
In a consequence the name of the processor changed to SANDY BRIDGE ("Pesochny Bridge"), and to final release it remained to the same. Issue in 2011 of family of the Intel processors executed on 32-nanometer technical process if won't cause revolution, will precisely bring essential changes in the market of personal computers.
Let's consider closer that prepared Intel.
The new chipset of the processor represents the tetracyclic 64-bit processor with own controler of memory to DDR3 integrated by a graphic kernel and changeable sequence of execution of teams. Also it is necessary to note the new ring tire Ring Interconnect, support of two data flows on a kernel in the Hyper-Threading mode and execution at once four teams for a step. To put such Intel processor it will be possible only on the new motherboard based on a chipset of Intel 6 Series, LGA 1155 topped with a socket. Short of Intel Atom, SANDY BRIDGE is the first microarchitecture, in which GPU and CPU are integrated on one general crystal. At the expense of such decision the processor and the video chip divide the general cache which has been also integrated into the chip that increases overall performance of all system as a whole.
Designation of Intel processors
In markings and designations of Intel didn't begin to invent something new. Sandy Bridge officially is considered the second generation of CPU of architecture of Core, so, habitual division of processors into groups i3, i5, i7 remains, and before figures of number of model there will be a two designating just accessory of a stone to the second generation.
Besides, by old tradition in the Intel company decided to give to some processors an index standing after the name of model. Some consider that such division confuses the buyer though anything difficult actually isn't present. Now it is known about three indexes: To, S and T.
The greatest interest is caused by stones with K.Im's index ardent pieces of silicon with the unblocked multiplier will be designated. Already now it is known that some engineering samples were dispersed to 5 GHz «on air»! Undoubtedly, since the moment of emergence in the market these processors will please to gamers and overkloker.
It is expected that at all capacity and will cost such Intel processors is slightly higher than the usual. It is necessary to note also that else in 2008 the AMD company presented Black Edition series processors with the multiplier unblocked on increase, and continues to let out them and until now. Such step with approval was met by fans of dispersal, however the present possibility, as a rule, isn't necessary to ordinary users.
The index of S will be received by TsPU with the lowered frequencies and a thermal emission. They will come in handy in the multimedia centers, NTRS, desktop systems and powerful laptops. At sufficient productivity such stone doesn't demand a powerful cooler and is very economic.
Letter T will designate the most economic models by which, most likely, will be equipped неттопы, laptops, netbooks and other portable devices from which it is not required solutions of complex challenges and which time of independent work and economy of resources play far not the last role. Some models remained without an index at all: according to preliminary data, it is ordinary processors with the blocked multiplier which, nevertheless, without problems will consult with all assigned to them corporate or daily tasks.
Held down by one chain
So it was moved that with each generation of Intel processors and not only she, tries to contain in a crystal more and more functional modules which settled down out of the processor - on the motherboard or in a chipset earlier. No wonder that at such configuration also requirements to capacity of intercomponental tires of data transmission repeatedly increased. Until recently it was possible to manage tires with cross topology - it was quite enough for Arrandale and Clarkdale chips at which the graphic processor was introduced in architecture of a crystal. In a case with SANDY BRIDGE developers decided to refuse already out-of-date decision in favor of ring topology of the 256-bit componental tire which is executed on the basis of the modifed version of the Quick Path Interconnect (QPI) technology. To this QPI it was realized in architecture of server Heon 7500. In a novelty the ring tire provides data exchange between the main components of a crystal: four processor kernels, L3 cache, system agent and graphic kernel. It is possible to present the tire as four 32-byte rings: tire of data, tire of inquiries, tire of confirmation and tire of monitoring of a condition. The communications protocol of the distributed arbitration operates tires, processing of inquiries thus occurs on clock frequency of kernels of the processor. The topology of the tire allows to transfer data with speed of 96 Gbytes / with to connection at frequency of 3 GHz. Actually these figures four times exceed indicators of Intel processors of the previous generation.
Memory at the last boundary
Let's address to the cache of L3 mentioned above which on the slides found in the Network from Intel is designated as LLC - Last Level Cache, or «a cache of the last level». Thanks to the ring tire it is distributed not only between processor kernels, but also between a graphic kernel and the system agent into which developers integrated the block of a video output and the module of hardware acceleration of graphics. Between kernels of the processor and graphics the special route-making mechanism is engaged in anticipation of the conflicts of access. Each of four kernels, in turn, in the new Intel processor got direct access to a "personal" segment of the cache, thus each segment provides a half of width of the tire for access from the general ring tire of data. Besides, each segment is allocated with the independent controler of access to the tire which is responsible for processing of inquiries on placement of physical addresses. Also this controler constantly exchanges data with the system agent, revealing unsuccessful appeals to L3 cache, fixing nekeshiruyemy addresses and supervising intercomponental data exchange.
System agent
The system agent in Intel call everything that doesn't treat a kernel: graphics, memory controler, cache, both other, and other... In a new chipset of Intel it is possible to call it elements of the northern bridge which with a view of acceleration in new architecture removed from a chipset to a stone. The controler of memory of DDR3, the PCIE 2.0 controler, the module of management got to System Agent a food, the video output block etc. The agent is reported with other elements of system by means of the same ring tire. It is necessary to note especially that the memory controler () works at a crystal in a two-channel mode. It, undoubtedly, will afflict owners of sets of a three-channel operativka who had to leave the considerable sums of cash. The supply voltage controler in the system agent provides dynamic change of tension and clock frequencies of kernels of the processor, GPU, interfaces and the memory controler, and management of a food and frequency occurs independently for the block of graphics and processor kernels. Especially it is necessary to note the Turbo Boost technology thanks to which both GPU, and CPU can work at the increased frequencies, with excess of declared TDP, but without any consequences for system.
On responses of the guests who were present on closed demonstration of a chipset of Intel at far San Francisco, the gain of productivity can reach fantastic levels. Thus, already now it is possible to say that when using system of cooling of sufficient capacity it is possible to turn future computer on the basis of a described stone into the superproductive monster.
Absolutely precisely it is known that the coolers calculated on cooling of Intel processors in конструктиве by LGA 1156 (Socket H), will rise without any completions on new (Socket H2) on LGA 1155, as geometry of openings at them identical.
Also engineers took care and about power system effectiveness - it is possible to transfer each kernel of the processor if necessary to a mode of the minimum power consumption, the same is provided and for graphics. The ring tire and a cache, of course, to "calm" it will not turn out, but for the first «waiting mode» during the idle time moments is provided, and the technology of shutdown of not used transistors is applied to a cache of L3 already familiar to us on last architecture of Intel.
At last it is necessary to tell about modules of multimedia hardware decoding and a video output. In considered architecture the productive and economic module is applied to processing of multimedia streams, and possibilities of a graphic kernel will be involved only at resursoyemky processes of processing of video data. Thanks to all these innovations and counters the hardware module of decoding of SANDY BRIDGE without effort will cope with pair of video streams independent from each other in Full HD permission.
Video with big advantage
Now we will consider shortly graphic system SANDY BRIDGE. The new graphics, according to representatives of Intel, first of all is intended for desktop computers and various portable devices. Considering the announced "clever" scheme of power consumption, it would be desirable to be put on that we will see these new processors and in tablet personal computers which in a large number will appear in the market in the near future.
Typical tasks for similar systems are first of all surfing in Networks, online video and music viewing, communication in social networks and by means of various messengers, work with documents, viewing of DVD and already become habitual Blu - ray - a content, start browser and DirectX of 10.1 games. Thus, Intel doesn't aspire to compete to competitors in the market of high-efficiency graphic chips for videocards where rules ball the 11th version of DirectX libraries and where integrated GPU doesn't need while to be put. However on already mastered field of the three-dimensional graphics built in the processor the company is ready to show very and very impressing result. In graphic system there is a lot of innovations, all of them also are aimed at increase in productivity at work with 3D-applications.
It is necessary to pay attention to one curious fact also: with the built-in graphics which resursoyemky appendices didn't interest, possibilities of processors of architecture of Clarkdale quite sufficed to the majority of owners of desktop and mobile personal computers, however experts from Intel didn't begin to waste time on trifles and in a new chipset of Intel increased capacity of integrated video already twice. The built-in graphic processor as it was specified earlier, was comfortably arranged together with other components on one crystal, unlike the previous generation of graphics where GPU together with henchmen "expelled" on the separate 45-nanometer chip which was connected to the main 32-nanometer stone according to the previous version of the QPI technology. In addition, unlike Clarkdale where for management of productivity and a temperature mode the softovy driver which has been rigidly adhered to the general system parameters answered, in SANDY BRIDGE other picture is observed absolutely. Various modes 3D in this microarchitecture cope the system agent at hardware level. It allows to change parameters of supply voltage, frequencies, power consumption and productivity in wider limits isn't less effective, than to make the same operations with processor kernels. Being based on such ample opportunities of management of the video chip, Intel promises to introduce the Turbo Boost technology in stationary and mobile devices.
DirectX, trigonometry and logarithms
Unlike the previous generation built in Sandy Bridge GPU supports DirectX 10.1 possibilities though developers did the main emphasis on the maximum use of hardware possibilities, first of all for processing of media and data and обсчета 3D-functions. This approach cardinally differs, for example, from solutions of NVIDIA where completely programmed hardware model is taken advantage. In the new chip the withdrawal from programmed flexibility is reasonable by benefit in the form of more high efficiency against economy of the power consumption, the simplified model of programming of drivers and that isn't less important, reduction of the physical sizes of the module of graphics. Also developers of microarchitecture increased the sizes of the register file for executive blocks that allowed to reach effective implementation of complex sheyder. Besides, in executive blocks branching for the best rasparallelivaniye of executed teams is optimized. According to Intel statements, new blocks possess the doubled capacity at the expense of what productivity of calculations with transcendental numbers - natural logarithms, the trigonometry - will jump up to 20 times!
Effective power consumption of the processor
Providing correct "diet" in the future architecture is assigned to System Agent. By means of the Turbo Boost technology power consumption of all four kernels of the processor and graphics is optimized, the rational scenario of loading of processor kernels and a graphic kernel is applied to each task. On IDF 2010 engineers of Intel emphasized that they managed to reduce considerably appetites of the new processor and to increase efficiency of functioning of Turbo Boost at the expense of what stone power consumption in a mode, for example, background reproduction of a HD content - decreases twice.

SANDY BRIDGE sales plans
To us promise active sales of SANDY BRIDGE after Christmas holidays. The first in the market will appear two - and tetracyclic processors of new architecture. Of preliminary information on them it is already full on the Internet and in order that the new thing didn't stand idle, Intel is going to allow sales of motherboards with new sets of logic of Intel 6 Series at the same time from the beginning of realization of processors. It is expected that fresh decisions from Intel will increase demand in sector of the house media centers growing now and entertaining systems, the benefit the integrated video of this architecture without effort supports reproduction even a "heavy" HD content, characteristic for Blu-ray of disks. From here it is easy to draw a conclusion that along with ordinary motherboards of the ATX/microATX format on sale sooner or later will appear a set of compact systems mini-ITX as the low thermal emission of processors as well as possible promotes it. At Core i5 2400Т, for example, declared TDP in face value of only 35 W.
Rough schedule of an exit of Intel processors and their range:
Processor
|
Number of kernels/streams
|
Frequency of kernels, GHz;
face value/Turbo Boost
|
Cache of L3, Mbayt
|
Frequency of video of a kernel, GHz;
face value/Turbo Boost
|
TDP, W
|
Exit date
|
Core i7-2600K |
4/8
|
3,4/3,8
|
8
|
850/1350
|
95
|
09.01.2011
|
Core i7-2600 |
4/8
|
3,4/3,8
|
8
|
850/1350
|
95
|
09.01.2011
|
Core i7-2600S |
4/8
|
2,8/3,8
|
8
|
850/1100
|
65
|
09.01.2011
|
Core i5-2500K |
4/4
|
3,3/3,7
|
6
|
850/1100
|
95
|
09.01.2011
|
Core i5-2500 |
4/4
|
3,3/3,7
|
6
|
850/1100
|
95
|
09.01.2011
|
Core i5-2500S |
4/4
|
2,7/3,7
|
6
|
850/1100
|
65
|
09.01.2011
|
Core i5-2500T |
4/4
|
2,3/3,3
|
6
|
650/1250
|
45
|
09.01.2011
|
Core i5-2400 |
4/4
|
3,1/3,4
|
6
|
850/1100
|
95
|
09.01.2011
|
Core i5-2400S |
4/4
|
2,5/3,3
|
6
|
850/1100
|
65
|
09.01.2011
|
Core i5-2300 |
4/4
|
2,8/3,1
|
6
|
850/1100
|
95
|
09.01.2011
|
Core i3-2390T |
2/4
|
2,7/3,5
|
3
|
650/1100
|
35
|
20.02.2011
|
Core i3-2120 |
2/4
|
3,3
|
3
|
850/1100
|
65
|
20.02.2011
|
Core i3-2100 |
2/4
|
3,1
|
3
|
850/1100
|
65
|
20.02.2011
|
Core i3-2100T |
2/4
|
2,5
|
3
|
650/1100
|
35
|
20.02.2011
|
The first there will be motherboards on the basis of sets of chips Н67 supporting built in GPU processor, and Р67, focused only on the discrete graphic adapter. Over time the budgetary set of logic should adjoin them Н61. He is on friendly terms with the integrated graphics, but is cut down by number of USB 2.0 ports and in possibilities of the organization of RAID massifs at chipset level. Won't be in it and the SATA interface of 6 Gb/s.
In all sets of chips capacity of the tire PCIE is doubled, one which two-forked line can transfer 1 Gbit/pages given for the speed. Traditional BIOS with an exit of new payments will be gradually sent on pension, giving way to EFI (Extensible Firmware Interface) that in practice will allow to support hard disks of large volume (more than 2 Tbyte) and to use any functions of the computer, for example access to the network, without OS loading.
Further development of Intel processors
As usual, coverage of all price segments with a world gain in far prospect. By the end of 2011 promised to prepare replacement to LGA 1366 processors - his successors will be inserted into a new socket, on hearings, it will be LGA 2011.
The following generation of Intel processors in 2012 declared 22-nanometer Ivy Bridge chips behind which mysterious Haswell approaches. But, judging by experience of previous years to wait for these miracles for certain still very long, and as a result everything leaves absolutely in a different way and names will for certain be replaced.
No comments:
Post a Comment