Platform Intel Kaby Lake will be released only in the first quarter of 2017, but today Intel revealed details of the seventh generation of Core processors, and announced several models. We received this information even 14 days ago at IDF, but we can publish it only today. In addition, some additional information leaked in the form of rumors, but so far there was no official confirmation. In particular, we published the news about ten desktop processor line 7000, eight Xeon E3-1200-v6 and the Intel 200 chipset, which the new motherboard will be released.
The current view of the seventh generation of the Core is that affects mobile versions of Y lines and U. The rest of Intel processors will present later this year. However, the new features, which we will discuss in this review, supported and desktop processors. Processors codenamed Kaby Lake Y will form the basis for models of Core M ultrathin devices and Kaby Lake U targeted at high-performance laptops.
Intel will continue to use all the processors Kaby Lake 14-nm process technology, although the company was able to increase efficiency, which is very important for the mobile segment. So in front of us, in fact, a logical evolution. But proceed directly to processors.
Intel has traditionally shown the seventh generation Core chip shot that will be used in the lines of Y and U. From above we can see not only the memory controller and I / O controller and System Agent blocks and Wireless Display. Below are two physical cores with support for Hyper-Threading, the center is visible cache L3. The latter (as before) is common to the two nuclei. The largest part is occupied by a picture box with numerous new features. On the right you can see the memory interfaces and I / O subsystem.
The photo shows the processor Kaby Lake Y. Namely, two crystals, the processor and chipset, which are located in the SoC PCB with dimensions of 20 x 16,5 mm. It changes happened over the previous generation.
Intel still uses the processor socket Kaby Lake Y BGA1515.
An important advantage of processor Y is very low altitude packaging. These processors are ideal for thin portable devices such as tablets, laptops or hybrid computing modules in the key fob format, where the new Core M processors will find their application.
Dimensions U CPU is much larger, 42 x 24 mm. SoC again consists of CPU and chipset, but they are separated by the PCB. Clearly visible thin trunks.
Compared to processors Skylake U socket has not changed. Processors Kaby Lake U is still installed in the socket BGA1365.
Expected processors Kaby Lake H will be installed in a socket BGA1440, they will be equipped with four cores have enabled Hyper-Threading.
Processors Kaby Lake Y (Core M) will work with a TDP of only 4.5 Watt. For Kaby Lake U Set higher package 15 W, but, in turn, the processors have higher clock frequencies, and therefore higher productivity. Seventh generation Intel compares the first pointing efficiency gains Kaby Lake U 10 times. But even compared to a generation Skylake Intel managed to optimize the 14-nm process technology and improve efficiency.
In particular, the Intel engineers were able to make the transistors more reliable, and optimized the structure. According to Intel, it managed to get a performance boost of up to 12%. Intel calls the new architecture “Architecture + 14nm +”. Name Taken from the new process technology 14+, who devoted his report to Mark Bohr (Mark Bohr) in the IDF.
To compare the performance of Intel Core i7-6500U processor took the sixth-generation Core (Skylake U) against Core i7-7500U upcoming seventh generation Core (Kaby Lake U). Both processors are based on the two physical cores and technology Hyper-Threading, meaning they can carry up to four threads simultaneously. According to the test SYSmark 2014 productivity growth was 12%. According to Core i7-7500U web-performance showed growth of 19% (measured using WebXPRT 2015).
However, increased productivity, for the most part, not associated with improvements architecture, and a difference frequency for effective Turbo. Core i7-6500U processor clocked at up to 3.1 GHz Core i7-7500U in the case of rate increases to 3.5 GHz.
If you plan to upgrade your PC in the next year, Intel gives a rough estimate of performance in comparison with the computer age of five years. Intel used to compare processors Core i5-7200U (Kaby Lake) and Core i5-2467M (Sandy Bridge). Chip giant points to 70% of mobile productivity growth. That is, we get into working applications 1.7 times faster performance. As for creating, editing and distributing video 4K UDH 360, here we have an increase of 8.6 times. Also it highlights and smooth process. As for the game, then Overwatch example from Blizzard we get a three-time increase in performance on processors Kaby Lake through faster graphics core.
On conversion of 4K video lasting an hour it took only 12 minutes. According to Intel, Sandy Bridge processor is required 6.8 times more time.
According to the slides Intel presentation, future processors Kaby Lake so effectively working with 4K video, that the battery will last for much longer battery life. Although it all depends on the capacity. “Premium Content” HEVC codec 10-bit can be seen from the battery for 9.5 hours. As an Intel-based Core i7-7500U processor takes and Core i7-6500U, a battery of 66 Wh and 4K panel. Why did it happen – we will show below, when we consider the GPU changes.
Quite interesting data are given for 4K and 4K video on YouTube with 360 codec VP9. Battery 66 Wh with a processor Core i7-6500U enough for four hours viewing, a new generation of battery life is 75% longer, and Core i7-7500U give seven hours. Here, too, the reason lies in improving the graphics unit.
Most Skylake graphics processors running Gen9 unit, which provides a good level of performance even for iGPU. For “Kaby Lake” Intel took as the basis of GPU Gen9 with 24 execution units, adding two hardware functions. Namely Multi-Format Codec (MFX) and Video Quality Engine (VQE). Adding the two functions has improved graphics unit performance as a whole, together with the power of the engines working in parallel was sufficient to cope with 4K video.
Multi-Format Codec (MFX) capable of decoding 10-bit HEVC and 8/10-bit VP9, and encoding 10-bit and 8-HEVC bit VP9. It is this block is the main reason for the improved performance of coding and decoding video 4K. In block Video Quality Engine (VQE) implemented the conversion of high dynamic range (HDR) in the standard range of colors with the change (Tone Mapping), also supported by a wide color gamut (Wide Color Gamut).
The improvements described above, lead to an increase in performance 4K. Graphic unit supports simultaneous playback of up to eight streams of 4K with a frame rate of 30 fps (for example, a wall of video monitors in the room). It is also possible decoding HEVC 4K video at 60 fps, and flow up to 120 Mbit / s. Recoding AVC-in-AVC on Kaby Lake Y processor runs twice as fast, and processors Kaby Lake U – three times faster than real time. Recoding AVC-in-HEVC on Kaby Lake Y is performed in real time, and Kaby Lake U – two times faster than real time.
The slide shows the differences in encoding and decoding between “Skylake” and “Kaby Lake”. The two-generation Core architecture supports encoding / 1080p HEVC decoding. But only “Kaby Lake” we get support for encoding and decoding 4K (2160p) HEVC 10-bit. Also in the list of functions of the seventh generation of Core and added VP9 decoding. Encoding VP9 Intel format “Kaby Lake” does not indicate, however, most likely, there is no hardware acceleration.
The trend continues to popularize 4K format. Google has moved to VP9 format for rendering on YouTube HTML 5 its platform, because it provides a higher compression ratio compared to other codecs, requiring twice less bandwidth for the former video quality. All this reduces the delay for buffering, the user can enjoy the video more often (U) HD without a pause for buffering. However, everything still depends on the quality of your Internet connection.
In April last year, tech blog YouTube, it was stated that in 2014 users watched 25 billion. Hours of HD-video codec vp9. Without codec VP9 this figure would have been impossible.
It is interesting to compare the load on the CPU and the power consumption of processors Core i7-6500U (Skylake) and Core i7-7500U (Kaby Lake) 4K video playback:
The first comparison is performed during playback 4K-video codec HEVC 10-bit. The load on the CPU Skylake when decoding video 4K was 50-70%. Kaby Lake Processor was much more efficient, with a load of only about five per cent. Of course, this difference leads to a substantial difference in power consumption. Intel imposed a 30-second two graphics processors, as well as resulted in average values.
The processor Skylake we have significant peaks of energy consumption from less than 5 watts to about 25 watts, which leads to an average level of 10.2 watts. At Kaby Lake skylake power consumption is significantly lower. The graph is not reached even 1 watt. The average is only 0.5 watts.
Therefore, the previous generation to the next change has led to an increase in performance 4K playback (CPU + iGPU) up to 20 times, SoC performance increase we have received up to 6 times the battery life – up to 2.6 times.
Intel evaluated the effectiveness of 4K video decoding codec on VP9 Google YouTube platform through corporate Chrome browser. Both generations of processors, according to the schedule, we got a heavy load on the CPU compared to the scenario above. At Core i7-6500U load was between 70 and 80%. The new generation of Intel it significantly below 15%.
The difference in power consumption is also very noticeable. Do we get a Core i7-6500U scatter powers up to 18 watts. The average level of 5.8 watts. The processor Core i7-7500U power consumption is much lower. We get a level of 0 to 2 W, an average of 0.8 watts – an excellent result.
In this test, Intel indicates an increase in performance of the new generation Core to 7 times (CPU + iGPU), an increase SoC performance up to 3 times, and increase battery life by 75%, as we have said above.
Our readers will remember Speed Shift technology. It was realized as early as Skylake processors, we just tested the Core i7-6700K, but Intel did not provide details then. The technology works in close cooperation with the function of Turbo Boost 2.0, it can accelerate the transition to a state of high frequency (Turbo), respectively, the processor will soon return to the idle state after the task.
For comparison again using Core i7-6500U and Core i7-7500U. Intel has tested Skylake model on and off the function Speed Shift. Without this technology Core i7-6500U reached a maximum frequency of 3.1 GHz after about 100 ms. The value does not seem so great, but look at the acceleration of the active function Speed Shift. Here, the same rate was achieved after 30 ms load.
At the Core i7-7500U processor the Intel with a maximum frequency of 3.5 GHz, we have not only a higher frequency of Turbo, but also more rapid transition to this rate – 15 ms. So optimization clearly traced.
From the outset it was clear that the processors Kaby Lake will not be a revolution on the CPU market, rather they can be regarded as a further evolution. As each new generation of Core, Intel added a small performance boost IPC (the number of instructions per clock), but significantly improved integrated graphics core Gen9, which proved to be the focus of attention Kaby Lake models. New hardware blocks Multi-Format Codec (MFX) and Video Quality Engine (VQE) with support HEVC-8/10-bit and VP9 allowed Intel to significantly increase coding performance and video decode 4K codec HEVC 10-bit, as well as increase the performance of the codec decoding vP9 (YouTube).
Moreover, thanks to the support of hardware decoding and encoding 4K could significantly reduce the load on CPU and, therefore, power consumption of processors. So mobile platforms based on Kaby Lake Y and Kaby Lake U can give a significantly higher battery life under these loads. Intel also finalized the mechanism of Turbo Boost 2.0, expanding its function Speed Shift, accelerating time to switch to the maximum Turbo frequency under load, which makes it easier to perform a task and go back to idle mode. In general, all these are quite important technology for mobile segment. In the case of desktop processors it seems interesting 14+ new process technology, which should provide higher overclocking frequency.
Many potential buyers disappoint the presence of the former restrictions on the four physical cores for mass Intel platform. Still, Intel reserves the prerogative of a larger number of cores on a platform for enthusiasts LGA2011-3 socket and chipset X99 (option Z97 with an increased number of SATA ports). Recently, Intel just introduced new processors for the platform: Core i7-6800K, Core i7-6850K, Core i7-6900K and Core i7-6950X, which are equipped with six-ten cores with support for Hyper-Threading, but Broadwell old architecture, let and 14-nm process technology. Update on this platform Kaby Lake will happen much later, even after the release of 4-core processors for desktop PCs by the end of the year.
Overall, Intel says about 10-15% performance boost with Kaby Lake, as we have repeatedly seen with updates CPU architectures in recent years.
AMD has high reliable at Zen processors, which will be released in early 2017. They will be installed in the new Socket AM4 and work with DDR4 memory. In the first tests, 8-core AMD Zen to support SMT (Simultaneous Multithreading) at the same frequency could compete with the Core i7-6900K (Broadwell-E), which was also equipped with eight cores with SMT support in the form of branded Intel Hyper-Threading technology, enabling simultaneously perform up to 16 streams. Perhaps, AMD processors increase pressure on the Intel, and a chip giant will accelerate, and consumers will gain greater than 10-15% per year.
In any case, the next six months on the market of processors are expected to be very hot. In our test lab has already received the first laptops to Kaby Lake, soon we will post tests.