Comparison of processors for laptops (AMD and Intel). AMD Mobile Processors Best AMD Processor on Kaveri Architecture

While choosing a processor from AMD, you encounter many incomprehensible letters and numbers. What do they mean? How to divide the average processor from weak? You will learn about it in our material.

Introduction

It will not be considered processors until 2010 release, also server solutions, chips on the AM1 platform, as well as the AMD Ontario line (currently not relevant), so the marking shown in this article may not approach them.

Here is a roller who will help you figure out, but we recommend to read the article, as it will be updated in the future in the future.

Architecture

At the market, the chips of the 4th last desktop architectures are currently represented, and in the second half of 2016 it is planned to present the world a new Zen architecture with a high jump performance on the tact and reduced to 14 nm, which may help to catch up intel in the top segment.

Sockets

Updated platforms at the beginning of 2016 include FM2, FM2 + and AM3 +

Processor Rulers

E - Series

Budget processors of the initial level intended for laptops and netbooks.

E1 have 2 kernels on board, and E2 - 4.

Belonging to a certain generation is determined by the first digit:

• 7- Carrizo-L
• 6 - Beema.
• 2, 3 - Kabini (without considering old chips until 2012, in which the same figure)

Chips of this series are quite small and if there is no need to familiarize yourself with the models of software.

APU

AMD processors with built-in graphics core (APU) are divided into rules:

• A4 - 2 kernels
• A6 - 2 kernels
• A8 - 4 kernels
• A10 - 4 kernels

A12-8800B falls out of this nomenclature, but you can read about it.

Accordingly, from weaker to more powerful, both in the graph and the processor part. Here is an example:

The first digit indicates the processor core (generation).

In our case, having a number 7, we get Kaveri kernels.

It is worth noting that the number 4 in the A4 series on the richland architecture means a reduced frequency, which leads to a decrease in performance.

850 - Indicates productivity among similar frequency processors (more - better)

• P - Standard power consumption in the case of mobile processors (35 W)
• B - Pro processor designation
• M - Mobile Processor (Old Designation)
• K - unlocked for overclocking
• T - reduced power consumption (stationary PC)

Interestingly, there are A-processors marked by the FX trademark. As a rule, these are the most powerful laptop processors of the company. They are also built on APU architecture.

Athlon

Now comprise Athlon. In essence, these are the same A - processors, but with a disconnected video engineer for a lower price.

As an example, take

• X4 - denotes 4 processor kernels
• 8 - is the Kaveri nuclei pointer (7 - Trinity)

Indicate on earlier models, we do not see the point, since even the top for this society, the Athlon X4 860K chip demonstrates the results of the average for modern chaiparations, so you do not advise you to take these processors in 2016. If at first it will arrange you, then in the upgrade you will have to change the motherboard, which will fly into a penny and will take money saved on this decision.

• 60 - as well as in the previous case indicates the position of the processor in the line
• K - has the same value

FX.

Now let's talk about the fastest AMD processors - the FX series. These chips have a large overclocking potential and a very democratic price tag. The main drawback flows out of enough outdated architecture and production technology - power consumption. TDP ratio - performance strongly loses Intel processors, but the price is performance on a very good level. The nomenclature set out below is not valid for FX 9xxx - these are the same 8xxx, but with an increased clock frequency. Here is the chip that we chose as an example:

The first digit indicates the number of cores, in this case 8.

The second indicates a generation

• 3 - Vishera kernels
• 1, 2 - Zambezi kernels

The remaining figures indicate the frequency of the chip within the framework of a single family, but we believe that it does not matter. We advise you to take the youngest model in the line, since the older is exactly the same, but with factory overclocking. And why overpay for factory overclocking, if "stones" and so well?

If you can visit the site, you can find some useful information there.

This article did not provide information on older chips, as well as on server solutions due to obsolete technology (technical process, architecture) in the first and specificity of application and high costs in the second. We hope that our material helped you understand AMD processor nomenclature and help you decide on the choice.

This article presents only the best AMD processors in 2017.

If you do not want to independently understand all the characteristics of each processor model or are not sure that you can choose the best option, pay attention to our CPU rating from AMD.

Building:

A good processor is the main power indicator and. AMD is one of the leaders of the processor market.

AMD produces the following types of processors:

• CPU. - Central computing units
• GPU - A separate device performing rendering video. Often used in gaming computers to reduce the load on the central unit and to ensure better quality of the video.
• APU - Central processors with built-in video accelerator. They are also called hybrid, because such a component is the associations of central and in one crystal.

№5 - Athlon X4 860K

AMD Athlon line is designed for socket Socket FM2 +. X4 860K is the best and most productive model from the entire series, in which three processors go:

• Athlon X4 860K;
• Athlon X4840;
• and model Athlon X2.

The Athlon family is designed for desktop personal computers. All model models are characterized by good multithreading.

The best results in the Athlon group showed the X4 860K model.

The first part that should be noted is support practically that consumes no more than 95 W along with quiet work and without loss in performance.

If the processor has been dispersed with special programs, noise increases in the cooling system.

Main characteristics:

• Family: Athlon X4;
• Number of processor nuclei: 4;
• Clock frequency - 3.1 MHz;
• There is no unlocked multiplier;
• Type of kernel: kaveri;
• Approximate cost: $ 50.

Integrated graphics in the CPU is missing.

The X4 860K processor is able to maintain fast operation of General-Purpose systems.

Testing the work of the CPU was carried out using the AIDA64 utility. In general, the model shows good results for the middle class processor.

If you are looking for an inexpensive CPU with multitask support for your home computer, Athlon X4 860K is one of the appropriate options.

testing Athlon X4 860K

№4 - AMD FX-6300

FX-6300 from AMD is a CPU with Piledriver architecture support. Processors with such architecture have already become worthy of competitors of innovations from Intel.

All processors from the AMD group FX have excellent acceleration potential.

Features FX-6300:

• Series: FX-Series;
• Supported connector: Socket AM3 +;
• Number of nuclei: 6;
• No integrated graphics;
• Clock frequency is 3.5 MHz;
• Number of contacts: 938;
• The cost of the model on average is $ 85.

The characteristic feature of the processor consists in its flexibility.

The clocking frequency declared by the developer is 3.5 MHz, which is quite a mediocre indicator among.

However, this CPU provides for the possibility of overclocking frequency to 4.1 MHz.

boxing devices FX series from AMD

Acceleration of work occurs during intensive loads. More often in the process of rendering video or working with games.

It should be noted that this CP model is equipped with a two-channel memory controller.

Testing the speed of the processor was carried out in Just Cause 2.

The final results showed that the Athlon X4 860K supports the maximum schedule resolution at the level of 1920 x 1200 points.

The computer also used an integrable GTX 580 video card.

In the figure below, you can see a comparative analysis of speed and other processors that have been tested with identical conditions of the software and hardware environment.

athlon X4 860K Testing Result

№3 - A10-7890K.

A10-7890K is a hybrid CPU from AMD. Despite the announcement of the development of a fundamental new technology and generation of processors, another A10 line model was solved in AMD.

The company positions this series of devices as an excellent choice for desktop PCs.

Model A10-7890K is the best-in-class solution for playback.

Of course, the graphics settings will have to be punished, but as a result you will get a good speed without strong overheating of the PC hardware.

packaging model A10-7890K.

This processor has a built-in graphic unit Radeon, which allows:

The processor comes with Wraith cooler, the feature of which is very quiet job. Also, the cooler supports the backlight mode. Specifications A10-7890K:

• CPU family - A-Series;
• Clock frequency: 4.1 MHz;
• The variety of connector: Socket FM2 +;
• Number of nuclei: 4 kernels;
• There is a unlocked multiplier;
• Number of contacts: 906;
• Approximate cost - $ 130.

The main plus A10-7890K is an improved interaction with Windows 10.

Detailed processor characteristics are indicated to us Figure below:

detailed features of APU A10-7890K

Results of component testing with standard test:

cinebench R15 test result

As you can see, the test component overtakes in its parameters some AMD models in the A-10 and Athlon line.

At the same time, the results obtained were not enough to exceed the analogues from Intel.

№2 - Ryzen 5 1600x

Two first places in our top are the models of the Ryzen line. It was in the past few years that the architecture of these processors has become a key key for Advanced Micro Devices.

The presented ZEN microarchitecture gradually returns the manufacturer leading positions on the market.

Ryzen 5 is a direct competitor for group processors. Bestly the CPU manifests itself in the game systems. This is also announced by the AMD CEO.

Characteristics:

• AMD Ryzen 5 family;
• 6 nuclei;
• Without integrated graphics;
• There is a unlocked multiplier;
• 3.6 MHz clock frequency;
• Socket AM4 connector;
• The cost is about $ 260.

Most 1600x modifications are devoid of native. Users will have to buy this component separately.

Basic frequencies do not cross the mark of 3.6 MHz installed. When working in the turbojym (as a result of acceleration of the processor), the clock frequency reaches a mark of 4.0 MHz.

All Fifth Generation Models Ryzen support SMT - surface mounting technology.

Thus, the CPU is easily mounted on the surface of the printed circuit board without the need to trim the parts of the component.

equipment Ryzen 5.

In the process of testing the work of the CPU, even with the most resource-proof programs, the maximum CPU temperature did not exceed 58 degrees. , Test results:

1600x model test

Together with the line of powerful CPP, the AMD has released a special firmware for their initial setup - AGESA.

The utility allows you to reconcile memory to avoid delays and interruptions to work.

№1 - Ryzen 7 1800x

Ryzen 7 1800X is an excellent choice for creating a powerful PC or for multi-level data support for data servers.

Currently, AMD is developing another powerful representative of the Ryzen family.

In March 2017, the APU Ryzen 2000 x model was announced, which should be on sale at the end of the year.

Characteristics:

• Family: AMD Ryzen 7;
• 8 cores;
• The clock frequency is 3.6 MHz with the possibility of overclocking up to 4 MHz;
• Support unlocked multiplier;
• No support for integrated graphics;
• The average price is $ 480.

1800x can simultaneously perform up to 16 program code streams. The processor works with SMT multithreading technology.

All Zen kernels provide efficient use of others. Increased access by supporting three-level cache.

Comparison of the results of testing Ryzen 7 1800x with competitive models from Intel.

AMD at a special event before CES 2018 has released new mobile processors and announced desktop chips with integrated graphics. And the Radeon Technologies Group, the AMD structural division, has announced the mobile discrete graphic chips vega. The company also disclosed plans for the transition to new technical processing and promising architectures: Graphic Radeon Navi and Processor Zen +, Zen 2 and Zen 3.

New processors, chipset and cooling

The first desktop Ryzen with VEGA graphics

Immediately two model of desktop Ryzen with built-in VEGA graphics will appear on sale February 12, 2018. Model 2200G refers to the processors of the initial segment of Ryzen 3, and 2400G - to the middle segment of Ryzen 5. Both models dynamically increase the frequency of 200 and 300 MHz from the base frequencies of 3.5 GHz and 3.6 GHz, respectively. In fact, they replace ultra-budget models Ryzen 3,1200 and 1400.

Blocks of graphics at 2200g are only 8 pieces, while 2400g is 3 more. The frequency of the graphic nuclei of 2200G reaches 1 100 MHz, and 2400g is more at 150 MHz. Each graphic block contains 64 shader.

The kernel of both processors wear the same code name as mobile processors with embedded graphics - Raven Ridge (letters. Vorona Mountain, rock in Colorado). But nevertheless, they are connected to the same LGA Nest AMD AM4, as well as all other processors Ryzen 3, 5 and 7.

Reference: Sometimes AMD calls processors with built-in graphics not CPU (Central Processing Unit, english Central Processor Device), and APU (Accelerated Processor Unit, Eng. Accelerated Processor Device, In other words, a video source processor).
AMD desktop processors with built-in graphics are marked with the letter G at the end, on the first letter of the word Graphics ( english graphics). Mobile processors and AMD and Intel label the letter U at the end, on the first letter of words Ultrathin ( english Ultra-thin) or Ultra-Low Power ( english Ultra-low power consumption), respectively.
It should not think that if the numbers of the models of the new Ryzen begin on the figure 2, the architecture of their nuclei belong to the second generation of ZEN microarchitecture. It is not so - these processors are still in the first generation.

New Mobile Ryzen with VEGA graphics

Last year, AMD has already brought to the market the first Mobile Ryzen codenamed Raven Ridge. All Mobile Family Ryzen is intended for game laptops, ultrabooks and hybrid laptop tablets. But there were only two such models, in the middle and senior segments: Ryzen 5 2500u and Ryzen 7,2700u. The younger segment was empty, but right on CES 2018 the company corrected - two models were added to the mobile family: Ryzen 3 2200u and Ryzen 3,3200u.

Vice President AMD Jim Anderson demonstrates Mobile Ryzen Family

The processor 2200U is the first dual-core CPU from all Ryzen, while 2300U is standard four-core, however, both of them work in four threads. At the same time, the base frequency of 2,200U nuclei is 2.5 GHz, and at 2300u lower - 2 GHz. But with increasing loads, the frequency of both models will rise to one indicator - 3.4 GHz. However, the ceiling of power can lower the manufacturers of laptops, because they still need to calculate energy costs and think over the cooling system. Also, there is a difference between the chips in the cache: 2200u has only two cores, and therefore two times less cache 1 and 2 levels.

The graphic blocks at the 2200u are only 3 pieces, but at 2300U - twice as many, as well as processor nuclei. But the difference in graphic frequencies is not so significant: 1,000 MHz against 1 100 MHz.

First Mobile Ryzen Pro

For the second quarter of 2018, AMD planned the release of Mobile versions of Ryzen Pro, corporate-level processors. Characteristics of mobile Pro are identical to consumer versions, with the exception of Ryzen 3 2200U, which has not received pro-implementation at all. Differences of desktop and mobile Ryzen Pro - in additional hardware technologies.

Ryzen Pro processors - full copies of ordinary Ryzen, but with additional features

For example, TSME is used to ensure security, hardware encryption of RAM "on the fly" (Intel has only software resource encryption SME). And for centralized control of the parking lot, the open standard Dash (Desktop and Mobile Architecture for System Hardware, English is available. Mobile and desktop architecture for system devices) - support for its protocols is built into the processor.

Laptops, ultrabooks and hybrid laptops with Ryzen Pro first must be interested in companies and state institutions that plan to buy them for employees.

New chipsets AMD 400th series

The second generation of Ryzen is relying the second generation of system logic: the 300th series of chipsets is replaced by 400th. The flagship series was expected to become AMD X470, and later the simpler and cheap sets of schemes, such as B450, will be released. The new logic has improved everything related to RAM: reduced access delay, raised the upper frequency limit and added a stock for overclocking. Also in the 400th series increased USB bandwidth and the power consumption of the processor has improved, and at the same time its heat dissipation.

But the processor nest has not changed. AMD AM4 Desktop Nest (and its mobile Valid Option AMD FP5) is a special advantage of the company. In the second generation, the same connector as in the first one. He is not changed in the third and fifth generation. AMD promised in principle not to change AM4 until 2020. And so that the Matplates of the 300th Series (X370, B350, A320, X300 and A300) earned with new Ryzen - just update the BIOS. Moreover, in addition to direct compatibility, there is both the opposite: old processors will work on new boards.

Gigabyte on CES 2018 has already shown the prototype of the first matplast on the new chipset - X470 Aorus Gaming 7 WiFi. This and other boards on X470 and junior chipsets will appear in April 2018, simultaneously with the second generation of Ryzen on the Zen + architecture.

New cooling system

AMD also introduced a new cooler AMD Wraith Prism (English. Prism Anger). While his predecessor Wraith Max was highlighted by one-color red color, Wraith Prism is equipped with a RGB-back-light controlled by a fan perimeter. The coolers of the cooler are made of transparent plastic and are also highlighted by millions of shades. RGB-backlit lovers will appreciate, and haters will be able to simply turn it off, although in this case the meaning of the purchase of this model is leveled.

Wraith Prism - full copy of Wraith Max, but with highlighting from millions of colors

The remaining characteristics are identical to Wraith MAX: direct contact heat tubes, programming profiles in overclocking mode and almost silent work on 39 dB under standard conditions.

There is no information about how much Wraith Prism will cost whether it will be supplied with processors and when it can be bought.

New laptops on Ryzen

In addition to mobile processors, AMD also promotes new laptops based on them. In 2017, HP Envy X360 models, Lenovo IdeaPad 720s and Acer SWIFT 3. In the first quarter of 2018, Acer Nitro 5, Dell Inspiron 5000 and HP are added to the mobile Ryzen. All of them work on last year's mobile Ryzen 7 2700u and Ryzen 5 2500U.

The Acer Nitro family is gaming machines. The Nitro 5 line is equipped with an IPS displays with a diagonal of 15.6 inches and a resolution of 1920 × 1080. And a discrete graphic chip Radeon RX 560 C 16 graphic blocks inside will be added to some models.

The Dell Inspiron 5000 laptop line offers models with a diagonal of displays of 15.6 and 17 inches, equipped with or hard drives or solid drives. Some model models will also receive a Discrete Radeon 530 video card with 6 graphic blocks. This is a rather strange configuration, because even in the integrated graph of Ryzen 5 2500U more graphic blocks - 8 pieces. But the advantage of the discrete card can be in higher clock frequencies and individual graphics memory chips (instead of the operational memory section).

Reducing prices for all Ryzen processors

Plans until 2020: Navi graphics, Zen 3 processors

2017 year for AMD has become a completely turning point. After many years of problems, AMD completed the development of the Zen nuclear microarchitecture and released the first generation of CPU: Ryzen Pro and Ryzen Threadripper PC Family, Ryzen Pro, and Ryzen Pro family, and the EPYC server family. In the same year, the Radeon group developed a VEGA graphic architecture: the VEGA 64 and VEGA 56 video card was released, and the end of the year the vega kernel was integrated into the Ryzen mobile processors.

Dr. Lisa SU, CEO AMD, assures that the company will release processors for 7 nanometers before 2020

New items not only attracted the interest of fans, but also seized the attention of ordinary consumers and enthusiasts. Intel and Nvidia had to hastily parry: Intel has released Coffee Lake six-core processors, unplanned the second "so" Skylake architecture, and NVIDIA has expanded the 10th series of video cards on the Pascal architecture to 12 models.

Rumors about AMD's future plans have been copied all 2017. Until now, Lisa Su, AMD general director, only noted that the company plans to exceed the 7-8% annual productivity growth rate in the electronic industry. Finally, at the CES 2018 exhibition, the company showed a "road map" not just until the end of 2018, and up to 2020. The basis of these plans is to improve chip architectures through miniaturization of transistors: a translational transition from the current 14 nanometers to 12 and 7 nanometers.

12 Nanometers: Second Generation Ryzen on Zen +

On the technical process of 12 nanometers, the Zen + microarchitecture is based, the second generation of the Ryzen brand. In fact, the new architecture is a modified Zen. The process of technological production of GlobalFoundries plants is translated from the 14-nanometer 14LPP (Low Power Plus, English. Low power consumption plus) on the 12-nanometer standard 12LP (Low Power, English. Low power consumption). The new technical process 12LP must provide 10% gains to chips.

Reference: The GlobalFoundries Factory Network is the former AMD production facilities allocated in 2009 to a separate company and combined with other contractors. By the share of the contract production market, GlobalFoundries divides second place with UMC, significantly inferior to TSMC. Chip developers are AMD, Qualcomm and others - order production both in GlobalFoundries and in other factories.

In addition to the new technical process, Zen + architecture and chips based on it will receive improved AMD Precision BOOST 2 technologies (ang. Accurate acceleration) and AMD XFR 2 (Extended Frequency Range 2, English. Advanced frequency range). In Mobile Processors, Ryzen can already find Precision Boost 2 and a special modification of XFR - Mobile Extended Frequency Range (MXFR).

In the second generation, the Ryzen Pro and Ryzen Threadripper PC and Ryzen Processor family will be released, but there are no information about updating generations of the Ryzen Mobile family and Ryzen Pro, and server EPYC. But it is known that some models of Ryzen processors from the very beginning will have two modifications: with a graph integrated into the chip and without it. Models of the initial and mean levels of Ryzen 3 and Ryzen 5 will be released in both options. And the high level of Ryzen 7 will not receive any graphic modification. Most likely, the code name of the Pinnacle Ridge (letters acute heroin of the mountain, one of the vertices of the Wind River Ridge Top River in Wyoming, is enshrined for the architecture of the nuclei for these processors.

The second generation of Ryzen 3, 5 and 7 will begin to be sold in April 2018 together with the 400 series chipsets. And the second generation of Ryzen Pro and Ryzen Threadripper will be raised to the second half of 2018.

7 Nanometers: Third Generation Ryzen on Zen 2, VEGA Discrete Graphics, Navi Graphic Core

In 2018, the Radeon group will release the discrete VEGA graphics for laptops, ultrabooks and laptop tablets. AMD is not divided into special details: it is known that discrete chips will work with compact multi-layer HBM2 type memory (RAM is used in the integrated graph). Separately, the Radeon emphasizes that the height of the memory chips will be only 1.7 mm.

Radeon Manager shows integrated and discrete vega graphics

And in the same 2018, Radeon will translate graphic chips on the VEGA architecture from the technical process of 14 nm LPP at once to 7 nm LP, completely jumping over 12 nm. But first, new graphics blocks will be supplied only for the Radeon Instinct line. This is a separate family of Radeon server chips for heterogeneous computing: machine learning and artificial intelligence - the demand for them is ensured by the development of unmanned cars.

And at the end of 2018 or early 2019, simple consumers wait for Radeon and AMD products on a 7-nanometer technical process: processors on the Zen 2 architecture and graphics on the Navi architecture. And work on the design of Zen 2 has already been completed.

The AMD partners will already be acquainted with the chips on Zen 2, which will create motherboards and other components under Ryzen. Such the AMD rates are gaining due to the fact that the company has two "jumping" each other teams to develop promising microarchitets. They began with parallel works on Zen and Zen +. When Zen was completed - the first team moved to Zen 2, and when Zen + was completed - the second team moved to Zen 3.

7 Nanometers "Plus": Fourth Generation Ryzen on Zen 3

While one AMD department solves the problem of mass production Zen 2, the other department already designs Zen 3 on the technological norm marked as "7 nm +". The company does not disclose details, but according to indirect data it can be assumed that the technical process will be improved due to the addition of the current deep ultraviolet lithography (DUV, DEEP Ultraviolet) with a new rigid ultraviolet lithography (EUV, EXTREME UltraViolet) with a long wave of 13.5 nm.

GlobalFoundries has already installed new equipment to transition to 5 nm

In the summer of 2017, one of the GlobalFoundries plants purchased more than 10 lithographic systems from the Twinscan NXE series from the Netherlands ASML. With partial use of this equipment, in the framework of the same technical process, 7 nm will be able to reduce energy consumption and increase the performance of chips. There is no accurate metrics yet - it will take some time to debug new lines and withdraw them for acceptable power for mass production.

AMD expects to start organizing chip sales at the rate of "7 nm +" with processors on the Zen 3 microarchitecture already until the end of 2020.

5 Nanometers: Fifth and subsequent generations Ryzen on Zen 4?

AMD's official announcement has not yet done, but you can safely speculate that the following borders for the company will become technical process 5 nm. Experienced chips on this norm were already produced by the IBM, Samsung and GlobalFoundries research alliance. The crystals on the 5 nm technical process will require no partial, but the full use of rigid ultraviolet lithography with an accuracy of 3 nm. It is such a permit that the Lithographic Systems of the TWINSCAN NXE: 3300B from ASML are purchased by the GlobalFoundries.

A layer of a thick molecule of the molybdenum disulfide (0.65 nanometer) demonstrates the leakage current of only 25 femtoamper / micrometer at a voltage of 0.5 volt.

But the difficulty lies also in the fact that on the process of 5 nm it will probably have to change the form of transistors. For the long-established finfet (transistors in the form of the fin, from the English FIN) can give way to a promising GAA FET (a form of transistors with surrounding shutters, from English Gate-All-Around). For commissioning and deploying the mass production of such chips take another year. The consumer electronics sector is unlikely to receive them earlier than 2021.

Further decrease in technological norms is also possible. For example, in 2003, Korean researchers created FINFET by 3 nanometer. In 2008, a nanometer transistor was created at the University of Manchesra based on graphene (carbon nanotubes). A sub-nanometer scale has been submitted to engineers-researchers of Berkeley's laboratory in 2016: in such transistors, both graphene and molybdenum disulfide (MOS2) can be used. True, at the beginning of 2018 it was not yet found a way to produce a whole chip or substrate from new materials.

A week ago, AMD conducted a small presentation dedicated to the new APU Ryzen Mobile previously known under the code name Raven Ridge. The speaker - however, as usual, - for a start, has permanently complained of the current situation in the world of processors. Like, the Moore law is no longer so strictly carried out and everyone has already managed to get used to "5-7% of growth a year" (it is known that this stone is known in whose garden). And even in desktops, where there are no special restrictions, five years ago, a massive competitor has 4 cores (and 8 threads) with a frequency of about 3.5 GHz, and until recently all the same 4C / 8T, but by about 4 GHz. Only this year a competitor has changed tactics, offering more cores for the same price as before. In the mobile segment in this sense, it was even worse in this sense - the stability of configurations is no longer a sign of skill. The lack of competition is bad for the market and end consumers. However, all this from AMD we have already heard earlier.

On the left - the CCX-block of the Zen nuclei, on the right - the GPU unit (blue)

The company itself has been developing new cores (CPU and GPU) for the last four years, and, according to AMD, it is important that they tried to make them the most scalable. On the same basis, powerful server solutions are made, and desktop systems, and now there are mobiles for laptops. Actually, AMD Ryzen Mobile 7 2700U and 5 2500U is one CCX for four Zen kernels (8 threads), Radeon Vega graphics and a slightly modified Infinity Fabric tire. The latter combines CPU, GPU, memory controller, display and multimedia blocks, as well as a peripheral controller. The basic version of both chips has a TDP 15 W, but the manufacturers of systems with AMD approval can independently configure the TDP in the range from 12 (the table indicates 9, but it was repeatedly announced that it was 12) to 25 W - everything will depend on the quality of the cooling system. The user is not available to the user.

At the microarchitecture level, the new APU differ from the desktop versions of crystals and. Changes relate to those areas that are critical for the mobile segment. Developers, for example, cut L3-keysh up to 4 MB - just in order not to inflate the size of the crystal. From HBM for GPU also had to refuse - video memory is cut off from the main DDR4. The specific volume depends on the OEM manufacturer of the laptop. For tests (benchmarks are shown below) AMD used configurations with 256 MB of video memory, but in general there will be options for 512-1024 MB, the benefit of the relatively large volume of RAM in modern laptops has long been not uncommon. And yes, the overall performance of the complex will again be partially dependent on the frequency of RAM.

The DDR4-2400 memory controller also has almost unchanged: it is two-channel, but for some ultraportative AMD solutions insists on the use of a single-channel configuration - in this case, the difference in the performance of graphics will be about 20-40%. ECC is supported, but in laptops we are unlikely to see. Differences between AMD Ryzen Mobile 7 2700U and 5 2500U are not so great. The older model has a basic and increased frequency of 2.2 and 3.8 GHz, respectively, and the youngest - 2.0 and 3.6 GHz. In 2500U there are eight Cu-modules with Radeon Vega with a frequency of 1.1 GHz, and in 2700U they are already ten and they work for 1.3 GHz. Yes, so far only two APU models will be available, but next year AMD promises to significantly increase their number. The crystal has an area of \u200b\u200b209.78 mm 2 and contains approximately 4.95 billion transistors. TEHPROTSESS - 14 nm.

However, some important changes in new chipsets are worth mentioning. The Dynamic Crystal Frequency Control Technology of Precision Boost acquired 2 in the title. She still changes frequencies with a step of 25 MHz, but in this case such a step is used in the GPU, and in the CPU. In addition, the new version is better coping with a multi-threaded load - the main limiting factor in the case of laptops will be more cooling efficiency than the power limit. In addition, the Mobile XFR subsystem appeared in the new APU - it also further increases the turbocarity above the nominal, but here it is the task to hold out of considerable acceleration as long as possible. The exact magnitude of the frequency growth, the number of activated nuclei and the specific APU models with the presence of MXFR were not declared, however, it is reported that this technology is designed rather to productive laptops with good cooling.

However, some additions are also provided in the nutrition subsystem. There are thousands of individual sensors (and regulators) in the crystals, which measure the voltages directly from the transistor blocks, and with an accuracy of Milvolt. That is, data on the state of external VREG is no longer so important. Voltage adjustment for individual ZEN nuclei has already been, and now it has been added for GPU. Curious approval of the AMD representative is curious that the worst version of the load when the peak occurs simultaneously on the CPU and GPU, in practical scenarios of work, it is not found. With this, of course, you can argue. Nevertheless, the main task in the case of APU is the correct and fast power distribution between the graphic and the processor part depending on which one really needs it. Actually, the main innovation in the APU is LDO regulators built into the GPU. It is argued that such an effective implementation of this technology is now not anyone.

New LDO unified for CPU / GPU, as AMD itself says, in the case of APU, reduce current requirements by 36%, in addition to 20% the maximum current to power CPU or GPU - in fact, you can make either a more powerful solution, leaving that The same system, or, on the contrary, reduce it, but maintain performance. In any case, the energy efficiency of the final solution increases, because the dynamic distribution of frequency and nutrition, depending on the load, occurs between the CPU cores, and between graphic and central processors. However, the specific details of the distribution algorithm are not disclosed. On the other hand, not only the algorithm is important, but also the speed of switching between different states of the CPU / GPU and their number, which, in particular, is necessary for more efficient use of the laptop battery.

In the new APU, the GPU has a special mode in which the power consumption of the card is reduced by 95%. It is activated when nothing happens on the screen in a literal sense, that is, a static picture is displayed, - for example, if the user just moved to the time from PC. There is a similar state for cpu cores. The transition between the main states in both cases occupies 100 microseconds or less (typical value - 50 microseconds), and for a deep sleep mode - up to 1.5 ms. In addition, the internal components of the APU are conditionally divided into two zones with different nutrition policies, which also contributes to energy efficiency. On the InFinity Fabric bus, these internal sensors and regulators are transmitted.

Also, developers note a small thickness of the finished product - only 1.38 mm. Earlier, as stated, not in all ultrabooks managed to put the available chips just because of their thickness. As for the GPU, it is worth noting the presence of Freesync technology 2. AMD will try to make it possible to support its support in the displays of their laptops. The video card itself supports multi-meter configurations, an image output with a 4K and HDR resolution. Right now, together with Microsoft, PLAYReady support is preparing, which is necessary for the correct operation of some video system services. Well, in general, AMD continues to adhere to a long-term strategy of 25 × 20, which was announced in 2014. According to it, by 2020, the APU's overall performance should grow 25 times compared with 2014 models.

Unfortunately, during the presentation, AMD never submitted the full characteristics of new products (no, for example, data on integrated controllers for the periphery), showing only some benchmarks. Note several important points in them. First, in some cases, the comparison is not with the solutions of a competitor, but only with AMD products on the old platform. Secondly, where such a comparison is still there, the chip of the eighth generation with the same nominal TDP 15 W, which was available on the market (and there are still a bit). Thirdly, various acceleration technologies were not involved or any other "Cheating", including, for example, laptop tests in a pre-cooled room. Below in the gallery presents test results, as well as comments and notes to them.

Benchmark AMD Ryzen Mobile

The best novelties show themselves in multi-threaded applications, as well as in software that actively uses the graphical subsystem. AMD notes that now on ultra-thin laptops you can, for example, calmly engage in video and graphics and graphics and does not greatly worry about the autonomy of the device. And of course, for them, according to the company, a new niche appears - games. Naturally, heavy igro farms here will feel uncomfortable here, but popular cybersport projects work well with acceptable resolution and quality of graphics. By the way, options with Dual Graphics are not yet expected, instead, developers can use DirectX 12 to share resources from different GPUs.