What socket does the intel i5 processor have? Intel processor sockets
Hello everyone, dear blog guests! Today we will look at generations of intel processors - a table by year, release date of each, as well as how to find out what generation of processor is in the computer. We'll talk about Core I7. Pentium and I5 are topics for separate posts.
From this article you will learn:
Brief description of the series
Core i7 – top processors from Intel, occupying flagship and sub-flagship positions. Before the advent of i9, they were the most powerful, second only to server Xeons. The model range has been produced for more than 10 years and is designed for use in powerful gaming and work computers.
Over all this time, 9 generations of this CPU model have been created. Unlike younger models, it is easier to get confused about them, since each line has several subseries that differ in operating parameters. 
Conventionally, these chips can be divided into stock and advanced. The latter have their own “ecosystem” of corresponding motherboards, chipsets and sockets. They belong to the so-called X series. The following designations are also used in the marking:
- K – unlocked multiplier and overclocking support;
- S – reduced energy consumption;
- T – very reduced;
- E – CPU for embedded systems;
- C and R – chips with Iris graphics.
Let's consider the history and features of all generations of this model
1st generation
The first series of this model went on sale in 2008. Even before the appearance of i3 and i5, this line switched to a new naming. Chips with model numbers 920, 930, 940, 950, 960, 965, 975 were created using the 45 nm process technology. All CPUs had 4 cores that worked in eight threads.
A new platform with a 1336-pin connector and DDR3 memory modules has been developed for these chips.
After the appearance of the more convenient socket 1156 in 2009, a series with numbers 860, 860, S 870, 875K and 880 was released. The characteristics did not differ from their predecessors, but the assembly was cheaper due to cheaper motherboards with this socket.
The controller was simplified so that only two memory channels were supported. 
The pinnacle of this generation was the Gulftown architecture CPU. Such CPUs received indexes 970, 980, 980X and 990X. They were created using a 32 nm process and were six-core. Supported three-channel memory mode and connected via socket 1366.
2nd generation
The architecture was changed to Snady Bridge and finally switched to a 32 nm process technology. In the basic series, processors 2600, 2600S, 2600K, 2700K were released - four-core, eight-thread, worked with single-channel memory and were mounted in the new 1155 sockets.
A logical continuation was the model for the 2011 platform, which replaced the outdated 1366. This is a CPU with codes 3820, 3930K, 3960X, 3970X. The younger model had 4 cores, the older ones 6. A new product was a four-channel controller for DDR III memory.
3rd generation
The Ivy Bridge architecture was used, a modified version of its predecessor with a 22 nm process technology. Within the line, chips with indices 3770, 3770S, 3770T, 3770K were created - quad-core, with support for two DDR3 channels. 
An integrated video card was used for the first time. The chips could be mounted on socket 1155.
As part of the X series, modifications with code numbers 4820K, 4930K and 4960X were released. Installed in socket 2001 and supported 4 DDR3 channels.
4th generation
A large number of modifications have been created on the Haswell architecture - 4765T, 4770, 4770K, 4770S, 4770T, 4770TE, 4771, 4785T, 4790, 4790T, 4790S, 4790K. They were mounted on boards with the new 1150 socket and had a built-in HD 4600 graphics chip.
5th generation
There was no mass production of processors in this series. The manufacturer mastered the 14 nm process technology using Broadwell architecture. Only two models were created: 5775C and 5775R - the same chip with the Iris Pro 6200 graphics accelerator.
The X series includes models 6800K, 6850K, 6900K and 6950X. They worked with four-channel DDR 4 memory and were installed in the 2011 slot of the third version.
6th generation
Using a 14 nm process technology, the manufacturer released the sixth generation, represented by models 6700, 6700K, 6700T and 6700TE. These CPUs had four cores, a built-in HD 530 graphics card, and were based on the SkyLake architecture.
The dual controller supported DDR3 and DDR4. Mounted on connector 1151. 
In the top category, three modifications were released: 7800X, 7820X, 9800X. They were installed in socket 2066.
7th generation
The upgraded Kaby Lake architecture was used, which was produced using the 14 nm process technology. Models 7700, 7700T and 7700K were released. Compatible with 1151 boards. Only one chip was released in the X-series - 7740X, a quad-core for the 2066 platform.
8th generation
The eighth generation chips, based on the Coffee Lake architecture, appeared in 2017. The model range includes 8700, 8700K and 8700T, which had 6 cores each. The socket has been updated to version 1151, support for DDR3 has been removed. The 8086K was released as a limited edition to celebrate the 40th anniversary of the Intel 8086 CPU.
9th generation
The chips released in 2019 did not receive any fundamental innovations. The same architecture and the same technical process were used. So far, the latest model range has two processors: 9700KF and 9700K. 
They work in the same boards as the previous generation CPUs. These chips already have eight cores.
When purchasing a new processor, you can determine which generation it belongs to by this description. No more models were produced, so it's easy to check.
| Ninth | |||
| i7-9700KF | 1151–2 | 14 nm | 2019 |
| i7-9700F | 2019 | ||
| i7-9700K | 2018 | ||
| i7-9800X | 2066 | 2018 | |
| Eighth | |||
| i7-8086K | 1151–2 | 14 nm | 2018 |
| i7-8700K | 2017 | ||
| i7-8700 | 2017 | ||
| i7-8700T | 2017 | ||
| Seventh | |||
| i7-7820X | 2066 | 14 nm | 2017 |
| i7-7800X | 2017 | ||
| i7-7740X | 2017 | ||
| i7-7700K | 1151–1 | 2017 | |
| i7-7700 | 2017 | ||
| i7-7700T | 2017 | ||
| Sixth | |||
| i7-6950X | 2011–3 | 14 nm | 2016 |
| i7-6900K | 2016 | ||
| i7-6850K | 2016 | ||
| i7-6800K | 2016 | ||
| i7-6700K | 1151–1 | 2015 | |
| i7-6700 | 2015 | ||
| i7-6700T | 2015 | ||
| Fifth | |||
| i7-5960X | 2011–3 | 22 nm | 2014 |
| i7-5930K | 2014 | ||
| i7-5820K | 2014 | ||
| i7-5775C | 1150 | 14 nm | 2015 |
| Fourth | |||
| i7-4960X | 2011 | 22 nm | 2013 |
| i7-4930K | 2013 | ||
| i7-4820K | 2013 | ||
| i7-4790K | 1150 | 2014 | |
| i7-4790 | 2014 | ||
| i7-4790S | 2014 | ||
| i7-4790T | 2014 | ||
| i7-4785T | 2014 | ||
| i7-4770K | 2013 | ||
| i7-4771 | 2013 | ||
| i7-4770 | 2013 | ||
| i7-4770R | BGA1364 | 2013 | |
| i7-4770S | 1150 | 2013 | |
| i7-4770T | 2013 | ||
| i7-4765T | 2013 | ||
| Third | |||
| i7-3970X | 2011 | 32 nm | 2012 |
| i7-3960X | 2011 | ||
| i7-3930K | 2011 | ||
| i7-3820 | 2012 | ||
| 1155 | 22 nm | 2012 | |
| 2012 | |||
| 2012 | |||
| 2012 | |||
| Second | |||
| i7-2700K | 1155 | 32 nm | 2011 |
| i7-2600K | 2011 | ||
| i7-2600 | 2011 | ||
| i7-2600S | 2011 | ||
| First | |||
| i7-995X | 1366 | 32 nm | 2011 |
| i7-990X | 2011 | ||
| i7-980X | 2010 | ||
| i7-980 | 2011 | ||
| i7-975E | 45 nm | 2009 | |
| i7-970 | 32 nm | 2010 | |
| i7-960 | 45 nm | 2009 | |
| i7-965E | 2008 | ||
| i7-950 | 2009 | ||
| i7-940 | 2008 | ||
| i7-930 | 2010 | ||
| i7-920 | 2008 | ||
| i7-880 | 1156 | 2010 | |
| i7-875K | 2010 | ||
| i7-870 | 2009 | ||
| i7-870S | 2010 | ||
| i7-860 | 2009 | ||
| i7-860S | 2010 | ||
You may also find the publications “
For an office, home or gaming computer, it is not so difficult to choose the right processor. You just need to decide on your needs, orient yourself a little in the characteristics and price ranges. There is no point in thoroughly studying the smallest nuances if you are not a “geek,” but you need to understand what to pay attention to.
For example, you can look for a processor with a higher frequency and cache memory, but without paying attention to the core of the chip, you can get into trouble. The core, in fact, is the main performance factor, and the rest of the characteristics are plus or minus. In general terms, I can say that the more expensive the product in the line of one manufacturer, the better, more powerful, and faster it is. But AMD processors are cheaper than those from Intel.
- The processor should be chosen depending on the tasks at hand. If in normal mode you have about two resource-intensive programs running, then it is better to buy a dual-core “stone” with a high frequency. If more threads are used, it is better to opt for a multi-core processor of the same architecture, even with a lower frequency.
- Hybrid processors (with a built-in video card) will allow you to save on the purchase of a video card, provided that you do not need to play fancy games. These are almost all modern Intel and AMD processors of the A4-A12 series, but AMD has a stronger graphics core.
- All processors marked “BOX” must be supplied with a cooler (of course, a simple model, which will not be enough for high loads, but is just what is needed for operation in nominal mode). If you need a cool cooler, then .
- Processors marked “OEM” are covered by a one-year warranty, while processors marked “OEM” are covered by a three-year warranty. If the warranty period provided by the store is shorter, it is better to think about looking for another distributor.
- In some cases, it makes sense to buy a percentage from hand, this way you can save about 30% of the amount. True, this method of purchase is associated with a certain risk, so you need to pay attention to the availability of a guarantee and the reputation of the seller.
Main technical characteristics of processors
Now about some characteristics that are still worth mentioning. It is not necessary to go into it, but it will be useful to understand my recommendations for specific models.
Each processor has its own socket (platform), i.e. the name of the connector on the motherboard for which it is intended. Whatever processor you choose, be sure to look at socket matching. At the moment there are several platforms.
- LGA1150 – not for high-end processors, used for office computers, gaming and home media centers. Entry-level integrated graphics, except Intel Iris/Iris Pro. Already going out of circulation.
- LGA1151 is a modern platform, recommended for future upgrade to newer hardware. The processors themselves are not much faster than the previous platform, i.e., there is little point in upgrading to it. But there is a more powerful integrated graphics core of the Intel Graphics series, DDR4 memory is supported, but it does not provide a significant performance gain.
- LGA2011-v3 is a top-end platform designed for building high-performance desktop systems based on Intel X299 system logic, expensive, outdated.
- LGA 2066 (Socket R4) - socket for HEDT (Hi-End) Intel processors of Skylake-X and Kaby Lake-X architecture, replaced 2011-3.
- AM1 for weak, energy-efficient processors
- AM3+ is a common socket, suitable for most AMD processors, incl. for high-performance processors without an integrated video core
- AM4 is designed for microprocessors with Zen microarchitecture (Ryzen brand) with and without integrated graphics, and all subsequent ones. Added support for DDR4 memory.
- FM2/FM2+ for budget versions of Athlon X2/X4 without integrated graphics.
- sTR4 is a connector type for the HEDT family of Ryzen Threadripper microprocessors. Similar to server sockets, the most massive for desktop computers.
There are outdated platforms that you can buy in order to save money, but you need to take into account that new processors will no longer be made for them: LGA1155, AM3, LGA2011, AM2/+, LGA775 and others that are not on the lists.
Kernel name. Each line of processors has its own kernel name. For example, Intel currently has Sky Lake, Kaby Lake and the newest eighth generation Coffee Lake. AMD has Richland, Bulldozer, Zen. The higher the generation, the more high-performance the chip, with lower energy consumption, and the more technologies are introduced.
Number of Cores: from 2 to 18 pieces. The bigger, the better. But there is such a point: programs that do not know how to distribute the load across the cores will work faster on a dual-core with a higher clock frequency than on a 4-core, but with a lower frequency. In short, if there is no clear technical specification, then the rule works: more is better, and the further, the more correct it will be.
Technical process, measured in nanometers, for example – 14nm. Does not affect performance, but does affect processor heating. Each new generation of processors is manufactured using a new technical process with a smaller nm. This means that if you take a previous generation processor and a new one that is approximately the same, the latter will heat up less. But, since new products are made faster, they heat up about the same. That is, improving the technical process allows manufacturers to make faster processors.
Clock frequency, measured in gigahertz, for example - 3.5 GHz. Always the more the better, but only within one series. If you take an old Pentium with a frequency of 3.5 GHz and some new one, then the old one will be many times slower. This is explained by the fact that they have completely different kernels.
Almost all “stones” are capable of accelerating, i.e. operate at a higher frequency than that specified in the specifications. But this is a topic for those knowledgeable, because... You can burn the processor or get a non-working system!
Level 1, 2 and 3 cache size, one of the key characteristics, the more, the faster. The first level is the most important, the third is less significant. Directly depends on the kernel and series.
TDP– dissipated thermal power, or how much at maximum load. A lower number means less heat. Without clear personal preferences, this can be ignored. Powerful processors consume 110-220 watts of electricity under load. You can see a diagram of the approximate energy consumption of Intel and AMD processors under normal load, the less the better:
Model, series: does not relate to the characteristics, but nevertheless I want to tell you how to understand which processor is better within the same series, without delving too much into the characteristics. The name of the processor, for example "Intel i3-8100", consists of the "Core i3" series and the model number "8100". The first number means the line of processors on a certain core, and the next ones are its “performance index,” roughly speaking. So, we can estimate that:
- Core i3-8300 is faster than i3-8100
- i3-8100 is faster than i3-7100
- But the i3-7300 will be faster than the i3-8100, despite the lower series, because the 300 strongly more than 100. I think you get the idea.
The same goes for AMD.
Will you play on the computer?
The next point that you need to decide in advance is the gaming future of the computer. For “Farm Frenzy” and other simple online games, any built-in graphics will do. If buying an expensive video card is not part of your plans, but you want to play, then you need to buy a processor with a normal graphics core Intel Graphics 530/630/Iris Pro, AMD Radeon RX Vega Series. Even modern games will run in Full HD 1080p resolution at minimum and medium graphics quality settings. You can play World of Tanks, GTA, Dota and others.
Comments (233)
- In contact with
Alexey Vinogradov
Answer
-
Answer
Answer
-
Elena Malysheva
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Basil
Feb 25, 2020Answer
Alexey Vinogradov
Answer
iUnhead
Feb 10, 2020Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Sergey
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
-
Answer
Alexey Vinogradov
Answer
-
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
-
Answer
-
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Gregory
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexander S.
Answer
-
Answer
-
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexander S.
Answer
Leonid
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexander S.
Answer
-
Answer
-
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexander
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
-
Answer
Andrey
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexander
Answer
-
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
-
Answer
Alexey Vinogradov
Answer
Alexander S.
Answer
-
Alexander S.
Answer
Newbie
Answer
Newbie
Answer
-
Answer
Newbie
Answer
-
Alexander S.
Answer
-
Answer
Alexander S.
Answer
-
Answer
Alexander S.
Answer
-
Alexander S.
Answer
Alexey Vinogradov
Answer
Alexey Vinogradov
Answer
Leonid
Answer
Alexey Vinogradov
Answer
Victor
Answer
Alexander S.
Answer
Alexey Vinogradov
Apr 19, 2019Answer
- Socket 370 – the most common socket for Intel processors. It is with this that the era of dividing Intel processors into inexpensive Celeron solutions with a trimmed cache and Pentium – more expensive full versions of the company’s product – begins. The connector was installed on motherboards with a system bus from 60 to 133 MHz. The socket is made in the form of a square plastic movable box; when installing a processor with 370 contacts, a special plastic lever presses the processor legs to the connector contacts. Supported processors Intel Celeron Coppermine, Intel Celeron Tualatin, Intel Celeron Mendocino, Intel Pentium Tualatin, Intel Pentium Coppermine. Speed characteristics of installed processors from 300 to 1400 MHz. Supported third party processors. Produced since 1999.
- Socket 423 – the first connector for Pentium 4 processors. It had a 423-pin grid of legs and was used on motherboards of personal computers. It existed for less than a year, due to the inability of the processor to further increase in frequency, the processor could not pass the frequency of 2 GHz. Replaced by Socket 478 connector. Production began in 2000.
- Socket 478 - released in pursuit of the competitor (AMD company) Socket A, since previous processors were unable to raise the bar of 2 Gigahertz, and AMD took the lead in the processor production market. The connector supports Intel solutions - Intel Pentium 4, Intel Celeron, Celeron D, Intel Pentium 4 Extreme Edition. Speed characteristics from 1400 MHz to 3.4 GHz. Produced since 2000.
- Socket 754 was developed specifically for the Athlon 64 processor. The release of new processor sockets was associated with the need to replace the Athlon XP processor line, which was based on Socket A. The first processors of AMD K8 platforms were installed in Socket 754 processor sockets measuring 4 by 4 centimeters. This need was dictated by the fact that the Athlon 64 processors had a new bus and integrated memory controllers. The voltage output from this socket was 1.5 volts. Of course, the 754 became an intermediate stage in the development of the Athlon 64. The high cost and initial shortage of these processors did not make this platform very popular. And by the time the availability and cost of components had just returned to normal, AMD presented the release of a new socket - Socket 939. By the way, it was he who helped make the Athlon 64 a popular and truly affordable processor.
- Socket 775 or Socket T - the first connector for Intel processors without sockets, made in a square form factor with protruding contacts. The processor was installed on the protruding contacts, the pressure plate was lowered, and using a lever it was pressed against the contacts. Still used in many personal computers. Designed to work with almost all fourth generation Intel processors - Pentium 4, Pentium 4 Extreme Edition, Celeron D, Pentium Dual-Core, Pentium D, Core 2 Quad, Core 2 Duo and Xeon series processors. Produced since 2004. Speed characteristics of installed processors range from 1400 MHz to 3800 MHz.
- Socket 939 , containing 939 contacts with an extremely small diameter, making them quite soft. This is a "simplified" version of the previous Socket 940, usually used in high-performance computers and servers. The absence of one hole in the socket did not make it possible to install more expensive processors into it. This connector was considered very successful for its time, as it combined good capabilities, dual-channel memory access and low cost of both the socket itself and the controller on computer motherboards. These connectors were used for computers with conventional DDR memory. Immediately after the transition to DDR2 memory, they became obsolete and gave way to AM2 connectors. The next step is the invention of new DDR3 memory and new AM2+ and AM3 sockets designed for the next models of AMD quad-core processors.
- Socket AM2 (Socket M2), developed by AMD for certain types of desktop processors (Athlon-LE, Athlon 64, Athlon 64 FX, Athlon 64 X2, Sempron-LE and Sempron, Phenom X4 and Phenom X3, Opteron). It replaced Socket 939 and 754. Despite the fact that Socket M2 has 940 pins, this socket is not compatible with Socket 940, since the older version of Socket 940 cannot support dual-channel DDR2 RAM. The first processors to support Socket AM2 were single-core models Orleans (or the 64th Athlon) and Manila (Sempron), some dual-core Windsor (for example, Athlon 64, X2 FX) and Brisbane (AthlonX2 and Athlon 64X2). In addition, Socket AM2 includes Socket F, designed for servers, and a Socket S1 variant for various mobile computers. Socket AM2+ i is absolutely identical in appearance to the previous one, the only difference is the support for processors with Agena and Toliman cores.
- LGA 1156 socket – Made using 1156 protruding contacts. Produced since 2009. Designed for modern Intel processors for personal computers. Speed characteristics from 2.1 GHz and higher.
- LGA 1155 socket or Socket H2 - designed to replace the LGA 1156 socket. Supports the latest Sandy Bridge processor and the future Ivy Bridge. The connector is made in 1155-pin design. Produced since 2011. Speed characteristics up to 20 GB/s.
- Socket R (LGA2011) - Core i7 and Xeon with integrated quad-channel memory controller and two QuickPath connections. Replacement Socket B (LGA1366)
- Socket FM1 is AMD's platform for Llano processors and looks like a tempting proposition for those who love integrated systems.
- Socket H3 or LGA 1150 is a processor socket for Intel processors of the Haswell (and its successor Broadwell) microarchitecture, released in 2013. LGA 1150 is designed as a replacement for LGA 1155 (Socket H2). Made using LGA (Land Grid Array) technology. It is a connector with spring-loaded or soft contacts, to which the processor is pressed using a special holder with a grip and a lever. It is officially confirmed that the LGA 1150 socket will be used with Intel Q85, Q87, H87, Z87, B85 chipsets. The mounting holes for cooling systems on sockets 1150/1155/1156 are completely identical, which means full comprehensive compatibility and identical installation procedures for cooling systems for these sockets.
- Socket B2 (LGA1356) - Core i7 and Xeon with integrated three-channel memory controller and QuickPath connections. Replacement Socket B (LGA1366)
- FM2 connector - Processor socket for hybrid processors (APU) from AMD with the Piledriver core architecture: Trinity and Komodo, as well as the canceled Sepang and Terramar (MCM - multi-chip module). Structurally, it is a ZIF connector with 904 pins, which is designed for installing processors in PGA-type cases. The FM2 connector was introduced in 2012, just a year after the FM1 connector. Although socket FM2 is an evolution of socket FM1, it is not backward compatible with it. Trinity processors have up to 4 cores, Komodo and Sepang server chips have up to 10, and Terramar have up to 20 cores.
- LGA 1151 socket - a socket for Intel processors that supports Skylake architecture processors. LGA 1151 is designed as a replacement for LGA 1150 (also known as Socket H3). LGA 1151 has 1151 spring-loaded contacts to contact the processor pads. According to rumors and leaked Intel advertising documentation, motherboards with this socket will feature DDR4 memory support. All Skylake architecture chipsets support Intel Rapid Storage Technology, Intel Clear Video Technology and Intel Wireless Display Technology (when supported by the processor). Most motherboards support various video outputs (VGA, DVI or - depending on the model).
- LGA 2066 (Socket R4) is a socket for Intel processors that supports Skylake-X and Kaby Lake-X processors without an integrated graphics core. Designed to replace the LGA 2011/2011-3 (Socket R/R3) socket for high-end Basin Falls desktops (X299 chipset), while the LGA 3647 (Socket P) will replace the LGA 2011-1/2011- 3 (Socket R2/R3) in server platforms based on Skylake-EX (Xeon “Purley”).
- AM4 (PGA or µOPGA1331) is a socket produced by AMD for microprocessors with Zen microarchitecture (Ryzen brand) and subsequent ones. The connector is a PGA (pin grid array) type and has 1331 contacts. It will be the company's first socket with support for the DDR4 memory standard and will be a single socket for both high-performance processors without an integrated video core (currently using Socket AM3+), and for low-cost processors and APUs (previously using various sockets of the AM / FM series).
- Socket TR4 (Socket Ryzen Threadripper 4, also Socket SP3r2) is a type of connector from AMD for the Ryzen Threadripper family of microprocessors, introduced on August 10, 2017. Physically very close to the AMD Socket SP3 server connector, however, it is incompatible with it. Socket TR4 became the first LGA-type socket for consumer products (previously LGA was used in the server segment, and processors for home computers were produced in FC-PGA packages). It uses a complex multi-stage process of mounting the processor into the socket using special holding frames: an internal one, secured with latches to the cover of the chip case, and an external one, secured with screws to the socket. Journalists note the very large physical size of the connector and socket, calling it the largest format for consumer processors. Due to its size, it requires specialized cooling systems that can handle up to 180W. The socket supports HEDT (High-End Desktop) segment processors with 8-16 cores and provides the ability to connect RAM via 4 DDR4 SDRAM channels. The socket has 64 generation 3 PCIexpress lanes (4 are used for the chipset), several 3.1 and SATA channels
- The fastest LGA1156 processor equipped with a graphics core (by the way, it costs the same as Core i7-2600)
- The fastest Core i5 of the previous generation (has the same starting frequency as the new Core i5-2300, and the selling price is the same as the Core i5-2500)
- Fastest Core i7 for LGA1156
- The fastest quad-core x86 processor
- Generally the fastest x86 processor :)
Minsk Repairman
Answer
BRedScorpius
Answer
aleksandrzdor
Answer
Dmitriy
Answer
Answer
Answer
Answer
Answer
Leonid
Answer
Leonid
Answer
Sergey
Answer
Answer
Stanislav
Answer
Vladislav
Answer
Answer
Alexander
Answer
Alexander
Answer
Igor Novozhilov
Answer
Answer
Answer
Answer
Answer
Alexander S.
Answer
Alexey Vinogradov
Answer
Answer
Answer
Answer
Answer
Alexander S.
Answer
Answer
Alexander S.
Answer
Answer
Vyacheslav
Answer
Dmitriy
Answer
Answer
Konstantin
Answer
Vitaly
Answer
Answer
Dmitriy
Answer
Answer
Alexander S.
Answer
Answer
Leonid
Answer
Answer
Vladimir
Answer
Answer
Seryoga
Answer
Answer
Leonid
Answer
Natalia
Answer
Andrey
Answer
Andrey
Answer
Andrey
Answer
Andrey
Answer
Andrey
Answer
Andrey
Answer
Andrey
Answer
Andrey
Answer
Alexander S.
Answer
Andrey
Answer
Alexander S.
Answer
Andrey
Answer
Alexander S.
Answer
Andrey
Answer
Alexander S.
Answer
Andrey
Answer
Andrey
Answer
Alexander
Answer
Answer
Answer
Maksim
Answer
Andrey
Answer
Andrey
Answer
Andrey
Answer
Answer
Alexey Vinogradov
Answer
Dmitriy
Answer
Maksim
Answer
Alexander
Answer
Alexander S.
Answer
Answer
Answer
Dmitriy
Answer
Answer
Answer
Alexander S.
Answer
Answer
Answer
Answer
little ducalis
Answer
Newbie
Answer
Answer
Konstantin
Answer
Answer
Iskandar
Answer
Answer
Answer
Vladimir
Answer
Answer
Andrey
Answer
Answer
Answer
Sergey
Answer
Leonid
Answer
Victor
Answer
Tatiana
Jan 04, 2019
Answer
Victor
Apr 19, 2019
Answer
A
Jul 12, 2019
Socket (colloquial - socket) of the central processor is a connector located on the computer motherboard to which the central processor is connected. The processor, before it is installed on the motherboard, must fit the socket. It is very easy to understand what a processor socket is, if you remember that the latter is a microcircuit, only of relatively large size. The socket is located on the motherboard and looks like a low rectangular structure with many holes, the number of which corresponds to the processor legs. To securely fix the inserted microcircuit in the socket, a specially designed mechanical latch is used. Note that Intel, unlike AMD, has recently been using a different principle of connecting the processor and board.
Sometimes on forums the question is asked about which socket to choose. In fact, you should first select a processor, and then a board with the appropriate socket for it. However, one important point must be taken into account. Intel is famous for the fact that often each new generation of processors involves the use of a new socket. This may lead to the fact that a recently purchased computer based on a processor from this company will be difficult to upgrade in a few years due to the incompatibility of the installed microprocessor and new ones offered on the market. AMD has a more loyal attitude towards customers: changing sockets occurs more slowly, and backward compatibility is usually maintained. Although, times are changing.
| Type | Purpose | Number of contacts | Year of issue |
| PIN DIP | 8086/8088, 65С02 | 40 | 1970 |
| CLCC | Intel 80186, 80286, 80386 | 68 | 1980 |
| PLCC | Intel 80186, 80286, 80386 | 68 | 1980 |
| Socket 80386 | Intel 386 | 132 | 1980 |
| Socket 486/Socket 0 | Intel 486 | 168 | 1980 |
| Motorola 68030 | Motorola 68030, 68LC030 | 128 | 1987 |
| Socket 1 | Intel 486 | 169 | 1989 |
| Type | Purpose | Number of contacts | Year of issue |
| Socket 2 | Intel 486 | 238 | 1989 |
| Motorola 68040 | 68040 | 179 | 1990 |
| Socket 3 | Intel 486, 5x86 | 237 | 1991 |
| Socket 4 | Pentium | 273 | 1993 |
| Type | Purpose | Number of contacts | Year of issue |
| Socket 5 | Intel 486 | 238 | 1994 |
| Socket 463 NexGen | Nx586 | 463 | 1994 |
| Motorola 68060 | 68060, 68l0C60 | 206 | 1994 |
| Socket 7 | Pentium, AMD K5, K6 | 321 | 1995(Intel), 1998(AMD) |
| Type | Purpose | Number of contacts | Year of issue |
| Socket 499 | DEC EV5 21164 | 499 | 1995 |
| Socket 8 | Pentium / Pentium 2 | 387 | 1955 |
| Socket 587 | DEC EV5 21164A | 587 | 1996 |
| Mini-Cartridge | Pentium 2 | 240 | 1997 |
| MMC-1 Mobile Module Connector | Pentium 2, Celeron | 280 | 1997 |
| Apple G3/G4/G5 | G3/G4/G5 | 300 | 1997 |
| MMC-2 Mobile Module Connector | Pentium 2.3, Celeron | 400 | 1998 |
| Type | Purpose | Number of contacts | Year of issue |
| G3/G4 ZIF | Power PC G3 G4 | 288 | 1996 |
| Socket 370 | Pentium 3, Celeron, Cyrix, Via C3 | 370 | 1999 |
| Socket A/Socket 462 | AMD Athlon, Duron, MP, Sempron | 462 | 2000 |
| Socket 423 | Pentium 4 | 423 | 2000 |
| Type | Purpose | Number of contacts | Year of issue |
| Socket 478 / Socket N / Socket P | Intel 486 | 238 | 1994 |
| Socket 495/MicroPGA 2 | Mobile Celeron/Pentium 3 | 495 | 2000 |
| PAC 418 | Intel Itanium | 418 | 2001 |
| Socket 603 | Intel Xeon | 603 | 2001 |
| PAC 611 / Socket 700 / mPGA 700 | Intel Itanium 2, HP8800, 8900 | 611 | 2002 |
| Type | Purpose | Number of contacts | Year of issue |
| Socket 604/S1 | Intel 486 | 238 | 2002 |
| Socket 754 | Athlon 64, Sempron, Turion 64 | 754 | 2003 |
| Socket 940 | Opteron 2, Athon 64FX | 940 | 2003 |
| Socket 479/mPGA479M | Pentium M, Celeron M, Via C7-M | 479 | 2003 |
| Socket 478v2/mPGA478C | Pentium4, Pentium Mobile, Celeron, Core | 478 | 2003 |
| Type | Purpose | Number of contacts | Year of issue |
| Socket 939 | Intel 486 | 939 | 2004 |
| LGA 775/Socket T | Pentium4, Celeron D, Core 2, Xeon | 775 | 2004 |
| Socket 563 / Socket A / Compact | Mobile Athon XP-M | 563 | 2004 |
| Socket M/mPGA478MT | Celeron, Core, Core 2 | 478 | 2006 |
| LGA771/Socket J | Xeon | 771 | 2006 |
Socket A. This connector is known as Socket 462 and is a socket for processors from Athlon Thunderbird to Athlon XP/MP 3200+, as well as for AMD processors such as Sempron and Duron. The design is made in the form of a ZIF socket with 453 working contacts (9 contacts are blocked, but despite this, the number 462 is used in the name). The system bus for Sempron, XP Athlon has a frequency of 133 MHz, 166 MHz and 200 MHz. The weight of coolers for Socket A, recommended by AMD, should not exceed 300 grams. The use of heavier coolers can lead to mechanical damage and even lead to failure of the processor power system. Processors with a frequency of 600 MHz (for example, Duron) and up to 2300 MHz (meaning the Athlon XP 3400+, which never went on sale) are supported.
| Type | Purpose | Number of contacts | Year of issue |
| Socket S1 | Athon Mobile, Sempron, Turion 64/X2 | 638 | 2006 |
| Socket AM2/AM2+ | Athon 64/FX/FX2, Sempron, Phenom | 940 | 2007 |
| Socket F/ Socket L/Socket 1207FX | Athon 64FX, Opteron | 1207 | 2006 |
| Socket/LGA 1366 | ,Xeon | 1366 | 2008 |
| rPGA988A/Socket Q1 | Core i3/i5/i7, Pentium, Celeron | 988 | 2009 |
LGA 1366 socket – Made in 1366 contact form, produced since 2008. Supports Intel processors – Core i7 series 9xx, Xeon series 35xx to 56xx, Celeron P1053. WITH speed characteristics from 1600 MHz to 3500 MHz. Core i7 and Xeon (35xx, 36xx, 55xx, 56xx series) with integrated three-channel memory controller and QuickPath connection. Replacement of Socket T and Socket J (2008)
| Type | Purpose | Number of contacts | Year of issue |
| Socket AM3 | AMD Phenom, athlon, Sempron | 941 | 2009 |
| Socket G/989/rPGA | G1/G2 | 989 | 2009 |
| Socket H1/LGA1156/a/b/n | Core i3/i5/i7, Pentium, Celeron, Xeon | 1156 | 2009 |
| Socket G34/LGA 1944 | Opteron 6000 series | 1944 | 2010 |
| Socket C32 | Opteron 4000 series | 1207 | 2010 |
| Type | Purpose | Number of contacts | Year of issue |
| LGA 1248 | Intel Itanium 9300/9600 | 1248 | 2010 |
| Socket LS/LGA 1567 | Intel Xeon 6500/7500 | 1567 | 2010 |
| Socket H2/LGA 1155 | Intel Sandy Bridge, Ivy Bridge | 1155 | 2011 |
| LGA 2011/Socket R | Intel Core i7, Xeon | 2011 | 2011 |
| Socket G2/rPGA988B | Intel Core i3/i5/i7 | 988 | 2011 |
| Type | Purpose | Number of contacts | Year of issue |
| Socket FM1 | AMD Liano/Athlon3 | 905 | 2011 |
| Socket AM3 | AMD Phenom/Athlon/Semron | 941 | 2011 |
| Socket AM3+ | Amd Phenom 2 Athlon 2 / Opteron 3000 | 942 | 2011 |
| Socket G2/rPGA989B | Intel Core i3/i5/i7, Celeron | 989 | 2011 |
| Socket FS1 | AMD Liano/Trinity/Richard | 722 | 2011 |
Socket AM3 is a processor socket for a desktop processor, which is a further development of the Socket AM2+ model. This connector has support for DDR3 memory, as well as higher speeds for HyperTransport buses. The first processors to use this socket were the Phenom II X3 710-20 and Phenom II X4 models 805, 910 and 810.
Socket AM3 + (Socket 942) is a modification of Socket AM3, developed for processors codenamed “Zambezi” (microarchitecture - Bulldozer). Some socket AM3 motherboards will allow you to update the BIOS to use socket AM3+ processors. But when using AM3+ processors on AM3 motherboards, it may not be possible to obtain data from the temperature sensor on the processor. Also, the power saving mode may not work due to the lack of support for fast core voltage switching in the Socket AM3 version. The AM3+ socket on motherboards is black, while the AM3 is white. The diameter of the holes for the pins of processors with Socket AM3 + exceeds the diameter of the holes for the pins of processors with Socket AM3 - 0.51 mm versus the previous 0.45 mm.
| Type | Purpose | Number of contacts | Year of issue |
| LGA 1356/Socket B2 | Intel Sandy Bridge | 1356 | 2012 |
| Socket FM2 | AMD Trinity/athlon X2/X4 | 904 | 2012 |
| Socket H3/LGA 1150 | Intel Haswell/Broadwell | 1150 | 2013 |
| Socket G3/rPGA 946B/947 | Intel Haswell/Broadwell | 947 | 2013 |
| Socket FM2/FM2b | AMD Kaveri/Godvari | 906 | 2014 |
| Type | Purpose | Number of contacts | Year of issue |
| LGA 2011-3 / LGA 2011 v3 | Intel Haswell, haswell-EP | 2011 | 2014 |
| Socket AM1/FS1b | AMD Athlon/Semron | 721 | 2014 |
| LGA 2011-3 | Intel Haswell / Xeon / haswell-EP / ivy Bridge EX | 2083 | 2014 |
| LGA 1151/Socket H4 | Intel Skylake | 1151 | 2015 |
| Type | Purpose | Number of contacts | Year of issue |
| LGA 2066 Socket R4 | Intel Skylake-X/Kabylake-X i3/i5/i7 | 2066 | 2017 |
| Socket TR4 | AMD Ryzen Threadripper | 4094 | 2017 |
| Socket AM4 | AMD Ryzen 3/5/7 | 1331 | 2017 |
Leave your comment!
Finally, the long-awaited moment for many has arrived when you can get acquainted with the performance of Intel processors for the new LGA1155 platform! True, like last year, it fell right on the holidays, but nothing - having recovered from the holiday, it’s all the more interesting to go to the store :) By the way, it’s not only the date that today’s event has in common with the announcement of processors based on the Clarkdale core a year ago. The fact is that the story with LGA1156 is essentially repeating itself - the announcement of new processors will be extended over several stages. Today we will learn all the details about the quad-core models of the Sandy Bridge architecture, but we will have to wait almost another month and a half for more affordable dual-core models. “People’s” Pentiums don’t even make it into the first quarter.
But still, one and a half is not four, there will be much more than one Pentium, the prices for them are expected to be more humane than for the only processor (well, one and a half) of this family for LGA1156, and Celeron is also visible on the horizon: in a word, the company took into account experience “Extended start” of LGA1156 and similar mistakes will most likely not be made. Thus, LGA1155, starting somewhere in the second or third quarter of this year, will finally make it possible to abolish the well-worn LGA775 design, and by the end of the year it will do away with LGA1156. But for some time these three platforms will exist in parallel, which, coupled with the surviving LGA1366 (and it will definitely live until the end of the year), will only increase the confusion in the market. However, these are the harsh realities of the modern market and it is unlikely that we can change them in any way. All you have to do is study everything carefully and always make the right choice :)
There will be no theoretical part today. The fact is that we already had materials on this topic, and more detailed studies of microarchitecture are just around the corner. In general, let's not beat theorists' bread :) Also, for now, let's leave behind the scenes the issue of performance and functionality of the graphics core - this is also a separate and serious topic, which we will return to in the near future for a detailed study. At the moment, the main thing is to study the performance of the processor part itself and compare it with competing products from both Intel and AMD. What do we suggest and move on to?
Test bench configuration
| CPU | Core i5-2300 | Core i5-2400 | Core i5-2500/2500K | Core i7-2600/2600K |
| Kernel name | Sandy Bridge | Sandy Bridge | Sandy Bridge | Sandy Bridge |
| Production technology | 32 nm | 32 nm | 32 nm | 32 nm |
| Core frequency (std/max), GHz | 2,8/3,1 | 3,1/3,4 | 3,3/3,7 | 3,4/3,8 |
| 28 | 31 | 33 | 34 | |
| How Turbo Boost works | 3-2-2-1 | 3-2-2-1 | 4-3-2-1 | 4-3-2-1 |
| 4/4 | 4/4 | 4/4 | 4/8 | |
| L1 cache, I/D, KB | 32/32 | 32/32 | 32/32 | 32/32 |
| L2 cache, KB | 4×256 | 4×256 | 4×256 | 4×256 |
| L3 cache, MiB | 6 | 6 | 6 | 8 |
| RAM | 2×DDR3-1333 | 2×DDR3-1333 | 2×DDR3-1333 | 2×DDR3-1333 |
| GMA HD graphics core | 2000 | 2000 | 2000/3000 | 2000/3000 |
| Graphics core frequency (max), MHz | 1100 | 1100 | 1100 | 1350 |
| Socket | LGA1155 | LGA1155 | LGA1155 | LGA1155 |
| TDP | 95 W | 95 W | 95 W | 95 W |
| Price | $275() | $236() | $229()/N/A() | $340()/N/A() |
| Wholesale price at the time of announcement | $177 | $184 | $205/$216 | $294/$317 |
In the family of processors for LGA1156, two processors of the Core i7 line and only one Core i5 appeared at first, but now the ratio is the opposite - one to three. The explanation is simple: older Core i7-800s are still on the market and have adequate performance, so you shouldn’t interfere with them too much. But the Core i5 is too much of a motley bunch, which includes fast but graphics-less 700-series processors and graphics-equipped but rather weak (due to only two cores) Core i5-600. It was this imbalance that Intel decided to eliminate first. Note that now Core i5 is Always four cores, and the “old” version of “two cores/four threads” is present only in the cheaper Core i3 family. But these processors will come out a little later, fortunately now the Core i3-500 is not so bad.
What does a comparison of technical characteristics show? If previously the Core i5-700 and Core i7-800 differed only in the presence/absence of Hyper-Threading support and frequencies, now the differences have become a little deeper: the i5 also has less cache memory. Moreover, the line is constructed in an interesting way - the step of the starting clock frequencies is uneven, but in terms of the maximum frequency in the boost mode, “everything is as it should”: a hundred in the index is equal to 300 MHz clock frequency. A very serious difference, since both Intel and AMD have already accustomed us to the fact that neighboring processors in the line differ only by one multiplier. It’s difficult to say for now whether the concept will be preserved in the future or whether the company will begin to tighten its ranks, so we will postpone this issue for the future. In our opinion, “not partial” is very useful - there are already too many processors on the market, in which it is too easy to get confused. But there may be some progress - otherwise the Core i5-2300 looks a little strange, its price is only slightly lower than that of the 2400, but the lag in clock frequency is greater than the difference between the older models. Is that in one or two-threaded applications it is reduced, but there are fewer and fewer of them. Moreover, the presence of background processes, which sometimes require quite a bit of computing resources, also “votes” in favor of multithreading (and some of these background applications have also become multithreaded).
But somehow with the Turbo Boost mode itself... More was expected. And the maximum increase decreased to 400 MHz (do not forget that one “new” step is equal to 3/4 of the old one), and the dependence on the number of working cores has not gone away, although there were rumors that it was now possible to increase the frequency of all cores to the maximum. The only significant change is that now processors have the right to overclock “to the last”: the boost mode is allowed up to the TDP level (previously it was turned off at a lower limit), and, if necessary, for a short time - even higher. Thus, a certain increase in performance under heavy load should be observed. Let's check which one.
The main thing for overclocking enthusiasts is that Turbo Boost in its new incarnation also supports such a function as “Limited Unlocked Core” - the ability to set multipliers to the value “Max Turbo +4”. That is, in other words, according to Intel documentation, a completely ordinary Core i7-2500 will be able to operate at a frequency of 3.9 GHz when all cores are loaded, and when only one is loaded, it will even reach a frequency of 4.1 GHz! The reality turned out to be even more interesting - the Gigabyte board, on which we tested the new family, of course limited the multipliers, but... But for 2600, for example, the maximum value (namely 42) could be set for any number of active cores, i.e., with a slight movement of the hand, a processor with a clock frequency of 3.4 GHz turns into a model with a frequency of 4.2 GHz. And we have strong suspicions that other boards based on the P67 chipset (with the possible exception of those produced by Intel itself) will behave in the same way.
P67 boards also support “Fully Unlocked Core,” which allows you to use a multiplier of up to 57 in any mode. However, this requires a K-series processor. Note that they are of interest not only to overclocking enthusiasts (and maybe not so much to them: as shown above, you can add 700-800 MHz on conventional processors): the K-series uses an HD 3000 series video core, but in conventional models - only only HD 2000, in which half of the executive modules are disabled. Thus, these processors will also be extremely useful for fans of integrated graphics, who will use them on boards based on the H67 chipset. But on the P67 it will not be possible to use the built-in video core (since it does not have an FDI link), but it will be possible to fully “have fun” during overclocking, as mentioned above. Moreover, when overclocking not only the cores, but also the memory: despite the fact that the officially supported maximum mode is DDR3-1333, this is only true for H67. On the P67, higher multipliers are also available, which gives memory frequencies up to 2133 MHz. And the TDP level on these boards can be adjusted manually, increasing it during overclocking or, conversely, decreasing it to save energy (which was previously only available for extreme processors). In general, when developing processors and chipsets for LGA1155, Intel took into account all past experience, putting things in order in their comparative positioning :)
| CPU | Core i5-680 | Core i5-760 | Core i7-880 | Core i7-975 Extreme | Core i7-980X Extreme |
| Kernel name | Clarkdale | Lynnfield | Lynnfield | Bloomfield | Gulftown |
| Production technology | 32/45 nm | 45 nm | 45 nm | 45 nm | 32 nm |
| Core frequency (std/max), GHz | 3,6/3,87 | 2,8/3,33 | 3,06/3,73 | 3,33/3,6 | 3,33/3,6 |
| Starting multiplication factor | 27 | 21 | 23 | 25 | 25 |
| How Turbo Boost works | 2-1 | 4-4-1-1 | 5-4-2-2 | 2-1-1-1 | 2-1-1-1-1-1 |
| Number of cores/threads | 2/4 | 4/4 | 4/8 | 4/8 | 6/12 |
| L1 cache, I/D, KB | 32/32 | 32/32 | 32/32 | 32/32 | 32/32 |
| L2 cache, KB | 2×256 | 4×256 | 4×256 | 4×256 | 6×256 |
| L3 cache, MiB | 4 | 8 | 8 | 8 | 12 |
| UnCore frequency, GHz | 2,4 | 2,13 | 2,4 | 2,66 | 2,66 |
| RAM | 2×DDR3-1333 | 2×DDR3-1333 | 2×DDR3-1333 | 3×DDR3-1066 | 3×DDR3-1066 |
| 733 | - | - | - | - | |
| Socket | LGA1156 | LGA1156 | LGA1156 | LGA1366 | LGA1366 |
| TDP | 73 W | 95 W | 95 W | 130 W | 130 W |
| Price | N/A() | N/A() | N/A() | N/A() | N/A() |
As expected when testing a new family of processors, there will be more competitors than test subjects. Especially competitors produced in the same factories. The company of Intel processors we selected at first glance looks too diverse, but the selection logic is simple - the table (from left to right) shows:
The last two models, of course, we need mainly out of curiosity - any processor announced today for LGA1155 is not ashamed to lose to them :) However, there are serious suspicions that the Core i7-2600 will not be able to lose to the “extreme” i7-975 Extreme (how no matter how hard he tries), but a comparison with the i7-980X on a wide range of applications is of considerable interest.
| CPU | Phenom II X4 970 | Phenom II X6 1090T |
| Kernel name | Deneb | Thuban |
| Production technology | 45 nm | 45 nm |
| Core frequency (std/max), GHz | 3,5 | 3,2/3,6 |
| Starting multiplication factor | 17,5 | 16 |
| Turbo CORE operation diagram | - | 3-3-3-0-0-0 |
| Number of cores/threads | 4/4 | 6/6 |
| L1 cache, I/D, KB | 64/64 | 64/64 |
| L2 cache, KB | 4×512 | 6×512 |
| L3 cache, MiB | 6 | 6 |
| UnCore frequency, GHz | 2,0 | 2,0 |
| RAM | 2×DDR3-1333 | 2×DDR3-1333 |
| Graphics core frequency, MHz | - | - |
| Socket | AM3 | AM3 |
| TDP | 125 W | 125 W |
| Price | N/A(0) | N/A(0) |
Now let's move on to AMD. It is obvious that when the heavy equipment of the “blues” enters the battlefield, the “greens” are left only with guerrilla warfare and ambush operations. In any case, this situation will last until Superwaffe, codenamed “Bulldozer,” rolls out of the laboratories, but there is quite a lot of time left until that moment. Today we will not touch the “green partisans”, in the form of hordes of various Athlon IIs, but we will look at a couple of “tank ambushes”. The first will be the Phenom II X4 970, already familiar to our readers, a processor with the highest guaranteed clock frequency of the quad-core ones on the market (Core i7-2600 reaches 3.5 GHz only in boost mode, while others are not capable of this). The second one is the Phenom II X6 1090T. The entry of this line into the market last spring allowed the company to once again return to the “$200-300” market segment, since the processors very successfully occupied a niche between the older Core i5 and the younger Core i7 - let’s see if they will be able to maintain their positions taking into account the update of the product range Intel. To be fair, both the X4 and X6 families are expected to be replenished in the near future (more precisely, the 1100T appeared at the end of last year, and the 975 now), but since we are talking about only a slight increase in clock frequency, it is obvious that the qualitative picture is a little more productive than used, Phenom II will not change.
| Motherboard | RAM | |
| LGA1155 | Gigabyte P67A-UD5 (P67) | |
| LGA1156 | Gigabyte P55A-UD6 (P55) | Kingston KVR1333D3N9K3/6G (2×1333; 9-9-9-24) |
| LGA1366 | Intel DX58SO (X58) | Kingston KVR1333D3N9K3/6G (3×1333; 9-9-9-24) |
| AM3 | Gigabyte 890FXA-UD7 (AMD 890FX) | Corsair CM3X2G1600C9DHX (2×1333; 7-7-7-20-1T, Unganged Mode) |
Testing
The performance testing methodology (list of software used and testing conditions) is described in detail in a separate article. For ease of perception, the results in the diagrams are presented as percentages (the result of the AMD Athlon II X4 620 in each test is taken as 100%). Detailed results in absolute values are available as a table in Microsoft Excel format.
3D visualization
The very first group of programs - and the first discoveries. As we already know, these tasks do not require a large number of computation threads, so what comes first is the speed with which these same threads (two or three) are “run” through the processor. That is, in other words, this is exactly the area where architectural optimizations can have the best impact. And they had an effect - the Core i5-2300 (the youngest and cheapest) has already outperformed all the processors that we tested earlier. Including the extreme Core i7-975, which no one has ever managed to beat in this test. The rest of the representatives of the new architecture, for obvious reasons, are even faster, so they simply have no one to compete with.
3D rendering
It seems to us that Sandy Bridge will have the last word in these tasks when programs begin to support the new set of AVX vector instructions. In the meantime, this is “pure” mathematics, and it is very well parallelized, so the more computation threads, the better: strength breaks straw. However, the high efficiency of each computation thread is reflected here as well. In particular, the new Core i5 is faster than the old ones with the same number of cores and at a comparable clock frequency by 10 percent (looking at the diagram, do not forget that the i5-760 in boost mode operates at a frequency of 2.93 GHz, and the i5-2300 - only 2.9 GHz). But the transition to a thinner technical process allows the new processors to operate at higher frequencies, as a result of which they can compete with both the old Core i7 and the six-core Phenom II X6. Moreover, with the latter - even despite their higher frequency;) However, miracles do not happen in the world, so six-core Core i7s are out of reach, but they are much more expensive. Therefore, the second place of the Core i7-2600 is in fact not a defeat, but a brilliant victory.
Scientific and engineering computing
Another basically low-thread group with small multi-threaded inclusions, which distinguishes it from the first. But not much - the first two places were taken by processors for LGA1155 (the first was shared by as many as two, which once again shows that Hyper-Threading technology is still far from “free”), and the “penny” Core i5-2300 was second only to “multi-ruble” ones extreme processors of previous families.
Graphic editor
As we have written more than once, the applications included in this group have very different preferences: Adobe Photoshop “loves” many computation threads, but the three “amateur” programs do not need them (and even sometimes interfere). Well, since there are three of them for one, it is not surprising that previously the dual-core (but high-frequency) Core i5-600 showed very good overall results. Only extreme sports, where there are a lot of cores and high frequencies, produced more. “Family 2000” suits these programs even better, and its results in Photoshop are very good - here are the new leaders. I was especially shocked by the Core i7-2600, which in the Adobe software package almost caught up with the much more expensive six-core Core i7-970, and in the three remaining applications it simply has no competitors. The Core i5-2400 in them showed similar performance to the Core i5-680 (previously the leader), but outperformed it in Photoshop by almost one and a half times, which allowed this inexpensive model to take its place among the former leaders in terms of total results. The Core i5-2500 is understandably faster than them and is only behind the Core i7-2600. In general, only the youngest Core i5-2300 did not blow our minds. Although if you remember that its wholesale price is only $177, and it didn’t “shock” against the background of processors worth a whole hundred (or even all four - if you remember how much the Core i7-880 costs, to which the “baby” from the new line somewhat closer than the equivalent-frequency Core i5-760) dollars more expensive, this is also just a wonderful result.
Archivers
7-Zip is capable of using as many cores as it can find, all three subtests are very fond of a large amount of cache memory, and the latter seems to be only interested in it - in general, it’s not surprising that the new Core i5 performed not so perfectly here , as in previous groups: only four threads and a cache reduced to 6 MB make themselves felt. But “not ideal” does not mean bad - they easily outperformed all AMD processors and managed to reach approximately the level of the old Core i7, which cost about a hundred more. But the new Core i7-2600 has support for Hyper-Threading and an 8 MB cache, so its only competitor is the extreme Core i7-980X (even 975 is slower).
Compilation
Visual Studio turned out to be not the most friendly application to new processors - apparently due to the fact that the compilation task was already one of the best optimized ones. However, the Core i5-2300 slightly beats the Core i5-760: taking into account the smaller cache memory capacity (and it is of considerable importance in this test) of the new product, this deserves a positive assessment. The increase (albeit small) is actually of strategic importance - as we remember, earlier in this program the Phenom II X6 was very good, located above the Core i5 and with older models reaching up to the younger Core i7. And now? And now with the compilation quad core(and “honest” - without any Hyper-Threading) Core i5-2400 copes at exactly the same speed as six-core Phenom II X6 1055T (albeit the youngest in the family, but more expensive)! And the next model with the 1075T index is not far behind, beating the Core i5-2500 by only one point. The older models, as we see, are still faster than even the new Core i5 and can already be compared with the old Intel processor at $294, but the new one for the same money has jumped far ahead and is only behind six-core processors Intel itself. Moreover, it cannot be said that it is very noticeable - only 10% separates it from the current extreme Core i7-980X.
Java
But SPECjvm surprised me a little, since we are already accustomed to citing this test as a good example of multi-core optimization. However, apparently, its capabilities extend to an area with eight to ten threads, but no more. While processors with different numbers of cores, but based on similar architectures, were competing, this gave obvious priority to more multi-threaded models, but as soon as we started comparing models with different efficiency per thread... In general, the Core i7-980X is still the fastest, but the superiority over the Core i7-2600 has become purely formal. Well, the Core i5-2400 somehow “didn’t notice” that the Core i7-880 supports twice as many computation threads and has a similar clock frequency, and almost caught up with it :)
Such an increase turned into a complete defeat for AMD processors - previously the Phenom II X4 970 was faster than all Core i5s, and the Phenom II X6 1090T outperformed any Core i7-800. Now the Phenom II X4 970 slower all Core i5 for LGA1155, and Phenom II X6 1090T lags behind Core i5-2500. And it’s not surprising that with the new Core i7 for LGA1155, AMD’s six-core processors, in principle, can no longer compete in performance.
Internet browsers
Previously, this group of applications was the most loyal to the Phenom II X4, since even the model with index 965 outperformed all Intel processors. Now, as we see, even the Core i5-2300 can repeat the results of the former tops, the Core i5-2400 overtakes the Phenom II X4 965 and is only slightly short of 970, and the 2500 and 2600 are simply the fastest on the market. Without any reservations :) However, as we have said more than once, from a practical point of view it makes no sense to attach great importance to the results of these tests on top processors, but from a research point of view, we tick off the fact that perhaps the last group has disappeared, where AMD processors held the lead.
Audio encoding
Another group of applications that over time could benefit a lot from the introduction of AVX, but for now operates only with “old” code. In addition, as has been said more than once, testing conditions most favor processors that can simultaneously perform a large number of computation threads. Therefore, at first glance, the new Core i5s are not that good here. But if you look closely, it becomes obvious that this is the level of the “old” Core i7 or Phenom II X6, i.e., more expensive CPUs. In any case, previously not a single quad-core crystal here scored 150 points, but now three at once score even more. The Core i7-2600, as one would expect, takes an honorable second place, behind only the six-core (and twelve-thread) Core i7-980X.
Video encoding
A similar picture to the previous one. Only now the gap between the 2600 and the 980X has become greater, but it is possible - after all, the devices are of completely different price classes. The main thing is that new devices are capable of defeating not only direct competitors, but also processors located one step higher.
Games
Even this group of applications has ended the stagnation. After which we began to run into a video card that was far from the slowest - for example, in Stalker and Resident Evil 5, all new processors showed the same results :) Which, it should be noted, turned out to be much higher than all the old ones. In general, the question of finding the best gaming processor should probably be considered resolved in all cases where you can spend more than $150 on a purchase - such is the Core i5-2300. Or, if finances are not so bad, then the Core i5-2400, which costs not much more, but “keeps up” at the level of former extreme sports enthusiasts. Top-end video cards or multi-GPUs remain behind the scenes, but here, it seems to us, the issue of processor price is not decisive. Moreover, even the Core i7-2600 is not too expensive. And you can overclock it to 400-800 MHz if desired... Or pay just a little extra for 2600K and overclock it even more. Or save a hundred and do the same procedure with the Core i5-2500K :) In general, the question of choice will only be faced by those who need a fast processor for games for $100 or who, on principle, want to take something very expensive.
Total
There was a time when older Phenom II X4 models sold for about $300, but the emergence of the Core i5-750 “driven” all AMD processors into the “under $200” price niche. The company was able to get out of it only by releasing the Phenom II X6. Now, it seems, history is repeating itself: even six-core Phenom IIs need to be sold at prices not exceeding $200 - to the delight of some fans, but to the horror of shareholders. (It is obvious that quad-core processors produced using the 32 nm process technology are cheaper to manufacture than six-core processors at 45 nm, despite the fact that the former have a video core.) So it will be interesting to see how the “green” ones will get out of this situation - before the release of Bulldozer- there is still quite a lot of time left.
Another family of processors was much more unlucky. Yes, in fact, the Core i5-600 can be consigned to the dustbin of history in its entirety. While it was necessary to make a choice: “four cores or integrated graphics?”, there was something to talk about. However, now the choice is obvious - four cores (faster than the old ones) And integrated graphics (faster than old ones) simultaneously. The new Core i5s are definitely better than the old ones. The current pricing policy looks a little strange: the 2400 differs from the 2300 by as much as 300 MHz and only 7 dollars, and from the 2500 by only 200 MHz and as much as 20 dollars, but this is quite understandable due to the premium for steepness. Moreover, perhaps, after the release of the new i3 (which will finally write off all processors based on the Clarkdale core) the “ladder” will be converted into 155-177-204, which will be more logical.
If the new i5 turned out to be so good, then what can we say about the Core i7-2600? An excellent processor, the absolute triumph of which was spoiled only by the extreme Core i7-980X. But even then only in the overall standings - it’s easy to notice that in half of the test groups even this expensive device can now compete only with the new Core i5, significantly getting ahead only in a few cases. Yes, this is still the difficult share of six-core processors in a desktop environment: an extremely small percentage of software can make good use of their potential capabilities. Intel, it seems to us, very correctly decided that the time for multi-core processors on the desktop has already come, but “many” still means “four”. For extreme sports enthusiasts, more is possible, but only if they are willing to pay for it :) And pay regularly - previously the same 980X competed only with the same extreme models, but now it does not always outperform budget ones. And the previous extreme one lost miserably everywhere to the regular Core i7-2600. Top, but ordinary. In general, it is standard practice for Intel - the new family of processors is unconditionally better than the old one, and the older models in it are no worse than the old extreme ones. Moreover, what is gratifying is that even fans of overclocking and other optimizations now do not have to prepare another thousand dollars: there are not so expensive Core i5-2500K and i7-2600K. And even more versatile than their K-series predecessors, since they are interesting not only with fully unlocked multipliers, but also with a more powerful graphics core.
To summarize, should the release of new processors be considered a success? Yes, count. Even despite the changed design, which will once again force upgrade enthusiasts to change boards: the new processors are good enough that even owners of systems with LGA1366 will be tempted to do this procedure (if only because they can replace some i7-920 with The i7-970 will be more expensive and less interesting than taking the i7-2600K on a new board) or LGA1156. Not to mention those who are still clinging to LGA775 - the time has come to finally retire any Core 2 Duo, and Core 2 Quad too. Well, those who buy assembled computers simply receive a small gift from the company - for the same money as in December last year, they can buy about 20 percent more processor power :)
