NEW skylake CPU Performance Benchmarks 19 hours ago (worth the wait ) ????

sosolola

Reputable
Nov 1, 2014
328
0
4,790
Intel-Skylake-Core-i7-6700K-Performance_Cinebench-R15.jpg


Intel-Skylake-Core-i7-6700K-Performance_Photoshop-CC.jpg


Intel-Skylake-Core-i7-6700K-Performance_Winrar.jpg


Intel-Skylake-Core-i7-6700K-Performance_wPrime-2.05.jpg


Intel-Skylake-Core-i7-6700K-Performance_Fritz-Chess-Benchmark.jpg


Intel-Skylake-Core-i7-6700K-Performance_TMPGEnc-5.4.1.jpg


not even close to 3-5% in procesor performance increase

http://wccftech.com/intel-skylake-core-i7-6700k-review-gaming-performance-5820k/
 
Solution
So much for earlier rumors that Skylake might be as much as 20% faster than Haswell, but I am not too surprised after seeing how much of a mixed bag Broadwell was - faster in some cases, worse in others.

Skylake might join the growing list of skippable CPUs for people who already own a Sandy Bridge or newer CPU.

InvalidError

Titan
Moderator
So much for earlier rumors that Skylake might be as much as 20% faster than Haswell, but I am not too surprised after seeing how much of a mixed bag Broadwell was - faster in some cases, worse in others.

Skylake might join the growing list of skippable CPUs for people who already own a Sandy Bridge or newer CPU.
 
Solution

crooked windows

Distinguished
Jul 4, 2013
191
0
18,690
If you are still alive to use it , then yes. delay or vacillation may have unfortunate or disastrous consequences. Simply put Need it or Want it what ever your choice, you will just shift the Bottle elsewhere.
 

SeanS6

Reputable
Aug 1, 2015
5
0
4,510
It does show how much Moore's law has stalled. Most software is still single core for reliability and cost of coding reasons. Hopefully the Java Hotspot just in time compiler will automatically spread the workload over many cores in future. It is the only compiler for which there is a real funding stream. I genuinely think that makes a difference.
 

InvalidError

Titan
Moderator

I think Intel's compiler team has plenty of funding too. The problem is that it is extremely difficult for compilers to infer threads directly from code in a sensible way. The more practical approach is for CPU designers, middleware writers and developers to put together collections of threading-optimized libraries to solve specific classes of problems like the Intel Math Kernel Library (MKL) does.
 

nichpas

Reputable
Jul 12, 2015
4
0
4,510
I think the real problem is the architecture. They have exhausted it. They need to move to a new architecture to really break grounds in performance again. But even still, the size scale they are working on has gotten so much smaller and it requires additional lithographic steps, I don't think that even with a new architecture Moore's Law would hold true. Until new or improved lithographic techniques are developed at least. Never say never with technology. :)
 

InvalidError

Titan
Moderator

Any architecture you could ever possibly imagine will inevitably run into performance and scalability brick walls due to data dependencies, conditional and indirect branches. There is absolutely nothing a CPU architecture can do about those.

A new ISA might reduce the complexity cost of some of those mitigation mechanisms but it will not magically make them go away since they are an intrinsic part of dynamic program execution. The only way to "get rid" of them is by doing stuff like loop unrolling to eliminate conditional branches and some data dependencies between loop iterations.
 

tommyhardon

Distinguished
Feb 10, 2015
9
0
18,510
Broadwell was 5 years in the making.
They said Broadwell wasn't much of an upgrade over Haswell. Wrong. It was.
They said wait for Skylake for better performance. Wrong. Keep waiting indefinately otherwise.
 

ngoni615

Reputable
Jul 23, 2015
15
0
4,510
Boy am I so glad I did not upgrade my 5960x lol skylake seems to be an utter and complete failure performance wise. Just like nvidia titan z lol.. I'm so glad I decided to stick to my 5960x..
 
G

Guest

Guest


I'm pretty sure everyone here knows the 5820K has 6 cores and the 6700K has 4 cores.

I believe the point here is that, if a i7-6700K goes for around $430, or so, and an i7-5820K is at $390, the latter is a better deal for less money with more cores, and more PCI lanes, and it beats the i7-6700K in many benchmark tests.

Each require their own special motherboard and DDR4 RAM, as the 6700K will in most cases, so the cost/benefit seems to favor the 5820K. One might be able to overclock a 6700K higher for a greater single core performance gain but is it really worth it in the long run? SLI and crossfire have better performance on the X99 - 5820K platform, too. So, the 5820K still seems the better choice, especially if one does any video coding/editing/Twitch/YouTube videos in addition to their gaming.

I suppose, in the end, the choice between which to go with may just be more of an individual preference, than anything else. Some just might want to overclock as high as they can go and think the 6700K offers them more there, while others might like the 5820K's solid, more core, beefy cpu performance that currently appears to cost less. Probably can't go wrong either way. Computing is getting so high end for even the average consumer that performance gains will most likely be unnoticed in real world computing activities.