After many months of delays, Intel’s true successor to the Core i7 4770/90K CPUs has arrived. Following a procession of rumoured specs, we finally have the first mass-market 14nm desktop CPU.
For many, this will be the first 14nm CPU they have used and even though that doesn’t mean much to the end user as such, it certainly affects CPU behaviour and what you may come to expect from the new line of CPUs.
Node (process): 14nm FinFET Cores (threads): 4 (8) L3 cache (process): 8,192KB CPU clock (Turbo): 4GHz (4.2GHz) Socket: LGA 1151 Memory type / max ratio: DDR4 / 4,133MHz TDP: 91W iGPU: Intel HD Graphics 530
Before we get into that, you should know that this review only examines the high-end K SKU: the Core i7 6700K. A Core i5 model does exist as well, dubbed the Core i5 6600K. The main difference between these two as with prior Core CPUs is the number of processing threads, frequency and total package cache. The Core i7 6700K features the usual four cores, each with hyper-threading support to deliver a total of eight CPU threads. L3 cache remains unchanged from the previous models at 8,192KB or 8MB. Base frequency is at an impressive 4GHz, while the Turbo frequency is 4.2GHz – lower than what we saw previously with the 4790K, which could reach frequencies as high as 4.4GHz.
As one would expect of Intel’s latest CPU generations, specifically over the last three years, graphics play a major role in the design process and as such, Intel has once again made tremendous gains in their IGP (Integrated Graphics Processor) design and performance. In fact a large part of the die area has been dedicated to this functionality along with a TDP (Thermal Design Power) budget to match. To that end, the new CPU actually has a higher TDP than the outgoing 4770K and 4790K. A bit strange you may think, given that it’s essentially replacing CPUs that were built on a larger node. The general consensus would be that with all things equal, the smaller 14nm process should make for not only a smaller CPU die, but a cooler-running one as well with a lower TDP. Reinforcing this would be the fifth-generation Core i7 577C CPU for example. Built on the same 14nm node, this CPU has a TDP of only 65W. Granted that the Core i7 577C operates at a significantly lower clock speed, it would cause one to wonder why the TDP for the 6700K is at a Sandy Bridge i7 2600K-matching 91W (slightly less because the 2600K was 95W).
Well, a large part of that has to do with the IGP. As stated previously, Intel has made significant gains here and it looks as if what was saved on the CPU die is allocated and used by the IGP. A sensible decision given the kind of workloads experienced by typical consumer devices.
Do keep in mind that while the 6700K is strictly a desktop CPU, there are other variations of the Skylake family which will see a home within ultra-portables, notebooks, server environments and perhaps even in small mobile devices. That is all related to the architecture, however, which is not what we are concerned with here. Primarily what we want to know is if this new generation of Intel CPUs makes a compelling case for themselves. This may be specifically important for those still on the older Ivy Bridge, third-generation Core CPUs.
If you skipped the entire Haswell generation of CPUs and motherboards then you may find that the 6700K in particular is exactly what you need, depending on what you view as a worthwhile spend or not. With the new CPU, it brings the additional cost of a new motherboard and memory. That being said you have little to no choice in the matter because should you decide to go with the enthusiast X99 platform instead, you would still be confronted with the same question and budget decisions. As such, you should not place too much importance on the cost element for the platform as a whole. Whichever direction you look and whatever you do, you will need at least three new components.
This is of particular interest because as much as the Core i7 6700K is an impressive CPU, it does face the issue of being reasonably close in pricing to the Core i7 5820K for the X99 chipset. This is a fifth-generation CPU and platform, but for the most part it has parity with all relevant features of the 6700K, barring the IGP. As such, the true value of the 6700K is perhaps in how well it does against the 5820K, rather than how much better it is compared to the outgoing fourth-generation CPUs. This standoff between the 5820K and the lower SKUs isn’t new nor unique to the 6700K; it’s something that, while rarely covered, also had a profound effect on the value proposition of the 4790K. The only redeeming point to be made for the 4770K is that it made use of DDR3, which was and remains very affordable.
This isn’t so with the Z170 platform. We are now squarely in the DDR4 generation and while DDR3 has had an incredible run, it’s time we retired the memory standard. As such, whatever the cost of DDR4 at present, it’ll be the same for both the 5820K and the 6700K. The motherboards (which we’ll cover in future reviews) may be cheaper by and large, but again the difference in pricing between the more high-end options is negligible and in fact in many cases non-existent.
It is in light of this that we have for the first round of benchmarks compared the 5820K, via non-GPU-centric tests, against the Core i7 6700K in Windows 10. We will look at IGP, discrete GPU and overclocking performance at a later stage, when we will detail the various gaming motherboards as well.
As you can see in the results, the Core i7 6700K offers solid performance against the 5820K with no HT. For the most part they are matched CPUs except where threading is the key performance indicator. With HT enabled, the 5820K runs away with the performance, despite the lower clock frequency. Testing with no HT enabled for the 5820K actually shows how closely matched these architectures are even if the node is different. What you won’t see here though and which again we will cover later is how much more flexible the 6700K is over and above all other CPUs Intel has offered to date.
For instance, the Skylake-S CPUs offer a step-less BCLK (base clock). Which means that, unlike with previous platforms where there were specific clock dividers at 100, 125 and 166MHz, the 6700K has an additional external clock generator which allows it to use any frequency as the BCLK. So if you want 115MHz or 200MHz, it’s possible and there’s no artificial limit placed on the system via these locked clock ratios.
This bodes well for DRAM overclocking as well. Not only is the integrated memory controller (IMC) improved, but in conjunction with the step-less base clock, it is possible to reach frequencies in excess of 4,500MHz. In fact, before the end of this year, 3,800 and 4,000MHz kits will become commonplace within the high-end enthusiast circles. The frequency scaling in conjunction with the dramatic fall in DDR4 prices means that DDR3 is effectively going to rise in price but not in performance, thus spelling the end for that memory type.
Powered by this new 14nm process, this processor family will by design allow for far more creative motherboard designs than the previous generation. Gone is the integrated voltage regulator and reliance on fixed clock ratios, so board vendors have the freedom to truly make great products that reflect more on their own engineering capabilities than that of the CPU architecture. It is still early days for the platform, but there’s no question about it: the Core i7 6700K is quite easily the most interesting CPU Intel has built to date.
Step-less base clock
Amazing IMC with plenty of ratios
Impressive IGP performance
5820K offers better performance at a similar price
Scaling under liquid nitrogen is limited
Weak rand is making the CPUs more expensive than they should be
8 Intel has made great strides over the Haswell family with the 6700K. Not only do the CPUs run cooler, they provide better performance and significantly more tuning options. The 6700K is precisely the kind of successor to the 4770K we were expecting.