Isokolon :
You're missing the point. It's not about upgrading now out of spite. If your CPU needs an upgrade, waiting for a year will be a long time. And by the time ice lake is releases we might be talking about 7nm dies and start the discussion all over.
But anyway, a few things to add here. Comparing the advancement of GPUs who become much much stronger with every generation and extrapolating that on CPUs doesn't work anymore.
In the begging you were suggesting 50%, now we're down to 5-10% in the later posts. So should someone wait for a year to gain 5-10%? For ultra high refresh rate gaming, maybe. But for that you need a moneybag anyway. Question was, will the 8700k be obsolete. And with 5-10% extra on new generations, that's a definite no. 5-10% don't render a CPU obsolete and don't justify an upgrade.
As for performance difference -- why are you arguing with synthetic benchmarks? In firestrike an i9 or Threadripper CPU get much better scores than an i7. But would you suggest anyone a Threadripper CPU for gaming? Don't think so.
For gaming, yes Far Cry 4 runs faster on a i7-6700k.
Differences are very small and within the margin of error often. There appears to be a small difference between games that were released prior to the Skylake launce and after the Skylake launch, but differences are still pretty minimal.
Major speed difference is the iGPU which improved significantly.
Synthetic benchmarks and transistor density is all nice and looks really impressive but is not important and doesn't have that much of an impact if at the end of the day if actual performance in real life scenarios is the roughly the same.
Firstly, I never said that the CPU performance difference would be 50% now did I? I clearly pointed to the performance difference between a 980 Ti and a 1080 Ti with everything else being mostly equal going from 28 to 16nm.
Second, the performance difference between a 4790k and a 6770k isn't "only 5-10%" as you still erroneously believe, and we can throw all of the benchmarks you linked to right in the trash because they ALL, every single one of them, are using games that are GPU intensive / bound (yes Fallout 4 with Ultra God Rays will bring any system to it's knees, my 980 Ti couldn't muster 60 FPS at 2560x1440 with this crap, simply disabling it yielded 20 FPS). Per my last post, a 6700k pushes Far Cry 4 at 95 FPS at 4.2 GHz and a 4790k at 78 FPS at 4.4 GHz. This is much more than "5-10%". Try over 20%? 78 to 95 is 17 FPS is over 20%? Considering 6700k overclocks as well as 4790k a 6700k at 4.X GHz is that much faster than a 4790k at 4.X GHz considering a 6700k is 20% or more faster in Far Cry 4, a CPU intensive game, vs a 4790k running 200 Mhz faster? Far Cry 4 is NOT a GPU intensive game, nowhere near as intensive as Metro LL, BF1 or Fallout 4 with God Rays on Ultra, the latter not being an example of gorgeous graphics but because Bethesda uses AA on their God Rays for whatever reason.
Or, we can use the following comparison, I have a 4930k, which is nearly identical to a 4790k in terms of single core speed:
4790k at 4.4 GHz = 10.5k CPU Firestrike
4930k (with two cores removed) at 4.4 GHz = 10.5k CPU Firestrike
They are both 22nm, Haswell brought no real improvements architecturally speaking over Ivy E as we can see, Haswell can overclock to 4.8 GHz or so from what I gather, whereas a 4930k 4.5 GHz is a good OC and the most you will see is 4.6-4.7 GHz at or under 1.4v.
Here's Cemu running Zelda: BOTW on the latest 1.11 build which introduced some new optimizations. In this video, a 7700k at 4.6 GHz is seen holding 37-40 FPS at a demanding area, "Dueling Rocks Stable", with the CPU the bottleneck (this game isn't GPU intensive and the creator is using a 980 Ti and pushing the game at 1440p).
For comparison, I get ~25 FPS here. At least 50% less. Cemu is insanely CPU intensive so this is a fantastic benchmark.
Given that a 7700k isn't really that much faster than a 6700k and given that your 4790k isn't any faster than my 4930k (it just has 4 cores instead of 6) it's easy to see that yeah, we aren't talking about incremental, 5-10% performance difference here. We're really looking at 20-40% or more between at 4790k and a 6700k. But you can believe whatever you want, just be prepared to have your faulty thinking challenged if you go spouting nonsense on the internet.
https://www.youtube.com/watch?v=BzROgsZHreM&t=1s
My comment, which can be found in the comment section of the video:
Vincent Tuminello
2 days ago
I'm using the Patreon Build (paid) of 1.11 and am seeing 25 FPS at Riverside stable. I don't know if it's due to architectural difference between my i7 4930k and your 7700k but I am at 4.5 GHz as well (this CPU does have HT), High Performance / Cores Unparked, SSD (850 Evo) on Win 7 with nothing else open. Out in the open I see about 45-50 FPS, not the 60 youre seeing. Still an improvement over 1.10 nonetheless but yeah, I wish I was seeing what youre seeing. I didn't want to upgrade my CPU this early (only 3 years into build) but yeah, maybe it's time. Sucks. This CPU was basically the fastest thing you could go with in 2014.
https://steamcommunity.com/discussions/forum/11/626329186832966007
https://www.futuremark.com/hardware/cpu/Intel+Core+i7-4790K/review
https://www.3dmark.com/fs/13403818
Facebook
Twitter
Instagram