Don't forget the IPC improvements. The architecture is faster than your Haswell. It's not just about frequency.

 

But some of the same frequency to frequency tests only show a very tiny difference between Haswell and Skylake, so it really does boil down to increased frequency when comparing the two. 

 

But of course if it is upgrade time Skylake is the best choice.

As with all such new tech, the greatest gains come from buying the old tech which is now so `yesterday` as to be hardly worth a mention - so the prices drop dramatically. 

 

Shop cleverly and you could gain 50% fps performance in the sim for not much money, especially if you have a mobo to support a CPU and RAM upgrade, and already have a top performing GPU. 

 

Of course that then just puts the bottleneck elsewhere, but you can't have everything.  :unsure:

 

As I said before, for this tech wait one year for the bugs to be ironed out, and another for the sim to catch up (assuming one is talking about X-P or P3D, no catching up possible with FSX). in this case its going to be the merge with DX12 that defines the real performance increase.

 

So that rules out FSX again !  <_<

But some of the same frequency to frequency tests only show a very tiny difference between Haswell and Skylake, so it really does boil down to increased frequency when comparing the two. 

 

But of course if it is upgrade time Skylake is the best choice.

"some of the tests". It's become quite clear that there's a decent IPC improvement. Intel wouldn't have bothered if switching to 14nm, plus all the other architecture improvements was pointless. Look at the consensus, look at all the tests, not just one. look at the tests relevant to us as gamers.

 

More's law is under stain, you won't get miracles.

IPC gain avarge.

SandyBrige-IvyBrige 6%

IvyBridge-Haswell 11%

Haswell-Skylake 6%

 

Overclocking seems ok 4.6-4.7 with 1.3-1.35v dont look at 4.8 with vcore +1.4V  and more.

 

Read Ians Review at Anandtech he nows overclocking

So it looks like 20% over Sandy plus perhaps another 8% on overclock, certainly less than I had thought. I do like the wireless capabilities however. I think that I'll wait for Kaby Lake, no tick this go round, or see what Skylake-e looks like. 

The die is rectangular they say. Thus the line method of applying TIM is dropping temps by up to 10 degrees.

The die is rectangular they say. Thus the line method of applying TIM is dropping temps by up to 10 degrees.

 

Are you talking TIM from CPU to cooler or do these CPU's also need you to crack them open and apply TIM under the cover ?

They say Skylake has polymer TIM between die and heat spreader Mark. I'm not sure if we'll be able to de-lid and scrape off the NGPTIM polymer TIM and replace with our own or not. Theoreticaly I guess, but Intel haven't released any details yet.

 

I meant between cooler and heat spreader. Rather than the small blob in the middle method, It's claimed the line method is better.

Excuse my ignorance but what is IPC and how does it play a role in P3d? Thanks.

Instructions per cycle, it's a measure of how much work each clock cycle can perform. Newer architectures perform more work, even if the clock speed is the same. With P3D, as mentioned, the relationship of frames per second to IPC/Ghz is linear.

IPC gain avarge.

SandyBrige-IvyBrige 6%

IvyBridge-Haswell 11%

Haswell-Skylake 6%

 

Overclocking seems ok 4.6-4.7 with 1.3-1.35v dont look at 4.8 with vcore +1.4V  and more.

 

Read Ians Review at Anandtech he nows overclocking

 

I read that one and a few others that seem to agree on about that 6% number on the IPC increase over Haswell. If there are others that show a bigger jump I would be eager to see them. (Not being sarcastic, I really would!)

Thanks -

 

Looks like there is a tremendous difference using higher clocked ram http://forums.anandtech.com/showpost.php?p=37607925&postcount=2692

So it seems the consensus is that there are definitely some IPC gains, which is good, no doubt. But, as CPU-bottlenecked flightsimmers, we simply need more than 6% gains here and there...

 

Even being conservative, if we assume we can only get 4.8Ghz (and truth is it sounds like maybe a bit more), IPC gains of 6% and clock gains/losses of a couple percent on either side of that 4.8, we're still just not getting the boosts we'd really need to break open our favorite aircraft around our favorite airports in our favorite densely-populated metro areas. Why did Moore's Law have to die now?!?!

 

Unless, of course, we get some architectural improvements from developers...

Thanks -

 

Looks like there is a tremendous difference using higher clocked ram http://forums.anandtech.com/showpost.php?p=37607925&postcount=2692

 

DDR4 3600 is about four times as much as DDR4 3000. I wonder if it is worth the price? In any case the faster RAM does bring out a better Skylake.

Question - does p3d take advantage of multi thread? If not I may just go with the 6600k. Considerable difference in price between this and the 6700k over here in Australia.