http://www.techpowerup.com/210997/intel-to-launch-just-two-lga1150-broadwell-parts.html
 
Damn, the i7 part will only have a boost clock of 3.7Ghz compared to the i7-4790K 4.4Ghz.  Maybe Intel realized they set the Haswell refresh too high with consumers complaining about overheating at stock setting on those chips?  Time to wait for Skylake.

Sooooo...  Are the new 5775C's even going to be any faster than the 4790K's? With the clock speed difference, right off the bat it appears they are going to be slower.  They are going to have to make up the difference in architecture.  Curious.

Skylake will probably be a true upgrade to Haswell. And DDR4 prices for entry level kits really aren't too bad compared to high performance DDR3 kits. Memory controllers should have some improvements too since X99.

I remember when I bought my Z97 motherboard and 4790K thinking that at least I can upgrade to Broadwell too.  Well played Intel...
 
On the other hand, not much to complain about since my 4790K will most likely easily handle games for years to come.

Being as I was coming from an i7-920 at 3.6, my choices were the 4790K or the 5960X.  Being as my rig is used for gaming only, I went with the 4790K.  I'm thinking I should get a good few years out of this before I even have to consider upgrading it again.  

 
It was just a pleasant thought that if Broadwell came out to be a spectacular chip, I could switch to it...but now not so much.  I probably won't upgrade until I start noticing GPU bottlenecks.

While they both have the letter "C", they are unlocked, so we can raise the clocks, who cares what it debuts at?  

That said, the lower frequency could be a hint at marketing, I don't think Intel wants to sacrifice Haswell LGA1150 sales [just yet].  Granted like Ivy Bridge, Broadwell could have a minor CPU-bound <10% improvement that won't show up in games.  That said, we aren't looking at a shrink of, say, Devil's Canyon (8MB L3 cache die).  We're looking at the shrink of a mobile CPU+IGP (Crystalwell, due to the 6MB L3 cache), but fused into a pin-compatible LGA1150 socket.  Intel is either saving those dies for another platform or they were never part of the plan.
 
As for Skylake, my guess:

• It will have at least the 10-15% ICP gains that Haswell had from Sandy Bridge.
• Have a larger L3 cache memory to boost performance, maybe the quad-core will have 12MB; this will add 5-8% performance.
• Like Sandy Bridge from Nehalem, Skylake could debut with higher starting frequency to make it seem worth it.
  • Last time, the 10-15% ICP from Nehalem coupled with the 28% frequency boost resulted in what many reviews regarded as a >40% gain from i7-920 to i7-2600K.  Everyone fell for an Intel factory overclock, they could again-- except if they wanted our attention on Broadwell-C for a while, then Skylake could debut at a lower clock.  Like I said, debut frequency won't matter if it is unlocked.

Granted Skylake quads are purported to be 95W unlike the 65W Broadwell quad-core, due to Skylake's bigger IGP: At 72-eu and DDR4-2400 (w/ eDRAM running equal to DDR4), it will perform on par with a GTX560 SE.  Nothing to do with the CPU, but Intel has to consider it into TDP.

i love my I7 4790.
in power management i have my plan set to high performance, i am using a corsair H-100 for cooling.
after 3 hours of gaming real temp shows my cpu high temp was 70cel. and that was at a 4.6o/c

Fennicillin

lehpronGranted Skylake quads are purported to be 95W unlike the 65W Broadwell quad-core, due to Skylake's bigger IGP:

I would love, and also kind of hate, to see Intel match the iGPU performance of Kaveri on a chip that at the same wattage runs circles in sheer I/O power.

In 3DMark11 scores, Kaveri's 512-sp gets a score of 2302 while Iris Pro HD5200 gets a score of 1957.  Broadwell would close the gap with 48-eu against 40-eu, and having the Core i5-5675C for LGA1150 effectively threatens Kaveri's monopoly of having the best APU per dollar.  At least I'm hoping it is down in the $200 range.  Even the i5-4570R was soldered on and that system was expensive.

Hypothetically Carrizo would extend AMD's lead in IGP, but it isn't coming to desktop, so Broadwell has the lead and so will Skylake by a fair margin (possibly a 3DMark11 score of 3000, equivalent to a GTX675M) unless AMD does something with SkyBridge this fall/winter.  I'd like to think they will refresh Carrizo and port it into desktop, I'd also like to think it will be shrunk further to 20nm or even using HBM; but I don't know.
 
The only reason Iris Pro works out so well is because of the cache memory, it triples the available bandwidth from system RAM which (square root of 3 = 1.73) adds a major performance boost comparing HD5100 w/eDRAM with a 3Dmark score of 1151.  Once AMD adds HBM to their APU, the performance should jump like we'll see from R9-290X to R9-390X; AMD could keep the 512-sp and match up with Skylake.  Certainly by the time Zen APUs come, it will be back to a level playing field.  

Broadwell-E is going to have PCI-E 4.0, and skylake will only have it for skylake-E as well.  Nothing mentioned regarding if Cannonlake will have PCI-E 4.0 for the mainstream.  But NVIDIA's NVLink is required for Pascal cards with an NVLink connector, safely assume there will be variants with PCI-E 4.0?  But maybe only the Titan series for Pascal will have this feature at first considering it is listed as launching in 2016, but I get a feeling it will launch February 2017 at the earliest since there must be mobos with a NVLink slot for the cards to work.  Havent seen anything about NVLink on Skylake, but it seems it would have to be integrated into motherboards by Cannonlake.  But even then anyone that will want a $1000 Pascal Titan will have to buy a new mobo and cpu if NVLink isnt integrated into mobos until Cannonlake since that launches 2017.