[ad_1]
The year 2017 is coming to an end and it’s time to remember everything that has happened over the past 12 months in the world of computer hardware. The most discussed topic of the outgoing year, of course, is the return of competition to the processor market. After several years of hard work, AMD introduced the Zen microarchitecture to the world, which became the basis for a large number of new products. The next stage of the “race of cores”, which began with the release of the eight-core “stones” Ryzen 7, forced Intel to rush to prepare the eighth generation of Core processors, while increasing the number of cores, both in mobile and desktop solutions. Advanced Micro Devices even dared to “encroach on the sacred” – the market for HEDT systems, which Intel had a hand in creating.
Unfortunately, it will not be possible to mention all the novelties in one article, so it was decided to divide it into three parts, while the author takes responsibility for choosing the most interesting and important topics. As always, you can leave your comments and suggestions in the comments. Well, let’s start summing up the “iron” results of 2017 from the cornerstone of modern PCs – central processors.
The Second Coming of Skylake
The outgoing year in the hardware world began with the release of the seventh generation Intel Core desktop processors belonging to the Kaby Lake-S family. The new microarchitecture, which is a slightly modified version of Skylake, quite expectedly did not bring a significant increase in performance, and the performance increase relative to the solutions of the previous generation was obtained in an extensive way – by increasing the operating frequency. Tests in our laboratory have shown that the specific performance of products from the Kaby Lake-S and Skylake-S families at the same frequencies is almost identical.
The obvious advantages of Kaby Lake-S desktop processors include the already familiar LGA1151 design, thanks to which motherboards based on Intel 100-series chipsets have acquired their support, as well as the use of an improved 14-nm process technology, which made it possible to increase working hours by 200–300 MHz. frequencies, while keeping power consumption at the same level.
Pentium G4560 is the best budget CPU of the first half of 2017. Photo by ComputerBase
Strange as it may sound, the most memorable representatives of the new CPU family are Pentium solutions, which received support for Hyper-Threading multi-threaded data processing technology, which made them one of the most attractive processors for budget systems. Also noteworthy is the very controversial overclocking chip Core i3-7350K with an unlocked multiplier, which, despite having only two physical cores with multithreading support, was sold at the price of the junior quad-core CPUs of the Core i5 line.
Together with the 7th generation Core central processors, new motherboards using Intel 200-series chipsets went on sale.
ASUS Maximus 9 Hero
Compared to their predecessors, the new line of chipsets can provide more PCI Express 3.0 lanes, and also support Intel Optane Memory devices with 3D XPoint memory (which we’ll talk about a little later), designed to speed up hard drives by early caching the most frequently used data. For comparison, the flagship Intel Z270 chipset allows you to use up to 24 PCI-E 3.0 lanes, while its predecessor (Z170) has 20 lanes.
Back in Red
While Intel has been systematically increasing the performance of its central processors, not forgetting the reduction of technological processes, the engineers at Advanced Micro Devices have been busy developing a new microarchitecture code-named Zen in recent years. As a result, it became the basis for a number of products from the Sunnyvale chipmaker, the first of which are AMD Ryzen desktop chips for the Socket AM4 platform.
AMD Ryzen 7 1800X (left) and FX-6100
The debut processors of the Summit Ridge family, namely the eight-core models of the Ryzen 7 line, appeared on store shelves in early March, announcing the start of a new stage of the “core race” in the desktop CPU segment. We can say with full confidence that the new generation of processors turned out to be very successful and made a worthy competitor to Intel solutions available at that time. Despite the frequency advantage of Kaby Lake-S chips, AMD Ryzen processors showed significant superiority in applied multi-threaded tasks along with acceptable gaming performance.
On the other hand, it cannot be said that the release of AMD Ryzen processors went smoothly and without any problems or childhood illnesses. The first 14-nanometer CPUs of the “reds” cannot boast of outstanding overclocking potential, and overclocking these “stones” to frequencies above 4 GHz is not an easy task even when using liquid cooling systems. In addition, the cranky dual-channel DDR4 memory controller in the first months after the debut of Ryzen processors caused a lot of headaches for pioneers who want to achieve high-speed RAM. It is fair to say that over time, the chipmaker has significantly expanded the list of RAM sets supported by new CPUs by releasing improved UEFI firmware.
As for the AMD AM4 platform itself, its formal debut took place last year along with Bristol Ridge hybrid processors, but at that time they were available exclusively to OEMs.
The life cycle of this socket, as well as the Zen microarchitecture itself, according to company representatives, will be about four years, and existing motherboards will most likely acquire support for more than one family of central or hybrid processors. However, it is too early to talk about such compatibility today.
Race of cores
The return of competition to the consumer processor segment pushed Intel to significantly accelerate the pace of development, and nine months after the release of Kaby Lake-S solutions, the first Coffee Lake-S products saw the light. In these CPUs, Santa Clara chipmaker engineers decided to increase the number of physical cores by one and a half times, thanks to which for the first time the mainstream Intel desktop platform was led by six-core chips capable of processing up to 12 threads simultaneously. Recall that for more than ten years, the top models of Intel processors for mass platforms contained only four cores.
In terms of microarchitecture, Coffee Lake-S solutions are practically the same as their predecessors, which is why the specific performance per core has remained unchanged. However, the overall level of computing power due to the presence of additional cores has grown by 25-50%, depending on the model. A more advanced 14-nm process technology made it possible to keep the nominal heat pack at a relatively low level, while the frequency potential of new processors did not suffer much from an increase in the number of cores.
Along with the new processors, motherboards made their debut using the hastily made Intel Z370 logic set. This chipset, despite belonging to the logic of the 300 series, differs from its predecessor only in the “innate” support for the new CPU family.
Not without a “fly in the ointment”. Despite the already familiar LGA1151 design, Coffee Lake-S products are not compatible with existing motherboards based on Intel 100 and 200 series logic, and Skylake-S and Kaby Lake-S solutions are not supported by the new boards. The reason for this, as the company representatives assured, was the need to rework the power supply scheme for new processors. The voltage converter on motherboards with Intel Z370 logic includes at least four phases (excluding doublers), and a larger number of processor socket pads are used to supply power to the CPU itself.
Battle for HEDT
While enthusiasts were talking about Ryzen 7 with might and main, predicting the imminent end of the monopoly in the consumer processor market, AMD was preparing to launch an attack on Intel’s position in the segment of high-end desktop systems of the HEDT (High-End Desktop) class. The high flexibility of the Zen microarchitecture allowed the chipmaker’s engineers to quickly prepare the Socket TR4 platform along with the Ryzen Threadripper processors.
For the new high-end chips, AMD used the available eight-core Zeppelin crystals, familiar from regular AMD Ryzen processors and server EPYCs. The Ryzen Threadripper’s heat-spreading cover contains four semiconductor chips, two of which are physically disabled. The remaining two operable dies are connected via the Infinity Fabric bus and, roughly speaking, are a two-processor system using one 4094-pin socket.
Scalped AMD Ryzen Threadripper 1950X. Der8auer Photos
As a result, AMD Ryzen Threadripper processors can offer up to 16 multi-threaded physical cores, up to 32 MB of L3 cache and a quad-channel DDR4 memory controller. They also include a PCI Express 3.0 interface controller for 64 lines, of which four pieces are used to communicate with the chipset, and the remaining 60 can be used to connect video cards, NVMe drives and other devices.
One of the main advantages of the Ryzen Treadripper chips, quite expectedly, was the high ratio of raw computing power to price. At the same recommended price as the 10-core Intel Core i9-7900X, the Ryzen Threadripper 1950X (16 cores / 32 threads) in professional applications optimized for multi-threaded computing is often noticeably faster than its rival.
In parallel with the new CPUs, compatible motherboards debuted using the tandem of the Socket TR4 processor socket and the AMD X399 system logic set. In terms of functionality, the new chipset is a complete copy of the X370 logic designed for the Socket AM4 platform and provides PCI-E 2.0 lanes, USB and SATA ports.
The socket itself deserves special attention. Ryzen Threadripper processors use LGA and are very large, which negatively affected compatibility with existing cooling systems.
Noctua coolers for Socket TR4 platform
In order to “cool off the heat” of one of these chips, you will need to get a cooler specially designed for it or use Asetek’s maintenance-free CBO.
Unlike the unexpected decisions of AMD Ryzen Threadripper, the upcoming update of the Intel HEDT platform has been known for a long time. In early summer, the LGA2011-3 processor socket was replaced by the LGA2066, along with Skylake-X and Kaby Lake-X chips, as well as the Intel X299 chipset.
The release of processors for the new platform took place in three stages: in June, solutions with the number of cores from 4 to 10 appeared on store shelves, in early September, the release of the 12-core Core i9-7920X took place, and at the end of the month, the flagship 14-, 16 – and 18-core “stones” Core i9-7940X, Core i9-7960X and Core i9-7980XE.
By the way, the release of older solutions was a surprise not only for computer enthusiasts, but also for motherboard manufacturers. For this reason, some vendors this fall released updated versions of boards with a reinforced processor socket power system to cope with overclocked 18-core CPUs.
ASRock X299 Taichi (left) and X299 Taichi XE: Find 10 Differences
Compared to Broadwell-E solutions, in the new chips, Intel engineers significantly redrawn their internal structure, abandoning the use of a ring bus in favor of a mesh network, implemented support for AVX-512 instructions, and also increased the amount of second-level cache memory for by reducing the capacity of the shared L3 cache. At the same time, the number of PCI Express 3.0 interface lines in the top Skylake-X processors has grown by only four.
Kaby Lake-X solutions, positioned by the chip maker as an “entrance ticket” to the world of HEDT, also deserve attention. With the exception of slightly increased operating frequencies, the Core i5-7640X (recommended price $242) and Core i7-7740X ($339) processors are complete copies of older solutions for the Intel Z270 platform. Both chips are equipped with only four physical cores, 16 PCI-E 3.0 lanes, and a dual-channel DDR4 memory controller, while Skylake-X chips interact with memory in a four-channel mode. Because of this, when assembling a HEDT system on the Intel LGA2066 platform, it is advisable to consider buying only Skylake-X solutions, especially since the younger “six-core” Core i7-7800X will cost only $50 more than the Core i7-7740X.
Summing up the struggle between Intel and AMD for the market of high-performance HEDT systems, we can safely say that the latter managed to shake the position of the “blue giant”, which is the ancestor of this segment of personal computers. Ryzen Threadripper processors can offer not only an attractive ratio of “raw” processing power and price, but also a greater number of PCI Express 3.0 lanes, for which the CPU is responsible for the operation. On the other hand, with the release of older Skylake-X solutions, Intel showed who is in charge in the house, offering enthusiasts and professionals the most powerful processors for desktop systems, albeit not distinguished by a humane price tag.
Meanwhile in the mobile market
While AMD was still preparing to release Ryzen Mobile APUs, Intel decided to take the initiative and in August introduced the first 8th generation Core mobile solutions belonging to the Kaby Lake Refresh family. In the new CPUs, the chipmaker’s engineers managed to double the number of physical cores, while maintaining their nominal heat pack at 15 watts.
Despite twice as many cores, Intel Kaby Lake Refresh processors are on average only 40% faster than their predecessors, due to lower nominal frequencies. It is fair to say that along the way, the Turbo Boost “ceiling” was increased by a couple of hundred megahertz, due to which the new chips are faster than the energy-efficient Core 7th generation solutions in a single-threaded load.
On the other hand, the performance of the graphics subsystem of the new processors has remained virtually unchanged. The integrated video core in Kaby Lake Refresh chips is called UHD Graphics 620 and differs from HD Graphics 620 used in previous generation solutions, mainly by 100 MHz (1150 vs. 1050 MHz) auto-overclocking frequency. True, it all depends on the specific processor model.
After the return of competition in the desktop CPU segment, Advanced Micro Devices decided to take on the mobile market with Ryzen Mobile APUs. At the end of October, the chipmaker introduced the first solutions of the Raven Ridge family, combining four Zen physical cores with multithreading support and a Vega-based graphics subsystem that can compete with entry-level discrete video adapters.
Thanks to the transition to Zen microarchitecture, AMD Ryzen Mobile APUs are only slightly behind Intel Kaby Lake-R in terms of x86-64 core performance, and the powerful integrated video core can compete with discrete GPUs used in ultrabooks. And the engineers of the “red” chipmaker managed to place all this in a small heat pack, which (depending on the device) is from 12 to 25 watts.
Another indisputable advantage of AMD Ryzen Mobile chips is their price: laptops with these APUs are often cheaper than counterparts with quad-core products from Intel.
One of the most discussed topics in the first half of November was the formal announcement of the upcoming Intel Core H-series processors equipped with AMD Radeon discrete graphics.
On a single substrate, Santa Clara-based chipmaker engineers managed to place a CPU die, a discrete GPU developed by specialists from the Radeon Technologies Group, and one HBM2 video memory stack.
As planned by Intel, this multi-chip module should become a new step in the evolution of mobile processors and will be used in thin laptops or mini-PCs that traditionally do not have a high-performance video subsystem.
Unfortunately, the “blue giant” did not disclose the technical characteristics of new products, and it is not known for certain which architecture the CPU and GPU will use as part of the above-described “hybrid”. However, the detailed specification of the eighth generation Intel Core H processors, according to the information available on the Web, will be published in the first quarter of the coming year.
On this note, we will finish the first part of our material and in the near future we will continue summing up the results of 2017. Stay with us!
Overview of the main events of 2017. Video cards
Overview of the main events of 2017. Memory, overclocking and new trends
[ad_2]