massive sweeping changes.

6.) Room temperature superconductors are quite a way off still...though there have been several advancements recently, it still won't be soon enough to benefit either company. That is the only way to change materials at 10nm and smaller that would make current designs at that node a reasonably attainable goal with larger performance gains than we are currently seeing per generation from Intel.

-my 2 cents.

That's easy...if you stuck 2 Phenom II X4 965 BE's together (140W TDP) and downclocked them to 2.9 GHz all you need to do is market it for servers and you have that chip...

http://www.cpu-world.com/CPUs/Bulldozer/AMD-Opteron%206386%20SE%20-%20OS6386YETGGHK.html

That's the 16 core AMD version...with 140W TDP.

EDIT: Notice the Intel version is $800 more?

8350rocks

Good info, know I appreciate it

I never said it was fake...you must've misread.

I said that what you showed me isn't that hard to do, and then linked you to a 16 core AMD Opteron CPU that was $800 cheaper with similar specs and similar power consumption.

Don't start trolling.

EDIT: Of course it's a server CPU...it's a Xeon. What else would it be?

By your logic, AMD are holding back...they could release a G34 socket 16 core Opteron to the general public for everyday desktop PCs just as easily.

Well, you seem to forget that Intel wanted to launch 8+8 core i9 SB processor, but then decided against it. The TDP was around 150W. But, since the introduction of the Core architecture AMD was forced to play second league, there was never need for such a consumer processor.
The pros will always consider Xeon, which is a premium priced 8 core, and releasing consumer and wallet friendly 8 core would diminish their sales. And no one halfway sane would want to compete with himself

Also, there is a lot interesting tech in the Intel server CPUs. And when the time is dire, there will surely be another major architecture overhaul like it was 7 years ago.

http://gpuboss.com/gpus/GeForce-GTX-780-vs-GeForce-GTX-680
well, the 780 gets around 20%-50% better frame rates.... but if it's really worth?
If I wanted to build a new computer, I'd go for the 770/780. But as an upgrade, I'm not so sure. Well it is better, but you also need to have power and means to feed it.

and for Tera - guess some system tweaking /disabling any unused services, programs and basically killing anything that eats some system resources and is not really necessary/ and my i7 help a little bit.

Pure performance wise the 2400's would provide more bandwidth so more could be done each clock cycle - here however have 16GB vs 8 is also a big factor, depending on what you do (or want to do, i.e. with multi tasking more is often better, or could take 4 of the 16GB or so and make a RAMDisk)