So.. This... Is... The... Power... Of.... Kaby.. Lake...

>>57177179
it's fucking nothing as always.

This is what you get for falling for the "new hardware" meme.

8% higher clocks, 4% faster.

Did IPC go down again?

>i5 6600k 3.5-3.9 GHz
wew lad

>4.25% faster in a CPU bound game with higher clocks


lol

I bet the CPU is now over 60% GPU, rest is uncore stuff, cores and cache.

When will this joke end.

>>57177541
When you stop buying it.

>>57177601
I'm still on a fucking Ivy Bridge.

>>57177179
>gaming benchmark

4k master race

>>57177607
Still on Sandy. My 5.4 life is awesome.

>>57177179
if it was over 10% they would have to invent a new socket for it :^)

>>57177721
Even if it wasn't you'd get a new socket next year.

And another 5% bump in WHOOAA PERFORMANCE

This shit makes me think they're saving their really good stuff for when AMD makes an actual competitor. Is there any evidence for that?

>video games

*scoff*

*guffaw*

*vanish*

>>57177741
That's just wishful thinking.
Intel is constrained both by architecture, power and silicon, mostly power, so they can't improve performance more.

>>57177741

14nm Coffee Lake > 10nm Cannonlake in 2018

shifted focus on teh internet of things, data center clouds and memoreh.

>>57177179
>gaymen bench

What is your point?

>>57177741
it's plausible

it's more likely that Intel is focusing R&D elsewhere though, trying to piece other markets

>>57177807
Gaymers are the ones buying new hardware. Or rather, buying new hardware at a reliable rate.

>>57177823
Probably the low power market to try and catch up to ARM.

Well now that this is guff, shoukd I just get Skylake instead of going for Kaby Lake?

I want to try to find a MOBO that doesnt have BCLK locked out

File: 1475466292695.jpg (91KB, 768x1024px) Image search: [Google]

91KB, 768x1024px

tfw all the lakes have integrated backdoors

>>57177836

Not really, enterprise hardware is where the money is.

File: 1472439761402.png (50KB, 801x446px) Image search: [Google]

50KB, 801x446px

I just want the i7-7770 because I like sevens.

Why doesn't intel just remove their ME if they need more efficiency?

Why doesn't intel release chips with just the modern subset of the x86_64 ISA to be more efficient and stop wasting silicon for "muh backward compatibility with 1985 OSs". or completely ditching x86 as desktop PCs are dying and ARM is in dire need of improvement?

>>57177943
ARMshit doesn't need improvement, it needs to be exterminated.

>>57177855
Too true. Look at atom. I even see Intel SBCs when I'm looking at ARM SBCs at Amazon.

>>57177883
Which is the main reason no progress ever gets made. Dumb gamers buy new hardware every 2 years regardless of the ~5 fps improvement so they just keep churning out crappy incremental chips.

Processors are changed far less frequently. If you're only swapping CPUs every 6 years or something, annual incremental change isn't as bad since it adds up.

File: 1477022826145.jpg (73KB, 640x628px) Image search: [Google]

73KB, 640x628px

>>57177179
>people are hyped about kaby lake
>its just a 10% increase in performance as always

>>57177712
Still on Westmere-EP lad.

File: cap65412.jpg (177KB, 1603x892px) Image search: [Google]

177KB, 1603x892px

>>57177179
~~~40-50%~~~ of that boost is coming from the higher stock clocks alone. that's pretty shit improvement. then again it is a skylel refresh.

wait for zen or go broadwell-e if you want a worth while upgrade.

All I'm praying for is Zen being reasonably priced.

>>57178049
10% faster because it has 300mhz faster stock/turbo boost clock. Clock to clock is like 2% faster

X86 has pretty much reached it's maximum potential.
The only improvement made in the past years always consist in adding some specific circuit to optimize a specific operation (e.g. avx).

>>57178049
>10% faster in one synthetic
>3% faster in real life workload

Whoa

>>57178181
x86 hasn't reached shit, what reached its end is the Von Neumann arch, specifically the bottleneck that's keeping us back is memory latency.

>>57177211
it's 9% better in the only cpu relavant benchmark there so it's the same