6th generation intel processors review. Skylake - processor from Intel
On January 3, the birthday of founding father Gordon Moore (he was born on January 3, 1929), Intel announced a family of new processors Intel Core 7th generation and new Intel 200 series chipsets. We had the opportunity to test the Intel Core i7-7700 and Core i7-7700K processors and compare them with the previous generation processors.
7th generation Intel Core processors
The new family of 7th generation Intel Core processors is codenamed Kaby Lake, and these processors are new by a stretch. They, like the 6th generation Core processors, are manufactured using a 14-nanometer process technology, and they are based on the same processor microarchitecture.
Recall that earlier, before the release of Kaby Lake, Intel released its processors in accordance with the “Tick-Tock” (“tick-tock”) algorithm: the processor microarchitecture changed every two years and the manufacturing process changed every two years. But the change in microarchitecture and process technology were shifted relative to each other by a year, so that once a year the process technology changed, then, a year later, the microarchitecture changed, then, again a year later, the process technology changed, etc. However, to maintain such a fast pace for a long time, the company could not and eventually abandoned this algorithm, replacing it with a three-year cycle. The first year is the introduction of a new process technology, the second year is the introduction of a new microarchitecture based on the existing process technology, and the third year is optimization. Thus, another year of optimization was added to Tick-Tock.
The 5th generation Intel Core processors, codenamed Broadwell, marked the transition to a 14nm process (“Tick”). These were processors with Haswell microarchitecture (with minor improvements), but produced using a new 14-nanometer process technology. The 6th generation Intel Core processors, codenamed Skylake ("Tock"), were manufactured on the same 14nm process as Broadwell, but with a new microarchitecture. And the 7th generation Intel Core processors, codenamed Kaby Lake, are manufactured on the same 14nm process (although now it is designated “14+”) and are based on the same Skylake microarchitecture, but all this is optimized and improved. What exactly is the optimization and what exactly improved - so far this is a mystery shrouded in darkness. This review was written before the official announcement of new processors, and Intel could not provide us with any official information, so there is still very little information about new processors.
In general, about the birthday of Gordon Moore, who in 1968 together with Robert Noyce founded Intel, we remembered at the very beginning of the article not by chance. Over the years, many things have been attributed to this legendary man that he never said. First, his prediction was elevated to the rank of law (“Moore's law”), then this law became the fundamental plan for the development of microelectronics (a kind of analogue of the five-year plan for the development of the national economy of the USSR). However, Moore's law had to be repeatedly rewritten and corrected, since reality, unfortunately, can not always be planned. Now you need to either rewrite Moore's law once again, which, in general, is already ridiculous, or simply forget about this so-called law. Actually, Intel did just that: since it no longer works, they decided to slowly consign it to oblivion.
However, back to our new processors. It is officially known that the Kaby Lake processor family will include four separate series: S, H, U and Y. In addition, there will be an Intel Xeon series for workstations. Kaby Lake-Y processors targeted at tablets and thin laptops, as well as some models of Kaby Lake-U series processors for laptops, have already been announced earlier. And in early January, Intel introduced only some models of H- and S-series processors. Desktop systems are focused on S-series processors, which have an LGA design and which we will talk about in this review. Kaby Lake-S has an LGA1151 socket and is compatible with motherboards based on Intel 100-series chipsets and the new Intel 200-series chipsets. We do not know the release plan for Kaby Lake-S processors, but there is information that a total of 16 new models for desktop PCs are planned, which traditionally make up three families (Core i7/i5/i3). For all processors desktop systems Kaby Lake-S will only use the Intel HD Graphics 630 (codenamed Kaby Lake-GT2).
The Intel Core i7 family will consist of three processors: 7700K, 7700 and 7700T. All models of this family have 4 cores, support simultaneous processing of up to 8 threads (Hyper-Threading technology) and have an L3 cache of 8 MB. The difference between them lies in power consumption and clock speed. In addition, the top model Core i7-7700K has an unlocked multiplier. Brief specifications for the 7th Gen Intel Core i7 processor family are listed below.
The Intel Core i5 family will consist of seven processors: 7600K, 7600, 7500, 7400, 7600T, 7500T and 7400T. All models of this family have 4 cores, but do not support Hyper-Threading technology. Their L3 cache size is 6MB. The top model Core i5-7600K has an unlocked multiplier and a TDP of 91W. Models with a "T" have a TDP of 35W, while regular models have a TDP of 65W. Brief specifications for the 7th Gen Intel Core i5 processor family are listed below.
| CPU | Core i5-7600K | Core i5-7600 | Core i5-7500 | Core i5-7600T | Core i5-7500T | Core i5-7400 | Core i5-7400T |
| Process technology, nm | 14 | ||||||
| connector | LGA 1151 | ||||||
| Number of Cores | 4 | ||||||
| Number of threads | 4 | ||||||
| L3 cache, MB | 6 | ||||||
| Rated frequency, GHz | 3,8 | 3,5 | 3,4 | 2,8 | 2,7 | 3,0 | 2,4 |
| Maximum frequency, GHz | 4,2 | 4,1 | 3,8 | 3,7 | 3,3 | 3,5 | 3,0 |
| TDP, W | 91 | 65 | 65 | 35 | 35 | 65 | 35 |
| Memory frequency DDR4/DDR3L, MHz | 2400/1600 | ||||||
| Graphics core | HD Graphics 630 | ||||||
| Recommended price | $242 | $213 | $192 | $213 | $192 | $182 | $182 |
The Intel Core i3 family will consist of six processors: 7350K, 7320, 7300, 7100, 7300T and 7100T. All models of this family have 2 cores and support Hyper-Threading technology. The letter "T" in the name of the model indicates that its TDP is 35 watts. Now the Intel Core i3 family also has an unlocked model (Core i3-7350K) with a TDP of 60W. Brief specifications for the 7th Gen Intel Core i3 processor family are listed below.
Intel 200 series chipsets
Along with the Kaby Lake-S processors, Intel announced new Intel 200-series chipsets. More precisely, so far only the top Intel Z270 chipset has been presented, and the rest will be announced a little later. In total, the Intel 200 series chipset family will include five options (Q270, Q250, B250, H270, Z270) for desktop processors and three solutions (CM238, HM175, QM175) for mobile processors.
If we compare the family of new chipsets with the family of 100-series chipsets, then everything is clear here: Z270 is a new version of Z170, H270 replaces H170, Q270 replaces Q170, and Q250 and B250 chipsets replace Q150 and B150 respectively. The only chipset that has not been replaced is the H110. The 200 series does not have the H210 chipset or equivalent. The positioning of the 200-series chipsets is exactly the same as that of the 100-series chipsets: Q270 and Q250 are targeted at the corporate market, Z270 and H270 are targeted at consumer PCs, and B250 is aimed at the SMB market sector. However, this positioning is very conditional, and motherboard manufacturers often have their own vision of chipset positioning.
So, what's new in Intel 200-series chipsets and how are they better than Intel 100-series chipsets? The question is not idle, because Kaby Lake-S processors are also compatible with Intel 100-series chipsets. So is it worth buying a motherboard based on the Intel Z270 if, for example, the motherboard based on the Intel Z170 chipset turns out to be cheaper (ceteris paribus)? Alas, there is no need to say that Intel 200-series chipsets have serious advantages. Almost the only difference between the new chipsets and the old ones is a slightly increased number of HSIO ports (high-speed input / output ports) due to the addition of several PCIe 3.0 ports.
Next, we will take a closer look at what and how much is added in each chipset, but for now, we will briefly consider the features of the Intel 200 series chipsets in general, focusing on the top options, in which everything is implemented to the maximum.
For starters, like Intel's 100-series chipsets, the new chipsets allow you to mix and match 16 PCIe 3.0 processor ports (PEG ports) to implement a variety of PCIe slot options. For example, the Intel Z270 and Q270 chipsets (as well as their Intel Z170 and Q170 counterparts) allow you to combine 16 PEG processor ports in the following combinations: x16, x8/x8 or x8/x4/x4. The remaining chipsets (H270, B250 and Q250) allow only one possible combination of PEG port distribution: x16. The Intel 200 series chipsets also support dual-channel DDR4 or DDR3L memory. In addition, Intel 200-series chipsets support the ability to connect up to three monitors to the processor graphics core at the same time (just like in the case of 100-series chipsets).
As for the SATA and USB ports, nothing has changed here. The integrated SATA controller provides up to six SATA 6Gb/s ports. Naturally, Intel RST (Rapid Storage Technology) technology is supported, which allows you to configure the SATA controller in RAID controller mode (though not on all chipsets) with support for levels 0, 1, 5 and 10. Intel RST technology is supported not only for SATA -ports, but also for drives with PCIe interface (x4/x2, M.2 connectors and SATA Express). Perhaps, speaking of Intel RST technology, it makes sense to mention the new technology for creating Intel Optane drives, but in practice there is nothing to talk about yet, there are no ready-made solutions yet. Top models of Intel 200 series chipsets support up to 14 USB ports, of which up to 10 ports can be USB 3.0, and the rest - USB 2.0.
Like the Intel 100-series chipsets, the Intel 200-series chipsets support Flexible I/O technology, which allows you to configure high-speed input/output (HSIO) ports - PCIe, SATA, and USB 3.0. Flexible I/O technology allows you to configure some HSIO ports as PCIe or USB 3.0 ports, and some HSIO ports as PCIe or SATA ports. Intel 200-series chipsets can have a total of 30 high-speed I/O ports (intel 100-series chipsets had 26 HSIO ports).
The first six high-speed ports (Port #1 - Port #6) are strictly fixed: these are USB 3.0 ports. The next four chipset high-speed ports (Port #7 - Port #10) can be configured as either USB 3.0 ports or PCIe ports. In this case, Port #10 can also be used as a GbE network port, that is, the Gigabit network interface MAC controller is built into the chipset itself, and the PHY controller (the MAC controller in conjunction with the PHY controller forms a full-fledged network controller) can only be connected to certain high-speed chipset ports. In particular, these can be Port #10, Port #11, Port #15, Port #18 and Port #19. Another 12 HSIO ports (Port #11 - Port #14, Port #17, Port #18, Port #25 - Port #30) are assigned to PCIe ports. Four more ports (Port #21 - Port #24) are configured as either PCIe or SATA 6 Gb/s ports. Port #15, Port #16 and Port #19, Port #20 have a feature. They can be configured as either PCIe ports or SATA 6Gb/s ports. The peculiarity is that one SATA 6 Gb / s port can be configured to either Port#15, or on Port #19 (that is, it is the same SATA port #0, which can be output to either Port #15 or Port #19). Similarly, another SATA 6Gb/s port (SATA #1) is routed to either Port #16 or Port #20.
As a result, we get that in total the chipset can implement up to 10 USB 3.0 ports, up to 24 PCIe ports and up to 6 SATA 6 Gb / s ports. However, here it is worth noting one more circumstance. A maximum of 16 PCIe devices can be connected to these 20 PCIe ports at the same time. Devices in this case are controllers, connectors and slots. A single PCIe device may require one, two, or four PCIe ports. For example, if we are talking about a PCI Express 3.0 x4 slot, then this is one PCIe device, which requires 4 PCIe 3.0 ports to connect.
The distribution diagram of high-speed I / O ports for Intel 200-series chipsets is shown in the figure.
Compared to what was in the Intel 100 series chipsets, there are very few changes: they added four strictly fixed PCIe ports (HSIO ports of the chipset Port # 27 - Port # 30), which can be used to combine Intel RST for PCIe Storage . Everything else, including the numbering of HSIO ports, remained unchanged. The distribution diagram of high-speed I / O ports for Intel 100-series chipsets is shown in the figure.
So far, we have considered the functionality of new chipsets in general, without reference to specific models. Further, in the summary table, we present brief characteristics each Intel 200-series chipset.
And for comparison, here are brief characteristics of the Intel 100 series chipsets.
The distribution diagram of high-speed I/O ports for five Intel 200-series chipsets is shown in the figure.
And for comparison, a similar chart for five Intel 100-series chipsets:
And the last thing worth noting when talking about Intel 200-series chipsets: only the Intel Z270 chipset supports processor and memory overclocking.
Now, after our quick review of the new Kaby Lake-S processors and Intel 200-series chipsets, let's move on to testing the new products.
Performance study
We were able to test two new products: the top-end Intel Core i7-7700K processor with an unlocked multiplier and the Intel Core i7-7700 processor. For testing, we used the stand of the following configuration:
In addition, in order to be able to evaluate the performance of new processors in relation to the performance of processors of previous generations, we also tested the Intel Core i7-6700K processor on the described stand.
Brief specifications of the tested processors are given in the table.
To evaluate performance, we used our new methodology using the iXBT Application Benchmark 2017 test suite. The Intel Core i7-7700K processor was tested twice: with default settings and in the overclocked state to a frequency of 5 GHz. Overclocking was done by changing the multiplier.
Results are calculated from five runs of each test with a 95% confidence level. Please note that the integral results in this case are normalized relative to the reference system, which also uses the Intel Core i7-6700K processor. However, the configuration of the reference system differs from the configuration of the test bench: the reference system uses the Asus Z170-WS motherboard based on the Intel Z170 chipset.
The test results are presented in the table and in the diagram.
| Logical group of tests | Core i7-6700K (ref. system) | Core i7-6700K | Core i7-7700 | Core i7-7700K | Core i7-7700K @5 GHz |
| Video conversion, points | 100 | 104.5±0.3 | 99.6±0.3 | 109.0±0.4 | 122.0±0.4 |
| MediaCoder x64 0.8.45.5852, with | 106±2 | 101.0±0.5 | 106.0±0.5 | 97.0±0.5 | 87.0±0.5 |
| HandBrake 0.10.5, with | 103±2 | 98.7±0.1 | 103.5±0.1 | 94.5±0.4 | 84.1±0.3 |
| Rendering points | 100 | 104.8±0.3 | 99.8±0.3 | 109.5±0.2 | 123.2±0.4 |
| POV-Ray 3.7, with | 138.1±0.3 | 131.6±0.2 | 138.3±0.1 | 125.7±0.3 | 111.0±0.3 |
| LuxRender 1.6 x64 OpenCL, with | 253±2 | 241.5±0.4 | 253.2±0.6 | 231.2±0.5 | 207±2 |
| Вlender 2.77a, with | 220.7±0.9 | 210±2 | 222±3 | 202±2 | 180±2 |
| Video editing and video content creation, points | 100 | 105.3±0.4 | 100.4±0.2 | 109.0±0.1 | 121.8±0.6 |
| Adobe Premiere Pro CC 2015.4, with | 186.9±0.5 | 178.1±0.2 | 187.2±0.5 | 170.66±0.3 | 151.3±0.3 |
| Magix Vegas Pro 13, with | 366.0±0.5 | 351.0±0.5 | 370.0±0.5 | 344±2 | 312±3 |
| Magix Movie Edit Pro 2016 Premium v.15.0.0.102, with | 187.1±0.4 | 175±3 | 181±2 | 169.1±0.6 | 152±3 |
| Adobe after effects CC 2015.3, since | 288.0±0.5 | 237.7±0.8 | 288.4±0.8 | 263.2±0.7 | 231±3 |
| Photodex ProShow Producer 8.0.3648, with | 254.0±0.5 | 241.3±4 | 254±1 | 233.6±0.7 | 210.0±0.5 |
| Digital photo processing, points | 100 | 104.4±0.8 | 100±2 | 108±2 | 113±3 |
| Adobe Photoshop CC 2015.5, s | 521±2 | 491±2 | 522±2 | 492±3 | 450±6 |
| Adobe Photoshop Lightroom CC 2015.6.1, s | 182±3 | 180±2 | 190±10 | 174±8 | 176±7 |
| PhaseOne Capture One Pro 9.2.0.118, with | 318±7 | 300±6 | 308±6 | 283.0±0.5 | 270±20 |
| Text recognition, points | 100 | 104.9±0.3 | 100.6±0.3 | 109.0±0.9 | 122±2 |
| Abbyy FineReader 12 Professional, with | 442±2 | 421.9±0.9 | 442.1±0.2 | 406±3 | 362±5 |
| Archiving, points | 100 | 101.0±0.2 | 98.2±0.6 | 96.1±0.4 | 105.8±0.6 |
| WinRAR 5.40 CPU, s | 91.6±0.05 | 90.7±0.2 | 93.3±0.5 | 95.3±0.4 | 86.6±0.5 |
| Scientific calculations, points | 100 | 102.8±0.7 | 99.7±0.8 | 106.3±0.9 | 115±3 |
| LAMMPS 64-bit 20160516, with | 397±2 | 384±3 | 399±3 | 374±4 | 340±2 |
| NAMD 2.11, with | 234±1 | 223.3±0.5 | 236±4 | 215±2 | 190.5±0.7 |
| FFTW 3.3.5, ms | 32.8±0.6 | 33±2 | 32.7±0.9 | 33±2 | 34±4 |
| Mathworks Matlab 2016a, s | 117.9±0.6 | 111.0±0.5 | 118±2 | 107±1 | 94±3 |
| Dassault SolidWorks 2016 SP0 Flow Simulation, with | 253±2 | 244±2 | 254±4 | 236±3 | 218±3 |
| File operations speed, points | 100 | 105.5±0.7 | 102±1 | 102±1 | 106±2 |
| WinRAR 5.40 Storage, with | 81.9±0.5 | 78.9±0.7 | 81±2 | 80.4±0.8 | 79±2 |
| UltraISO Premium Edition 9.6.5.3237, with | 54.2±0.6 | 49.2±0.7 | 53±2 | 52±2 | 48±3 |
| Data copying speed, s | 41.5±0.3 | 40.4±0.3 | 40.8±0.5 | 40.8±0.5 | 40.2±0.1 |
| Integral result of CPU, points | 100 | 104.0±0.2 | 99.7±0.3 | 106.5±0.3 | 117.4±0.7 |
| Integral result Storage, points | 100 | 105.5±0.7 | 102±1 | 102±1 | 106±2 |
| Integral performance result, points | 100 | 104.4±0.2 | 100.3±0.4 | 105.3±0.4 | 113.9±0.8 |
If we compare the test results of processors obtained on the same bench, then everything is very predictable here. The Core i7-7700K at default settings (no overclocking) is slightly faster (by 7%) than the Core i7-7700 due to the difference in their clock speeds. Overclocking the Core i7-7700K processor to 5 GHz allows you to get a performance gain of up to 10% compared to the performance of this processor without overclocking. The Core i7-6700K (not overclocked) is slightly faster (by 4%) than the Core i7-7700, which is also explained by the difference in their clock speeds. At the same time, the Core i7-7700K model is 2.5% more productive than the previous model. Generation Core i7-6700K.
As you can see, the new 7th generation Intel Core processors do not provide any jump in performance. In fact, these are the same 6th generation Intel Core processors, but with slightly higher clock speeds. The only advantage of the new processors is that they run better (of course, we are talking about K-series processors with an unlocked multiplier). In particular, our copy of the Core i7-7700K processor, which we did not specifically choose, overclocked to 5.0 GHz without any problems and worked absolutely stable when using air cooling. It was possible to run this processor at a frequency of 5.1 GHz, but in the stress testing mode of the processor, the system hung. Of course, it is not correct to draw conclusions on a single instance of the processor, but the information of our colleagues confirms that most Kaby Lake K-series processors race better than Skylake processors. Note that our sample of the Core i7-6700K processor overclocked at best to a frequency of 4.9 GHz, but only worked stably at a frequency of 4.5 GHz.
Now let's look at the power consumption of processors. Recall that we connect the measuring unit to the break in the power circuits between the power supply and the motherboard - to the 24-pin (ATX) and 8-pin (EPS12V) power supply connectors. Our measuring unit is able to measure the voltage and current on the 12V, 5V and 3.3V buses of the ATX connector, as well as the supply voltage and current on the 12V bus of the EPS12V connector.
The total power consumption during the test is the power delivered on the 12V, 5V, and 3.3V buses of the ATX connector and the 12V bus of the EPS12V connector. The power consumed by the processor during the test is the power transmitted through the 12 V bus of the EPS12V connector (this connector is used only to power the processor). However, keep in mind that in this case we are talking about the power consumption of the processor together with its voltage converter on the board. Naturally, the processor supply voltage regulator has a certain efficiency (certainly below 100%), so that part of the electrical energy is consumed by the regulator itself, and the real power consumed by the processor is slightly lower than the values we measure.
The measurement results for the total power consumption in all tests, except for tests on the performance of the drive, are presented below:
Similar results of measuring the power consumed by the processor are as follows:
Of interest is, first of all, a comparison of the power consumption of the Core i7-6700K and Core i7-7700K processors in the operating mode without overclocking. The Core i7-6700K processor has lower power consumption, that is, the Core i7-7700K processor is slightly more productive, but it also has higher power consumption. Moreover, if the integrated performance of the Core i7-7700K processor is 2.5% higher compared to the performance of the Core i7-6700K, then the average power consumption of the Core i7-7700K processor is as much as 17% higher!
And if we introduce such an indicator as energy efficiency, which is determined by the ratio of the integral performance indicator to the average power consumption (in fact, performance per watt of energy consumed), then for the Core i7-7700K processor this indicator will be 1.67 W -1 , and for the processor Core i7-6700K - 1.91 W -1 .
However, such results are obtained only if we compare the power consumption of the 12 V bus of the EPS12V connector. But if we consider the full power (which is more logical from the user's point of view), then the situation is somewhat different. Then the energy efficiency of a system with a Core i7-7700K processor will be 1.28 W -1 , and with a Core i7-6700K processor - 1.24 W -1 . Thus, the energy efficiency of the systems is almost the same.
conclusions
We have no disappointments about the new processors. Nobody promised what is called. Let us remind you once again that we are not talking about a new microarchitecture and not about a new technical process, but only about optimizing the microarchitecture and technical process, that is, optimizing Skylake processors. Of course, it is not necessary to expect that such optimization can give a serious performance boost. The only observable result of the optimization is that it was possible to slightly increase clock frequencies. In addition, the Kaby Lake family of K-series processors overclock better than their Skylake family counterparts.
Speaking of the new generation of Intel 200-series chipsets, the only thing that differentiates them from the Intel 100-series chipsets is the addition of four PCIe 3.0 ports. What does this mean for the user? And it means absolutely nothing. There is no need to wait for an increase in the number of connectors and ports on motherboards, since there are already too many of them. As a result, the functionality of the boards will not change, except that it will be possible to slightly simplify them during the design: there will be less need to come up with ingenious separation schemes to ensure that all connectors, slots and controllers work in the face of a shortage of PCIe 3.0 lanes/ports. It would be logical to assume that this will lead to a reduction in the cost of motherboards based on 200-series chipsets, but this is hard to believe.
And in conclusion, a few words about whether it makes sense to change the awl for soap. It makes no sense to change a computer based on a Skylake processor and a board with a 100-series chipset to a new system with a Kaby Lake processor and a board with a 200-series chipset. It's just throwing money away. But if it's time to change the computer due to the obsolescence of hardware, then, of course, it makes sense to pay attention to Kaby Lake and a motherboard with a 200-series chipset, and you should look first of all at prices. If the system on Kaby Lake turns out to be comparable (with equal functionality) in cost with the system on Skylake (and the board with Intel chipset 100th series), then there is a sense. If such a system turns out to be more expensive, then it makes no sense.
On August 5, Intel announced two new 6th generation Intel Core processors (codename Skylake): Core i7-6700K and Core i5-6600K. In addition, the new Intel Z170 chipset was also announced, and at the same time, leading motherboard manufacturers announced their solutions based on the Intel Z170 chipset.
We had the opportunity to test the Intel Core i7-6700K and Core i5-6600K processors and compare them with the previous generation processors.
Skylake processors
This article was being prepared, as they say, in emergency mode even before the announcement of the new platform, when there was little official information regarding new processors. Therefore, we will leave some questions regarding new processors out of consideration. In particular, we will not consider the microarchitecture of the new processors and the features of the new Intel graphics core. Intel is going to announce the details of the new microarchitecture at IDF 2015, which will be held at the end of August.
So, let's start with the fact that the new family of 6th generation Intel Core processors is known under the code name Skylake. These are processors made using a 14-nanometer process technology. Recall that Intel releases its processors in accordance with the Tick-Tock rule, which was invented by Intel itself. The meaning of the rule is that the processor microarchitecture changes every two years, and the manufacturing process changes every two years. But the change in microarchitecture and process technology is shifted relative to each other by a year. That is, once a year the technical process changes, then, a year later, the microarchitecture changes, then, again a year later, the technical process changes, etc. It occurred to someone very creative to associate such periodic changes in the microarchitecture and technical process with the movement of the pendulum in the clock and the "Tick-Tock" rule arose. Moreover, a change in the technical process is a “Tick” cycle, and a change in microarchitecture is a “Tock” cycle. It cannot be said that Intel strictly adheres to the time frame of this rule, but, in any case, tries to adhere to this rule.
So, the processors of the previous generation, known under the code name Broadwell, marked the transition to the 14-nanometer process technology (“Tick”). These were processors with Haswell microarchitecture (with minor improvements), but produced using a new 14-nanometer process technology.
Accordingly, the Skylake family of processors are "Tock" cycle processors, that is, they are manufactured using the same 14-nanometer process technology as Broadwell processors, but they have a new microarchitecture.
As already noted, on August 5, Intel announced only two models of the Skylake family of processors for desktop PCs. But this, of course, does not mean that the Skylake family will consist of only two models. According to unofficial information, in late August - early September, 8 more models of Skylake processors for desktop PCs will be announced. So far, we are talking about only two models that have an unlocked multiplier (K-series).
In general, the Skylake processor family will include four separate series: Skylake-S, Skylake-H, Skylake-U, and Skylake-Y. Processors of the Skylake-H, Skylake-U and Skylake-Y series will be BGA-based and targeted at laptops, tablets and all-in-ones. Moreover, the processors of these series are SoC (System-on-Chip), that is, they do not require a separate chipset (Platform Controller Hub, PCH).
Desktop systems are focused on the Skylake-S series processors, which have an LGA design and work only in conjunction with a single-chip chipset (PCH). It is about these processors that we will continue to talk.
The Skylake-S series processors have an LGA1151 socket and, of course, are only compatible with motherboards based on the new Intel 100 series chipsets.
One of the innovations in Skylake-S processors is that the processor voltage regulator (Fully Integrated Voltage Regulator, FIVR), which in Haswell processors was located inside the processor itself (and which, in fact, Intel was very proud of), is now moved outside limits of the processor and is located on the motherboard.
Another innovation is that Skylake-S processors will support both DDR3L memory (with a lower supply voltage) and DDR4 memory. Moreover, the memory controllers are dual-channel and support up to two memory modules per channel.
Just like Haswell and Broadwell processors, Skylake processors have a 16-port PCI Express 3.0 (PCIe 3.0) controller that can be used to organize slots for discrete graphics cards or expansion cards.
The new Skylake-S processors also have a new graphics core. For desktop processors, Skylake-S will only use the Skylake-GT2 graphics core, while for the laptop processor family there will be models with Skylake-GT2, Skylake-GT3e and Skylake-GT4e graphics cores.
Recall that the graphics cores, in the code designation of which the letter “e” appears (GT3e, GT4e), use additional memory eDRAM (embedded DRAM). Such memory appeared in the top models of Haswell mobile processors, and Haswell processors for desktop PCs did not have this memory. The eDRAM memory was a separate die, which was located on the same substrate as the processor die. This crystal also became known under the code name Crystalwell.
In mobile Haswell processors, eDRAM was 128 MB in size and manufactured using a 22-nanometer process technology. But the most important thing is that this eDRAM memory was used not only for the needs of the GPU, but also for the computing cores of the processor itself. That is, in fact, Crystalwell was an L4 cache shared between the GPU and the processor cores.
The Broadwell desktop processor family also features a separate 128MB eDRAM die that acts as an L4 cache and can be used by the processor's graphics and compute cores. Moreover, the eDRAM memory in the 14-nanometer Broadwell processors is exactly the same as in the top Haswell mobile processors, that is, it is performed according to the 22-nanometer process technology.
The Skylake-S family of processors will not use eDRAM.
In general, there was practically no data on the graphics core in Skylake-S processors at the time of this writing. It is only known that in the Core i7-6700K and Core i5-6600K models, the graphics core is called Intel HD Graphics 530 (codenamed Skylake-GT2). As for the number of actuators (EU), there is no information about this yet. It is only known that the top version of the new Intel graphics core (apparently, we are talking about the GT4 core) will have 72 EU.
Now we list the characteristics of the Intel Core i7-6700K and Core i5-6600K processors that were known at the time of this writing:
Both processors (Intel Core i7-6700K and Core i5-6600K) have an unlocked multiplier, that is, they are focused on overclocking. The processor multiplier can vary from 8 to 83.
We also note that the Intel Core i7-6700K processor is the top model in the Skylake-S family.
Intel 100-series chipsets
Along with the new 14nm Skylake-S processors, Intel also announced a new Intel 100-series chipset (codenamed Sunrise Point). On August 5th, only one chipset was introduced: the Intel Z170. Later, in early September, several more models of 100-series chipsets will be presented. In total, the Intel 100 series chipset family will include six models: Z170, H170, H110, Q170, Q150 and B150.
The Q170 and Q150 models are aimed at the corporate market and are replacing the Q87 and Q85 chipsets, respectively.
Models Z170, H170, H110 are focused on user PCs and replace models Z97, H97 and H81, respectively. The B150 chipset is a replacement for the B85 chipset and is aimed at the SMB market sector.
Note that if the Intel 9-series chipsets practically did not differ from their predecessors, the Intel 8-series chipsets, then the differences between Intel 100-series chipsets and Intel 9-series chipsets are very significant.
Next, we will consider the features of the Intel 100 series chipsets as a whole, without reference to a specific model, while focusing on the top chipset models in which everything is implemented to the maximum, and we will consider the features of each chipset separately a little later.
To begin with, all Intel 100-series chipsets now have an integrated PCI Express 3.0 controller (earlier chipsets had a PCI Express 2.0 controller), and therefore, you need to distinguish PCIe 3.0 ports from the processor and from the chipset. As noted, Skylake processors have 16 PCIe 3.0 (PEG) ports. The Intel 100-series chipsets allow these 16 PCIe 3.0 processor ports to be combined to provide a variety of PCIe slot options. For example, the Intel Z170 and Q170 chipsets (as well as their Intel Z97 and Q87 counterparts) allow you to combine 16 PEG PCIe 3.0 ports in the following combinations: x16, x8/x8 or x8/x4/x4. Thus, boards with an Intel Z170 or Q170 chipset based on PCIe 3.0 processor ports can have one PCIe 3.0 x16 slot, two PCIe 3.0 x8 slots, or one PCIe 3.0 x8 slot and two PCIe 3.0 x4 slots. The Intel H170, B150 and Q150 chipsets allow only one possible combination of PEG port allocation: x16. That is, on boards with these chipsets, only one PCIe 3.0 x16 slot based on PCIe 3.0 processor ports can be implemented.
Intel 100 series chipsets also support dual-channel DDR4 or DDR3L memory.
In addition, Intel 100-series chipsets support the ability to connect up to three monitors to the processor graphics core at the same time (just like in the case of 9-series chipsets).
The Skylake processor is connected to the Intel 100-series chipset using the new DMI 3.0 bus. Recall that the Intel 9 and 8 series chipsets used the DMI 2.0 bus with a bandwidth of 20 Gb / s in each direction (the bandwidth of the DMI 2.0 bus corresponds to the bandwidth of the PCI Express 2.0 x4 bus). However, given that the Intel 100-series chipsets now have a PCIe 3.0 controller built in, using the DMI 2.0 bus to communicate the processor to the chipset would be counterintuitive, as this bus could become a bottleneck. That is why the chipset communicates with the processor using a faster DMI 3.0 bus with twice the bandwidth.
It is worth paying attention to the fact that, apart from the DMI 3.0 bus, there is no more connection between the processor and the chipset. That is, there is no longer an FDI bus, which previously allowed analog video output through the chipset. Thus, with the advent of a new platform, the VGA connector is becoming a thing of the past. If VGA support will be implemented on motherboards, then it will be due to an additional circuit for converting a digital video signal into analog. But this is unlikely, because it simply does not make sense.
As already noted, one of the main features of the new Intel 100 series chipsets is that they implement a PCI Express 3.0 controller. Moreover, in the top models of chipsets, up to 20 PCIe 3.0 ports are supported (only up to 8 PCIe 2.0 ports were supported in Intel 9-series chipsets).
In addition, as before, there is an integrated SATA controller in the new chipsets, which provides up to six SATA 6 Gb / s ports.
And, of course, Intel RST (Rapid Storage Technology) technology is supported, which allows you to configure the SATA controller in RAID controller mode (though not on all ports) with support for levels 0, 1, 5 and 10. An innovation is the fact that that Intel RST technology is now supported not only for SATA ports, but also for drives with PCIe (x4/x2) interface (M.2 and SATA Express connectors). This option is called Intel RST for PCIe Storage. Moreover, Intel 100 series chipsets support Intel RST for PCIe Storage technology for three PCIe x4 / x2 interfaces, which can be implemented as M.2 or SATA Express connectors. We also note that up to three SATA Express connectors can be implemented on the board using the Intel 100-series chipset.
The number of USB 3.0 ports in the new chipsets has become larger. So, in the Intel 9-series chipsets (just like in the 8-series chipsets) there were only 14 USB ports, of which up to 6 ports could be USB 3.0, and the rest - USB 2.0. Intel 100-series chipsets also have a total of 14 USB ports, but up to 10 ports can be USB 3.0 and the rest USB 2.0. Note that one USB 3.0 port supports the OTG (USB On-The-Go) function (this was not previously the case). Theoretically, this allows you to directly connect two USB-host devices to each other without using a special cable. However, it is not a fact that this feature of the USB port can be used in practice. It all depends on the manufacturer of the motherboard and the availability of the appropriate driver. For example, the Asus Z170-Deluxe board we tested did not support OTG.
Just like Intel 9 and 8 series chipsets, Intel 100 series chipsets support Flexible I/O technology, which allows you to configure high-speed I/O ports (PCIe, SATA, USB 3.0), removing some ports and adding others. However, there is a significant difference between the Flexible I/O technology in Intel 9/8 series chipsets and this technology in Intel 100 series chipsets.
Recall that in Intel 9/8 series chipsets, there could only be 18 high-speed I / O ports in total. All high-speed chipset ports are numbered. Moreover, 14 ports were strictly fixed: these are four USB 3.0 ports, six PCIe 2.0 ports and four SATA 6 Gb / s ports. And here are four more ports that can be reconfigured: two of them can be either USB 3.0 or PCIe ports, and the other two can be either PCIe or SATA 6 Gb / s. In this case, the total number of PCIe ports cannot be more than eight.
The distribution diagram of high-speed I / O ports for Intel 9/8 series chipsets is shown in the figure.
In the Intel 100 series chipsets, a total of 26 high-speed I / O ports can be implemented in total (in the Intel technical documentation, these ports are called High Speed I / O lanes (HSIO)).
The first six high-speed ports (Port #1 - Port #6) are strictly fixed. These are USB 3.0 ports. The next four chipset high-speed ports (Port #7 - Port #10) can be configured as either USB 3.0 ports or PCIe ports. Moreover, Port #10 can also be used as a GbE network port. That is, we are talking about the fact that the chipset itself has a built-in Gigabit network interface MAC controller, but the PHY controller (the MAC controller in conjunction with the PHY controller forms a full-fledged network controller) can only be connected to certain high-speed ports of the chipset. In particular, these can be Port #10, Port #11, Port #15, Port #18 and Port #19.
Another eight high-speed chipset ports (Port #11 - Port #14, Port #17, Port #18, Port #25 and Port #26) are assigned to PCIe ports.
Four more ports (Port #21 - Port #24) are configured as either PCIe or SATA 6 Gb/s ports.
Port #15, Port #16 and Port #19, Port #20 have a feature. They can be configured as either PCIe ports or SATA 6Gb/s ports. The peculiarity is that one SATA 6 Gb / s port can be configured either on Port # 15 or on Port # 19 (that is, it is the same SATA port # 0, which can be output to either Port # 15 , or on Port #19). Similarly, another SATA 6Gb/s port (SATA #1) is routed to either Port #16 or Port #20.
As a result, we get that in total you can implement up to 10 USB 3.0 ports, up to 20 PCIe ports and up to 6 SATA 6 Gb / s ports on the chipset. However, here it is worth noting one more circumstance. A maximum of 16 PCIe devices can be connected to these 20 PCIe ports at the same time. Devices in this case are controllers, connectors and slots. A single PCIe device may require one, two, or four PCIe ports. For example, if we are talking about a PCI Express 3.0 x4 slot, then this is one PCIe device, which requires 4 PCIe 3.0 ports to connect.
The distribution diagram of high-speed I / O ports for Intel 100-series chipsets is shown in the figure.
So far, we have considered the functionality of the Intel 100 series chipsets in general, without reference to specific models. Further, in the summary table, we provide brief characteristics of the Intel 100 series chipsets.
| Chipset | Q170 | Q150 | B150 | H110 | H170 | Z170 |
| Number of high-speed I/O ports | 26 | 23 | 21 | 16 | 26 | 26 |
| Number of PCIe 3.0 ports | up to 20 | 10 | 8 | 6 (PCIe 2.0 only) | up to 16 | up to 20 |
| Number of SATA 6 Gb/s ports | until 6 | until 6 | until 6 | 4 | until 6 | until 6 |
| Number of USB 3.0 ports | to 10 | up to 8 | 6 | 4 | up to 8 | to 10 |
| Total number of USB ports (USB 3.0+USB 2.0) | 14 | 14 | 12 | 10 | 14 | 14 |
| Number of SATA Express (PCIe x2) connectors | until 3 | 0 | 0 | 0 | up to 2 | until 3 |
| Support for Intel RST for PCIe Storage (M2 PCIe x4 or SATA Express PCIe x2) | until 3 | 0 | 0 | 0 | up to 2 | until 3 |
| Possible combinations of 16 PCIe 3.0 processor ports | x16 x8/x8 x8/x4/x4 | x16 | x16 | x16 | x16 | x16 x8/x8 x8/x4/x4 |
A diagram of the distribution of high-speed I / O ports for six Intel 100-series chipsets is shown in the figure.
As you can see, Intel 100-series chipsets are fundamentally different from Intel 9/8-series chipsets.
As already noted, the Intel Z170 (top-end version), H170 (mass solutions) and H110 (budget sector) chipsets are intended for custom motherboards. Most likely, boards based on the Z170 chipset will support DDR4 memory, boards based on the H110 chipset will support DDR3 memory, and boards based on the H170 chipset will most likely be found in both DDR4 and DDR3 memory versions.
It is interesting to note that boards with the Z170 chipset will differ from boards with the H170 chipset not only in the number of PEG slots implemented on the basis of PCIe 3.0 processor lanes. The Z170 and H170 chipsets implement Flexible I/O slightly differently, resulting in H170 chipset boards with fewer USB 3.0 ports and fewer PCIe 3.0 ports that can be used for additional controllers, slots, and connectors.
Now, after our express review of the new Skylake-S processors and Intel 100-series chipsets, let's move on to testing the new products.
test stand
To test the Intel Core i7-6700K and Core i5-6600K processors, we used the stand with the following configuration:
In addition, in order to evaluate the performance of new processors in relation to the performance of processors of previous generations, we also present the results of testing two Broadwell processors (Core i7-5775C and Core i5-5675C models) and a top Haswell processor (Core i7-4790K ). To test the Core i7-5775C, Core i5-5675C and Core i7-4790K processors, the stand of the following configuration was used:
Test Methodology
We tested the Intel Core i7-6700K and Core i5-6600K processors using the same methodology as testing the Broadwell processors. However, being under time pressure, we slightly reduced the testing methodology by excluding such test packages as SPECviewperf v.12.0.2 (most of the tests from the SPECviewperf v.12.0.2 package are included in the SPECwpc 1.2 package) and SPECapc for Maya 2012.
Recall that tests from our scripted benchmarks iXBT Workstation Benchmark 2015 , iXBT Application Benchmark 2015 and iXBT Game Benchmark 2015 were used for testing. As a result, the following applications and benchmarks were used to test the processors:
- MediaCoder x64 0.8.33.5680,
- SVPmark 3.0
- Adobe Premiere Pro CC 2014.1 (Build 8.1.0),
- Adobe After Effects CC 2014.1.1 (Version 13.1.1.3),
- Photodex ProShow Producer 6.0.3410,
- Adobe Photoshop CC 2014.2.1,
- ACDSee Pro 8,
- Adobe Illustrator CC 2014.1.1,
- Adobe Audition CC 2014.2
- Abbyy FineReader 12,
- WinRAR 5.11,
- Dassault SolidWorks 2014 SP3 (Flow Simulation package),
- SPCapc for 3ds max 2015,
- POV Ray 3.7,
- Maxon Cinebench R15
- SPEC wpc 1.2.
In addition, games and gaming benchmarks from the iXBT Game Benchmark 2015 package were used for testing. Testing in games was carried out at a resolution of 1920×1080 in two game settings: for maximum performance and for maximum quality.
Note that due to lack of time for full-fledged testing, we will leave some aspects unattended for now, however, we will definitely return to them. In particular, while we have not considered the overclocking potential of Skylake processors, the possibility of overclocking DDR4 memory (Intel declares that Skylake processors have improved memory overclocking capabilities), as well as processor power consumption.
Test results
Tests from iXBT Application Benchmark 2015
Let's start with the tests included in the iXBT Application Benchmark 2015. Note that we calculated the integral performance result as the geometric mean of the results in logical groups of tests (video conversion and video processing, video content creation, etc.). To calculate the results in logical groups of tests, the same reference system was used as in the iXBT Application Benchmark 2015.
The full test results are shown in the table. In addition, we present test results for logical groups tests on diagrams in normalized form. The result of the Core i7-4790K processor is taken as a reference.
| Logical group of tests | Core i5-6600K | Core i7-6700K | Core i5-5675C | Core i7-5775C | Core i7-4790K |
| Video conversion and video processing, points | 289,8 | 406,6 | 272,6 | 280,5 | 314,0 |
| MediaCoder x64 0.8.33.5680, seconds | 152,2 | 105,0 | 170,7 | 155,4 | 132,3 |
| SVPmark 3.0 points | 2572,8 | 3495,0 | 2552,7 | 2462,2 | 2627,3 |
| Creation of video content, points | 284,7 | 339,8 | 273,3 | 264,5 | 290,9 |
| Adobe Premiere Pro CC 2014.1, seconds | 587,6 | 442,2 | 634,6 | 612,0 | 556,9 |
| Adobe After Effects CC 2014.1.1 (Test #1), seconds | 775,0 | 599,0 | 802,0 | 758,8 | 695,3 |
| Adobe After Effects CC 2014.1.1 (Test #2), seconds | 296,0 | 269,0 | 327,3 | 372,4 | 342,0 |
| Photodex ProShow Producer 6.0.3410, seconds | 456,7 | 426,1 | 435,1 | 477,7 | 426,7 |
| Digital photo processing, points | 219,9 | 305,1 | 254,1 | 288,1 | 287.0 |
| Adobe Photoshop CC 2014.2.1, seconds | 1091,2 | 724,9 | 789,4 | 695,4 | 765,0 |
| ACDSee Pro 8, seconds | 323,5 | 252,7 | 334,8 | 295,8 | 271,0 |
| Vector graphics, scores | 161,9 | 177,0 | 140,6 | 147,2 | 177,7 |
| Adobe Illustrator CC 2014.1.1, seconds | 318,0 | 291,0 | 366,3 | 349,9 | 289,8 |
| Audio processing, points | 220,4 | 270,3 | 202,3 | 228,2 | 260,9 |
| Adobe Audition CC 2014.2, seconds | 475,0 | 387,3 | 517,6 | 458,8 | 401,3 |
| Text recognition, points | 213,8 | 350,9 | 205,8 | 269,9 | 310,6 |
| Abbyy FineReader 12 seconds | 256,6 | 156,3 | 266,6 | 203,3 | 176,6 |
| Archiving and unarchiving data, points | 160,4 | 228,4 | 178,6 | 220,7 | 228,9 |
| WinRAR 5.11 archiving, seconds | 172,9 | 106,7 | 154,8 | 112,6 | 110,5 |
| WinRAR 5.11 unzipping, seconds | 9,1 | 7,4 | 8,2 | 7,4 | 7,0 |
| Integral performance result, points | 216,4 | 287,31 | 212,8 | 237,6 | 262,7 |
So, as you can see from the test results, the Intel Core i7-6700K processor is the leader in terms of integrated performance. However, it outperforms the Intel Core i7-4790K by only 9%. As you can see, the performance difference between these processors is quite modest.
As for the Intel Core i5-6600K processor, in terms of its integrated performance it is a complete analog of the Intel Core i5-5675C processor.
Despite the fact that the Core i7-6700K processor outperforms the Core i7-4790K processor by only 9% in terms of integrated performance, there are a number of tasks in which the advantage of the new Skylake processor is more significant. These are tasks such as video conversion and video processing (MediaCoder x64 0.8.33.5680 and SVPmark 3.0), video content creation (Adobe Premiere Pro CC 2014.1 and Adobe After Effects CC 2014.1.1), as well as text recognition (Abbyy FineReader 12).
But there are also such applications (and there are many of them) in which the Core i7-6700K processor does not have any advantage at all over the Core i7-4790K processor, or this advantage is very insignificant. In particular, in applications such as Photodex ProShow Producer 6.0.3410, Adobe Photoshop CC 2014.2.1, Adobe Illustrator CC 2014.1.1, Adobe Audition CC 2014.2, WinRAR 5.11, the Core i7-6700K demonstrates almost the same performance as the Core processor i7-4790K.
Calculations in Dassault SolidWorks 2014 SP3 (Flow Simulation)
The test based on the Dassault SolidWorks 2014 SP3 application with the optional Flow Simulation package was taken separately, since this test does not use a reference system, as in the tests of the iXBT Application Benchmark 2015 benchmark.
Recall that in this test we are talking about hydro / aerodynamic and thermal calculations. A total of six different models are calculated, and the results of each subtest are the calculation time in seconds.
Detailed test results are presented in the table.
| Test | Core i5-6600K | Core i7-6700K | Core i5-5675C | Core i7-5775C | Core i7-4790K |
| conjugate heat transfer, seconds | 338,0 | 331,1 | 382,3 | 328,7 | 415,7 |
| textile machine, seconds | 440,0 | 391,9 | 441,0 | 415,0 | 510,0 |
| rotating impeller, seconds | 260,1 | 242,3 | 271,3 | 246,3 | 318,7 |
| cpu cooler, seconds | 746,2 | 640,7 | 784,7 | 678,7 | 814,3 |
| halogen floodlight, seconds | 321,0 | 291,0 | 352,7 | 331,3 | 366,3 |
| electronic components, seconds | 455,0 | 477,1 | 559,3 | 448,7 | 602,0 |
| Total calculation time, seconds | 2560,3 | 2274,1 | 2791,3 | 2448,7 | 3027,0 |
In addition, we also give the normalized result of the calculation speed (the reciprocal of the total calculation time). The result of the Core i7-4790K processor is taken as a reference.
As can be seen from the test results, in these specific calculations, the leadership is on the side of the Skylake-S processor. A system based on the Core i7-6700K processor outperforms a system based on the Core i7-4790K processor by 28%. Moreover, in this test, even the Core i5-6600K demonstrates 18% faster computing speed compared to the Core i7-4790K.
SPCapc for 3ds max 2015
Next, consider the results of the SPECapc for 3ds max 2015 test for an Autodesk 3ds max 2015 SP1 application. The detailed results of this test are presented in the table, and the normalized results for the CPU Composite Score and GPU Composite Score are shown in the charts. The result of the Core i7-4790K processor is taken as a reference.
| Test | Core i5-6600K | Core i7-6700K | Core i5-5675C | Core i7-5775C | Core i7-4790K |
| CPU Composite Score | 4,28 | 5,24 | 4,09 | 4,51 | 4,54 |
| GPU Composite Score | 1,66 | 1,75 | 2,35 | 2,37 | 1,39 |
| Large Model Composite Score | 1,77 | 1,86 | 1,68 | 1,73 | 1,21 |
| Large Model CPU | 2,68 | 2,96 | 2,50 | 2,56 | 2,79 |
| Large Model GPU | 1,17 | 1,17 | 1,13 | 1,17 | 0,52 |
| Interactive Graphics | 1,85 | 1,94 | 2,49 | 2,46 | 1,61 |
| Advanced Visual Styles | 1,45 | 1,49 | 2,23 | 2,25 | 1,19 |
| Modeling | 1,40 | 1,49 | 1,94 | 1,98 | 1,12 |
| CPU Computing | 3,23 | 3,76 | 3,15 | 3,37 | 3,35 |
| CPU Rendering | 5,57 | 7,17 | 5,29 | 6,01 | 5,99 |
| GPU Rendering | 2,00 | 2,12 | 3,07 | 3,16 | 1,74 |
In tests that depend on CPU performance (CPU Composite Score), the platform based on the Core i7-6700K processor shows the best result. Moreover, the difference in the result between platforms based on the Core i7-6700K processor and based on the Core i7-4790K processor is 15%.
But in tests that depend on the performance of the graphics core (GPU Composite Score), the leaders are Broadwell processors, which are significantly ahead of both the Core i7-4790K processor and the Skylake-S processors. If we compare the Core i7-6700K and Core i7-4790K processors, then the Core i7-6700K processor shows 26% more high performance.
POV Ray 3.7
In the POV-Ray 3.7 test (3D model rendering), the leader is the Core i7-6700K processor. Although, of course, its advantage over the Core i7-4790K processor is very small (only 8%).
| Test | Core i5-6600K | Core i7-6700K | Core i5-5675C | Core i7-5775C | Core i7-4790K |
| Render average, PPS | 1492,9 | 1889,7 | 1396,3 | 1560.6 | 1754,48 |
Cinebench R15
In the Cinebench R15 benchmark, the result was mixed. In the OpenGL test, Broadwell-C processors significantly outperform Skylake-S processors, which is natural, since they integrate a more powerful graphics core. Moreover, in this test, the Core i7-6700K and Core i5-6600K processors demonstrate higher performance than the Core i7-4790K processor.
But in the processor test, the leader, although with a slight advantage on the Core i7-4790K processor, is the Core i7-6700K processor.
| Test | Core i5-6600K | Core i7-6700K | Core i5-5675C | Core i7-5775C | Core i7-4790K |
| opengl fps | 49,8 | 51,1 | 72,57 | 73 | 33,5 |
| CPU, cb | 598 | 879 | 572 | 771 | 850 |
SPECwpc v.1.2
Well, the last benchmark is a specialized test package for SPECwpc v.1.2 workstations.
The test results are presented in the table, as well as in a normalized form in the diagrams. The result of the Core i7-4790K processor is taken as a reference.
| Test | Core i5-6600K | Core i7-6700K | Core i5-5675C | Core i7-5775C | Core i7-4790K |
| Media and Entertainment | 2,73 | 3,29 | 2,84 | 3,26 | 2,36 |
| Blender | 2,15 | 2,68 | 1,82 | 2,38 | 2,59 |
| handbrake | 2,01 | 2,78 | 1,87 | 2,22 | 2,56 |
| LuxRender | 2,07 | 3,02 | 1,97 | 2,62 | 2,86 |
| IOMeter | 15,34 | 15,52 | 16,07 | 15,87 | 16,06 |
| Maya | 1,1 | 1,11 | 1,71 | 1,68 | 0,24 |
| product development | 2,52 | 2,82 | 2,6 | 2,44 | 2,49 |
| Rodinia | 2,36 | 3,18 | 2,54 | 1,86 | 2,41 |
| CalculiX | 1,88 | 2,05 | 1,49 | 1,76 | 1,97 |
| WPCcfg | 1,93 | 2,13 | 1,98 | 1,63 | 1,72 |
| IOmeter | 18,81 | 19,49 | 20,91 | 20,89 | 21,13 |
| catia-04 | 0,93 | 0,93 | 1,28 | 1,32 | 0,81 |
| showcase-01 | 0,73 | 0,74 | 0,99 | 1,00 | 0,55 |
| snx-02 | 0,19 | 0,21 | 0,19 | 0,19 | 0,2 |
| sw-03 | 1,23 | 1,28 | 1,38 | 1,4 | 1,08 |
| life sciences | 2,32 | 2,74 | 2,39 | 2,61 | 2,44 |
| Lampps | 2,21 | 2,79 | 2,08 | 2,54 | 2,29 |
| namd | 2,16 | 2,8 | 2,1 | 2,46 | 2,63 |
| Rodinia | 1,95 | 2,66 | 2,23 | 2,37 | 2,3 |
| Medical-01 | 0,69 | 0,69 | 0,69 | 0,72 | 0,54 |
| IOMeter | 10,53 | 10,68 | 11,49 | 11,45 | 11,5 |
| financial services | 2,15 | 2,71 | 1,95 | 2,42 | 2,59 |
| Monte Carlo | 2,2 | 2,81 | 2,21 | 2,55 | 2,63 |
| Black Scholes | 2,25 | 2,95 | 1,62 | 2,56 | 2,68 |
| Binomial | 2,01 | 2,37 | 1,97 | 2,12 | 2,44 |
| Energy | 2,11 | 2,56 | 2,18 | 2,62 | 2,72 |
| FFTW | 1,88 | 1,76 | 1,52 | 1,83 | 2,0 |
| Convolution | 1,16 | 2,54 | 1,35 | 2,98 | 3,5 |
| Energy-01 | 0,5 | 0,5 | 0,78 | 0,81 | 0,6 |
| srmp | 2,12 | 3,12 | 2,49 | 3,15 | 2,87 |
| Kirchhoff Migration | 3,19 | 3,93 | 3,12 | 3,54 | 3,54 |
| Poisson | 2,25 | 2,39 | 1,56 | 1,41 | 2,12 |
| IOMeter | 11,05 | 11,04 | 12,22 | 12,27 | 12,25 |
| General Operation | 3,64 | 4,25 | 3,53 | 3,83 | 4,27 |
| 7zip | 1,95 | 2,56 | 1,96 | 2,46 | 2,58 |
| Python | 1,71 | 2,16 | 1,48 | 1,64 | 2,06 |
| Octave | 1,52 | 1,64 | 1,44 |
Translating... Translate Chinese (Simplified) Chinese (Traditional) English French German Italian Portuguese Russian Spanish Turkish
Unfortunately, we are unable to translate this information right now - please try again later.
Introduction
6th Gen Intel® Core™ Processors (Skylake) introduced in 2015. With a host of core, system-on-a-chip and platform-level enhancements over the previous generation 14nm (Broadwell) processor, the Skylake processor is hugely popular in a wide variety of device types for work, creativity and play. This article provides an overview of key Skylake features and enhancements, as well as new usage models such as Wake on Voice and Biometric Login in Windows* 10.
Skylake architecture
The 6th generation Intel Core processors are manufactured using 14nm technology, with a more compact processor size and overall platform for use in a variety of device types. At the same time, the performance of the architecture and graphics is also improved, and advanced security features are implemented. On fig. 1 shows these new and improved features. Actual configuration in OEM devices may vary.
Picture 1.Skylake architecture and summary improvements .
The main directions of development of processors
Performance
Productivity improvements are directly driven by providing more instructions to the executing unit: more instructions are executed per clock cycle. This result is achieved through improvements in four categories.
- Improved frontend. With more accurate branch prediction and increased capacity, instruction decoding speed is increased, and prefetching is faster and more efficient.
- Improved instruction parallelism. More instructions are processed per cycle, and instruction parallel execution is improved with more efficient buffering.
- Improved execution units (IBs). The operation of the executing units has been improved compared to previous generations due to the following measures:
- Shortened delays.
- Increased the number of IB.
- Improved power efficiency by shutting down unused units.
- Improved execution speed of security algorithms.
- Improved memory subsystem. In addition to improving the external interface, parallel processing of instructions and execution units, the memory subsystem has also been improved in accordance with the bandwidth and performance requirements of the above components. For this, the following measures were used:
- Increased download and save bandwidth.
- Improved prefetch module.
- Storage at a deeper level.
- Fill and writeback buffers.
- Improved page miss handling.
- Improved throughput on L2 cache misses.
- New cache management instructions.
Figure 2.Skylake core microarchitecture diagram
On fig. Figure 3 shows the improvement in parallel processing in Skylake processors over previous generations (Sandy Bridge is the second and Haswell is the fourth generation of Intel® Core™ processors).
Figure 3Improved parallelism compared to previous generations of processors
With the improvements shown in Fig. 3, CPU performance has increased by 60% compared to a five-year-old PC, while video transcoding is 6 times faster, and graphics performance has been increased by 11 times.
Figure 46th Gen Intel® Core™ Processor Performance vs. Five Year Old PC
- Source: Intel Corporation. Based on SYSmark* 2014 results for Intel® Core™ i5-6500 and Intel® Core™ i5-650 processors.
- Source: Intel Corporation. Based on the results of the Intel® Core™ i5-6500 and Intel® Core™ i5-650 processors in the Handbrake test with QSV.
- Source: Intel Corporation. Based on 3DMark* Cloud Gate scores for Intel® Core™ i5-6500 and Intel® Core™ i5-650 processors.
For detailed desktop vs laptop performance comparison results, please visit the following links:
Desktop Performance: http://www.intel.com/content/www/us/en/benchmarks/desktop/6th-gen-core-i5-6500.html
Notebook performance: http://www.intel.com/content/www/us/en/benchmarks/laptop/6th-gen-core-i5-6200u.html
Energy saving
Setting resources based on dynamic consumption
Legacy systems use Intel® SpeedStep® Technology to balance performance and power consumption with an on-demand resource connection algorithm. This algorithm is controlled by the operating system. This approach is not bad for a constant load, but is not optimal for a sharp increase in load. With Skylake processors, Intel® Speed Shift Technology transfers control to the hardware instead of the operating system and allows the processor to reach its maximum clock speed in approximately 1ms, providing more precise power management.
Figure 5Comparison of Intel® Speed Shift and Intel® SpeedStep® Technologies
The graph below shows the responsiveness of the Intel® Core™ i5 6200U processor with Intel Speed Shift Technology compared to Intel SpeedStep Technology.
- Response speed increased by 45%.
- Photo processing is 45% faster.
- Graphing is 31% faster.
- Local notes are 22% faster.
- The average response time increased by 20%.
[Based on WebXPRT* 2015 by Principled Technologies*, which measures the performance of web applications in general and in specific areas such as photo processing, note-taking, graphing. additional information see www.principledtechnologies.com .]
Additional power optimization is achieved by dynamically tuning resources based on their consumption: by reducing the power of unused resources by limiting the power of the Intel® AVX2 Vector Extensions when they are not in use, and by reducing the power consumption when idle.
Multimedia and graphics
The Intel® HD Graphics features a host of enhancements in terms of 3D processing, media processing, display, performance, power, customization, and scalability. This is a very powerful member of the in-processor graphics family (first introduced with 2nd generation Intel® Core™ processors). On fig. Figure 6 compares some of these enhancements, providing over 100x improvement in graphics performance.
[Shader Peak FLOPS at 1 GHz]
Figure 6Features of the graphics subsystem in different generations of processors
Figure 7Improved graphics and media handling across generations
9th generation microarchitecture
The 9th Gen graphics architecture is similar to the 8th Gen graphics microarchitecture of Intel® Core™ Broadwell (5th Gen) processors, but improved in terms of performance and scalability. On fig. Figure 8 shows a block diagram of the generation 9 microarchitecture, which consists of three main components.
- Screen. From the left side.
- Out of cut. L-shaped part in the middle. Includes a threaded command handler, a global thread manager, and a graphical user interface (GTI).
- Slice. Includes executing blocks (IB).
Compared to the 8th generation, the 9th generation microarchitecture features higher peak performance per watt, increased bandwidth, and a separate power/clock circuit for the off-cut component. This allows for more efficient power management in usage modes such as media playback. The slicer is a custom component. For example, GT3 supports up to two slices (each slice with 24 execution units), GT4 (Halo) can support up to 3 slices (the number after the letters GT indicates the number of execution units based on their usage: GT1 supports 12 execution units, GT2 supports 24, GT3 - 48, and GT4 - 72 execution units). The architecture is configurable enough to use a minimum number of execution units in low load scenarios, so power consumption can range from 4W to over 65W. Support for the 9th Gen GPU API is available in DirectX* 12, OpenCL™ 2.x, OpenGL* 5.x, and Vulkan*.
Figure 89th generation GPU architecture
For more information on these components, see (IDF link)
Media processing enhancements and capabilities include:
- Less than 1W consumption, 1W consumption for videoconferencing.
- Accelerate raw camera video playback (RAW) with new VQE features to support playback of RAW video up to 4K60 resolution on mobile platforms.
- New New Intel® Quick Sync Video Fixed Function (FF) mode.
- Support for a wide range of fixed function codecs, GPU decoding acceleration.
On fig. Figure 9 shows the 9 generation GPU codecs.
Note. Support for media codecs and processing may not be available on all operating systems and applications.
Figure 9Codec support for Skylake processors
Screen enhancements and features include:
- Blending, scaling, rotating and shrinking an image.
- High pixel density support (over 4K resolution).
- Supports wireless image transmission up to 4K30 resolution.
- Self Update (PSR2).
- CUI X.X - new features, improved performance.
The Intel® Core™ I7-6700K processors provide the following features for gamers (see Figure 10). It also supports Intel® Turbo Boost Technology 2.0, Intel® Hyperthreading Technology, and overclocking capability. The performance increase compared to a five-year-old PC reaches 80%. See this page for more information: http://www.intel.com/content/www/us/en/processors/core/core-i7ee-processor.html
- Source: Intel Corporation. Based on the results of the Intel® Core™ i7-6700K and Intel® Core™ i7-875K processors in SPECint*_rate_base2006 (copy factor 8).
- Source: Intel Corporation. Based on the results of the Intel® Core™ i7-6700K and Intel® Core™ i7-3770K processors in SPECint*_rate_base2006 (copy factor 8).
- Features described are available in selected processor and chipset combinations. Warning. Changing the clock frequency and/or voltage may: (i) result in reduced system stability and reduced system and processor life; (ii) cause failure of the processor and other system components; (iii) lead to a decrease in system performance; (iv) cause additional heat or other damage; (v) affect the integrity of the data in the system. Intel does not test or guarantee operation of processors with specifications other than those specified.
Figure 10.Intel® Core™ i7-6700K Processor Features
Scalability
The Skylake microarchitecture is a customizable core: a single design for two directions, one for client devices and one for servers, without compromising the power and performance requirements of both segments. On fig. Figure 11 shows different processor models and their power efficiencies for use in devices of various sizes and types, from ultra-compact Compute Sticks to powerful Intel® Xeon®-based workstations.
Figure 11.Availability of Intel® Core™ Processors for various types devices
Advanced security options
Intel® Software Guard Extensions (Intel® SGX): Intel SGX is a set of new instructions in Skylake processors that enables application developers to protect sensitive data from unauthorized changes and access. extraneous programs running with a higher level of rights. This gives applications the ability to maintain the confidentiality and integrity of sensitive information , . Skylake supports instructions and threads to create secure enclaves, allowing the use of trusted memory areas. For more information about Intel SGX extensions, see this page:
Intel® Memory Protection Extensions (Intel® MPX): Intel MPX is a new set of instructions to check for runtime buffer overflows. These instructions allow you to check the boundaries of stack buffers and heap buffers before accessing memory, so that a process accessing memory has access only to the memory area that is assigned to it. Support for Intel MPX has been added to Windows* 10 using built-in Intel MPX functionality in Microsoft Visual Studio* 2015. Most C/C++ applications will be able to use Intel MPX by simply recompiling applications without changing source and links to legacy libraries. Running Intel MPX-enabled libraries on non-Intel MPX-enabled systems (5th Gen Intel® Core™ processors and earlier) does not improve or degrade performance. You can also dynamically enable and disable support for Intel MPX , .
We looked at the enhancements and enhancements to the Skylake architecture. In the next section, we will look at Windows components 10 optimized to take advantage of the Intel® Core™ architecture.
What's new in Windows 10
The 6th generation Intel Core processors are complemented by the Windows 10 operating system. Below are some of the key features of Intel hardware and Windows 10 that help you Intel platforms® running Windows 10 are more efficient, more stable, and faster.
Ϯ Intel and Microsoft are working together to provide further support on Windows.
Figure 12.Skylake and Windows* 10 Features
Cortana
Microsoft's Cortana Voice Assistant is available in Windows* 10 and gives you the ability to control your PC with your voice after you speak key phrase"Hey Cortana!" The Wake on Voice feature uses the audio processing pipeline on the CPU to improve recognition fidelity, but it is possible to outsource this feature to a hardware DSP audio with built-in Windows support 10.
Windows Hello*
With biometric hardware and Microsoft Passport* windows service Hello supports various login mechanisms via face, fingerprint or iris recognition. The system supports all of these passwordless login options without installing any additional components. Front camera Intel review® RealSense™ (F200/SR300) supports face-based biometric authentication.
Figure 13.Windows* Hello with Intel® RealSense™ Technology
Photos in fig. 13 shows how the fiducials detected on the face by the F200 are used for user identification and login. Based on the location of 78 fiducial points on the face, a face template is generated the first time a user tries to log in using face recognition. On the next login attempt, the saved fiducial location received by the camera is compared with the saved template. The power of Microsoft Passport, combined with the power of the camera, results in a 1 in 100,000 false admission rate and 2-4% false admission denial.
Links
- Intel's next generation microarchitecture code-named Skylake by Julius Mandelblat: http://intelstudios.edgesuite.net/idf/2015/sf/ti/150818_spcs001/index.html
- Next-generation Intel® processor graphics architecture, code-named Skylake, by David Blythe: http://intelstudios.edgesuite.net/idf/2015/sf/ti/150818_spcs003/index.html
- Intel® architecture code-named Skylake and Windows* 10 better together, by Shiv Koushik: http://intelstudios.edgesuite.net/idf/2015/sf/ti/150819_spcs009/index.html
- Skylake for gamers: http://www.intel.com/content/www/us/en/processors/core/core-i7ee-processor.html
- Intel's best processor ever: http://www.intel.com/content/www/us/en/processors/core/core-processor-family.html
- Skylake Desktop Performance Benchmark: http://www.intel.com/content/www/us/en/benchmarks/desktop/6th-gen-core-i5-6500.html
- Skylake Laptop Performance Benchmark: http://www.intel.com/content/www/us/en/benchmarks/laptop/6th-gen-core-i5-6200u.html
- The compute architecture of Intel® processor graphics Gen9:
This article will take a closer look at the latest generations of Intel processors based on the Core architecture. This company occupies a leading position in the computer systems market, and most PCs are currently assembled on its semiconductor chips.
Intel development strategy
All previous generations of Intel processors were subject to a two-year cycle. A similar strategy for releasing updates from this company was called "Tick-Tock". The first stage, called "Tick", was the transfer of the CPU to a new technological process. For example, in terms of architecture, the Sandy Bridge (2nd generation) and Evie Bridge (3rd generation) generations were almost identical. But the production technology of the first was based on the norms of 32 nm, and the second - 22 nm. The same can be said about Haswell (4th generation, 22 nm) and Broadwell (5th generation, 14 nm). In turn, the “So” stage means a fundamental change in the architecture of semiconductor crystals and a significant increase in performance. Examples of transitions are:
1st generation Westmere and 2nd generation "Sunday Bridge". The technological process in this case was identical - 32 nm, but the changes in terms of chip architecture are significant - the north bridge of the motherboard and the integrated graphics accelerator were transferred to the CPU.
3rd generation "Evie Bridge" and 4th generation "Haswell". The power consumption of the computer system has been optimized, the clock frequencies of the chips have been increased.
5th generation "Broadwell" and 6th generation "SkyLike". The frequency has been increased again, power consumption has been further improved, and several new instructions have been added that improve performance.
Segmentation of processor solutions based on the Kor architecture
Intel central processing units have the following positioning:
The most affordable solutions are Celeron chips. They are suitable for assembling office computers that are designed to solve the most simple tasks.
The CPUs of the Pentium series are located one step higher. In architectural terms, they are almost completely identical to the younger Celeron models. But the increased level 3 cache and higher frequencies give them a definite advantage in terms of performance. The niche of this CPU is gaming PCs entry level.
The middle segment of the CPU from Intel is occupied by solutions based on Core Ai3. The previous two types of processors, as a rule, have only 2 computing units. The same can be said about Kor Ai3. But the first two families of chips do not have support for HyperTrading technology, while Core Ai3 does. As a result, at the software level, 2 physical modules are converted into 4 program processing threads. This provides a significant performance boost. On the basis of such products, it is already possible to assemble a mid-level gaming PC, or even an entry-level server.
The niche of solutions above the average level, but below the premium segment, is filled with chips, occupied by solutions based on Core Ai5. This semiconductor crystal boasts the presence of 4 physical cores at once. It is this architectural nuance that provides an advantage in terms of performance over the Core I3. More recent generations of Intel i5 processors have higher clock speeds and this allows you to constantly get a performance boost.
The niche of the premium segment is occupied by products based on Core Ai7. The number of computing units they have is exactly the same as that of Kor Ai5. But here they, just like Core Ai3, have support for technology code-named Hyper Trading. Therefore, at the software level, 4 cores are converted into 8 processed threads. It is this nuance that provides a phenomenal level of performance, which any price can boast of these chips.
Processor sockets
Generations are installed on different types of sockets. Therefore, it will not work to install the first chips on this architecture in the motherboard for the 6th generation CPU. Or, on the contrary, a chip with the code name "SkyLike" cannot physically be put into the motherboard for the 1st or 2nd generation of processors. The first processor socket was called "Socket H", or LGA 1156 (1156 is the number of pins). It was released in 2009 for the first CPUs manufactured to 45 nm (2008) and 32 nm (2009) tolerance standards based on this architecture. To date, he is outdated both morally and physically. In 2010, the LGA 1155, or "Socket H1" comes to replace. Motherboards of this series support 2nd and 3rd generation Cor chips. Their code names are, respectively, "Sandy Bridge" and "Evie Bridge". 2013 was marked by the release of the third socket for chips based on the Core architecture - LGA 1150, or Socket H2. It was possible to install CPUs of the 4th and 5th generations in this processor socket. Well, in September 2015, the LGA 1150 was replaced by the last current socket - LGA 1151.
First generation of chips
The most affordable processor products of this platform were Celeron G1101 (2.27 GHz), Pentium G6950 (2.8 GHz) and Pentium G6990 (2.9 GHz). All of them had only 2 cores. The niche of middle-level solutions was occupied by Core Ai3 with the designation 5XX (2 cores / 4 logical information processing flows). One step higher were "Cor Ai5" marked 6XX (their parameters are identical to "Cor Ai3", but the frequencies are higher) and 7XX with 4 real cores. The most productive computer systems were assembled on the basis of Kor Ai7. Their models were designated 8XX. The fastest chip in this case was marked 875K. Due to the unlocked multiplier, it was possible to overclock such a price, but he had the corresponding one. Accordingly, it was possible to obtain an impressive increase in performance. By the way, the presence of the prefix "K" in the designation of the CPU model meant that the multiplier was unlocked and this model could be overclocked. Well, the prefix "S" was added to the designation of energy-efficient chips.
Planned renovation of the architecture and the "Sandy Bridge"
The first generation of chips based on the Core architecture was replaced in 2010 by solutions code-named Sandy Bridge. Their key "features" were the transfer of the north bridge and the integrated graphics accelerator to the silicon chip of the silicon processor. The niche of the most budgetary solutions was occupied by the Celerons of the G4XX and G5XX series. In the first case, the L3 cache was truncated and only one core was present. The second series, in turn, could boast of having two computing units at once. The Pentiums of the G6XX and G8XX models are one step higher. In this case, the difference in performance was provided by higher frequencies. It was the G8XX that, because of this important characteristic, looked preferable in the eyes of the end user. The Cor Ai3 line was represented by 21XX models (it is the number "2" that indicates that the chip belongs to the second generation of the Cor architecture). Some of them had a “T” index added at the end - more energy efficient solutions with reduced performance.
In turn, the decisions of "Kor Ay5" had the designations 23XX, 24XX and 25XX. The higher the model number, the more high level CPU performance. The "T" index at the end is the most energy efficient solution. If the letter "S" is added at the end of the name - an intermediate option for power consumption between "T" - the chip version and the standard crystal. Index "P" - the graphics accelerator is disabled in the chip. Well, chips with the letter "K" had an unlocked multiplier. This marking is also relevant for the 3rd generation of this architecture.
The emergence of a new more progressive technological process
In 2013, the 3rd generation of CPUs based on this architecture saw the light of day. Its key innovation is an updated technical process. In the rest, no significant innovations were introduced into them. They were physically compatible with the previous generation of CPUs and could be installed on the same motherboards. Their designation structure remained identical. "Celerons" had the designation G12XX, and "Pentiums" - G22XX. Only at the beginning, instead of “2”, there was already “3”, which indicated belonging to the 3rd generation. The Cor Ai3 line had indexes 32XX. More advanced "Cor Ai5" were designated 33XX, 34XX and 35XX. Well, the flagship solutions of Kor Ay7 were marked 37XX.
The fourth revision of the architecture "Cor"
The next step was the 4th generation of Intel processors based on the Core architecture. The marking in this case was:
CPU economy class "Celerons" were designated G18XX.
"Pentiums" had indexes G32XX and G34XX.
For "Cor Ay3" such designations were assigned - 41XX and 43XX.
"Cor Ai5" could be recognized by the abbreviation 44XX, 45XX and 46XX.
Well, 47XX were allocated to designate "Cor Ai7".
Fifth generation of chips
based on this architecture was mainly focused on use in mobile devices. For desktop PCs, only the chips of the AI 5 and AI 7 lines were released. And only a very limited number of models. The first of them were designated 56XX, and the second - 57XX.The most recent and promising solutions
The 6th generation of Intel processors debuted in early autumn 2015. This is the most current processor architecture at the moment. Entry-level chips are designated in this case G39XX ("Celeron"), G44XX and G45XX (this is how "Pentiums" are marked). Core Ai3 processors are designated 61XX and 63XX. In turn, "Cor Ay5" is 64XX, 65XX and 66XX. Well, only the 67XX marking is allocated for the designation of flagship solutions. The new generation of Intel processors is only at the beginning of its life cycle, and such chips will be relevant for quite a long time.
Overclocking Features
Almost all chips based on this architecture have a locked multiplier. Therefore, overclocking in this case is possible only by increasing the frequency. In the latest, 6th generation, even this possibility of increasing performance will have to be disabled in the BIOS by motherboard manufacturers. An exception in this regard are the processors of the "Cor Ai5" and "Cor Ai7" series with the "K" index. Their multiplier is unlocked and this allows you to significantly increase the performance of computer systems based on such semiconductor products.
Owners opinion
All generations of Intel processors listed in this material have a high degree of energy efficiency and a phenomenal level of performance. Their only drawback is high price. But the reason here lies in the fact that the direct competitor of Intel, represented by AMD, cannot oppose it with more or less worthwhile solutions. Therefore, Intel, based on its own considerations, sets the price tag for its products.
Results
In this article, generations of Intel processors for desktop PCs were considered in detail. Even this list is enough to get lost in the designations and names. Other than that, there are also options for PC enthusiasts (platform 2011) and various mobile sockets. All this is done only so that the end user can choose the most optimal one for solving their problems. Well, the most relevant now of the options considered are the 6th generation chips. It is on them that you need to pay attention when buying or assembling a new PC.
In August 2015, the 6th generation of computing chips from Intel - Skylake was introduced. The processor belonging to this generation received a significantly redesigned architecture, which allowed for a 10-15% increase in performance compared to the previous generation CPU, codenamed Haswell. It is about their technical parameters, capabilities and types that will be discussed further.
Background of appearance
At the moment, every 2 years, Intel updates processor sockets. So, in 2013, the LGA1150 was released along with the Haswell line of CPUs. This is the 4th generation CPU based on the Core architecture. Then, a year later, Haswell chips were replaced by Broadwell. This is already the 5th generation of the CPU of the Core architecture. Their key difference is the updated technological process, which is 14 nm. But the processor part has not changed. Then, in 2015, the 4th and 5th families of chips based on the Intel Core architecture were replaced by the 6th, which was codenamed Skylake. The processor of any model of this generation is produced according to a similar technological process- 14 nm (like Broadwell or the 5th generation of the Core architecture). But at the same time, the architecture of the computing part was redesigned, and this made it possible to obtain a certain performance increase of 10-15%. Also, the power supply subsystem of the semiconductor crystal has been redesigned. The CPU voltage regulators are now on the motherboard. Such an engineering approach made it possible to keep the power subsystem practically unchanged, but at the same time improved the overclocking potential of the central processor.
Socket and chipsets
It is the LGA1151 socket that is designed to install any desktop chip of the Skylake family. The new generation processor in this case is designed to be installed in a new socket and is not compatible with the previous generation CPU. Also, a new generation of system logic sets was released to support the new generation of central processors. The most modest among them from the position of a functional set is H110, users note. But at the same time, it has a corresponding cost. It is perfect for budget, entry-level systems. The most functional and most expensive set of logic in this case is the Z170. Its key difference from all other chipsets is the ability to overclock a CPU with an unlocked multiplier (it is aimed at installing such CPUs), an integrated graphics accelerator, and even random access memory. This is an excellent solution for creating the most productive PCs. The remaining options H170, B170, Q150 and Q170 are intermediate between the two previously given sets of system logic, and their main purpose is to build a PC of an average price level and exactly the same speed.
Technical features
As noted earlier, the Skylake processor core has been significantly redesigned, and due to this, an additional increase in performance has been obtained. But most of it has not changed significantly. This is the first level of the cache. Its total size for one block is 64kb, which are divided into 2 parts of 32kb for data and instructions. The second level no longer has such a division, and its volume is 256 kb. The third cache level is common to all CPU computing resources, and its size depends on the specific model: from 2 MB for Celeron processors to 8 MB for i7. The manufacturing process, as noted earlier, has not changed in comparison with its predecessors - 14 nm. chipset, as in previous generations of processors, is part of its semiconductor crystal. That is, the CPU, in addition to the computing part and the graphics accelerator, also includes a PCI-Express controller and a dual-channel RAM controller. The latter can already work with DDR4.
Entry Level Solutions
The entry-level Skylake is made up of chips from the Celeron and Pentium lineups. Physically and programmatically, these semiconductor chips have only 2 computing modules and the same number of data processing threads. The first of them have the most affordable cost, but at the same time their performance is much lower. A higher level of performance of the Pentium line of chips is provided by larger and larger level 3 caches. Also in the latter case, a more productive HD Graphics subsystem with an index of 530 is used, while Celeron is equipped only with a solution with the designation 510. An exception in this regard is the Pentium G4400 with a shortened version of the integrated 510 video card. The Celeron G3900T model with a thermal package of only 35W and a reduced clock frequency of 2.6 GHz. Otherwise, more detailed specifications of the 6th generation Celeron and Pentium processors are given in the table.
Processor Model and Index | Level 3 cache, Mb | Fixed chip frequency, GHz | Number of chip cores / threads | Thermal package, W | Video card model HD Graphics |
||
Middle segment
In the middle segment, this generation of CPUs is represented by Core i3 processors. In total, 6 chips belong to this niche at the moment. All of them include 2 physical computing units and 4 software threads. That is, in these processors there is support for proprietary technology from Intel, which calls HyperTrading.
It is this feature that makes it possible to double the number of information processing threads at the program level. But in this case there is no talk of supporting TurboBoost technology, and the processor frequency is fixed. Two representatives of this family with indices 6100T and 6300T have reduced clock frequencies and a reduced thermal package of 35 watts. These are energy efficient solutions aimed at creating compact computer systems. One chip marked 6098P is equipped with a less productive graphics system HD Graphics with an index of 510. All processors of the 60XX and 61XX series have 3 MB of L3 cache, and the 63XX series have 4 MB. The integrated video accelerator in all other cases has an index of 530. More detailed characteristics of all sixth generation i3 processors are indicated in the table below.
Processor name | Third level cache, MB | Processor clock frequency, GHz | Number of real cores/software threads | Thermal package values, W | Cost, USD | HD Graphics accelerator model |
|
The most productive quad-core solutions
The most popular semiconductor solution in this case is the Intel Core i5 processor. Skylake architecture in this case is represented by 9 chip models at once. All of them have 4 computing cores. Two models with indexes 6685R and 6585R have an improved graphics subsystem HD Graphics model 580, one, 6402P, less productive - 510. Three chips 6400T, 6500T and 6600T are energy-efficient solutions with reduced frequencies and a reduced thermal package. The remaining processors 6400, 6500 and 6600 are standard representatives of this line of devices. More detailed technical specifications of the i5 CPU of this generation are given in the table.
Marking | Level 3 cache, Mb | Frequency range min/max, GHz | Number of physical cores/processing threads | Thermal package value, W | Current price, USD | HD Graphics Video Accelerator |
|
Eight-thread chips with maximum performance
Any Core Skylake belonging to the i7 line has a full set of various technologies (HyperTrading and TurboBoost). It can process data in 8 threads and dynamically change its frequency.
In terms of performance, these processors lose only to the most expensive solutions for computer enthusiasts, in which the frequency multiplier is unlocked, and due to this, you can get a significant increase in performance. Currently, this line includes only 3 chips, and their specifications are shown in the table below. One of the models has the index 6700T, and it is an energy-efficient CPU for building high-performance compact systems. The second is the 6785R. It is equipped with an improved graphics accelerator model with an index of 580. And the last one, 6700, is a typical flagship with a locked multiplier and maximum performance (except for chips for enthusiasts).
CPU designation | Cache 3 levels, Mb | Frequency formula min/max, GHz | Number of cores / processing threads | Claimed thermal package, W | Declared value, USD | HD Graphics video adapter |
|
Products for PC Enthusiasts
As with previous generations of Core processors, only 2 chip models have an unlocked multiplier. The first one is 6600K. This is a typical i5 quad-core processor. Skylake architecture has excellent overclocking potential. With a high-quality cooling system, its frequency can be increased without any problems from 3.9 GHz to 4.6-4.7 GHz by simply raising the multiplier. If you also change the voltage on the semiconductor chip of the processor, you can even get 5.0 - 5.1 GHz.
The second representative of this family is 6700K, which already belongs to the i7 line. It has parameters identical to all other chips of this model range. The key difference that experts note is the unlocked multiplier. Well, the frequencies that can be obtained during overclocking are similar to 6600K. Their technical specifications are shown in Table 5.
Reviews. Results
Users claim that Skylake CPUs have become a worthy continuation of previous generations of chips. The processor of this family, in their opinion, has improved both in terms of speed and energy efficiency.
The life cycle of this platform is only just beginning, and according to Intel, it will still be relevant for the next 3 years. So it's time to buy a new high-performance and energy-efficient personal computer.
