Intel core i5 3570k comparison. New technologies on an old core. Main competitor in testing
This article will be dedicated to the wonderful processor Intel Core i5-3570K. Thanks to the presence of an unlocked multiplier, this device can eliminate all possible competitors in comparisons, both in applications with games and in synthetic tests.
A description, overview of technical characteristics and reviews from owners will enable potential buyers to obtain all the necessary information about the device.
Intel Core i5-3570K: new technologies on an old core
The Intel Core i5-3570K processor is based on the Ivy Bridge QC core. However, if you compare its characteristics with the Sandy core, which was used in the production of the Core i5-2550K, you can find complete correspondence between the architectures. This factor raises many questions. According to the manufacturers, the only difference is the presence of a GMA HD4000 graphics core on the processor. The exact same video core is used in top-end Core i7 crystals. For the manufacturer, the introduction of high-quality integrated graphics into the processor core makes it possible to save on the purchase of a discrete video adapter. For those users who buy a computer not for high-performance games, but for work, it is quite nice to save money on purchasing an entry-level discrete video adapter.
Intel Core i5-3570K: technical specifications
If we move away from the existing rules for describing the characteristics of processors, which not all users understand, I would like to note that the new product has much in common with its older brother Core i7 – 3770 K. Such a significant difference lies only in the number of computation threads that have an impact on the overall performance of the chip. For this reason, it is better to study the technical characteristics in comparison with previous model. Both processors are manufactured using the 32 nm process technology. They have four cores. The Intel Core i5-3570K has four computation threads, while the Core i7 has 8. The L1 cache size for both processors is 32 KB. The volume of the second level L2 cache is 4×256 KB according to the number of cores. The volume of the third level cache is slightly different: the Core i5 has 6 MB, and the Core i7 has 8 MB. At the hardware level, both processors support DDR3 RAM – 1600 MHz with dual-channel mode. The power consumption of both processors is identical – 77 W.
Intel Core i5-3570K: first acquaintance
First of all, it is worth noting the fact that the manufacturer presented two interpretations of its wonderful product to the market at once: this is an Intel Core i5 die - 3570K BOX and an OEM version without a fan. Usually the first option is found on the domestic market. We will consider it.
The heart of the PC comes in a large cardboard box, which is protected from shock during transport. As you would expect, the packaging is quite informative. The manufacturer described all the technologies on the main screen and on the side edges of the box. On one side of the package there is a small window through which the processor and markings on the case will be visible. This is especially convenient for those buyers whose priority is the country of origin. All devices manufactured by Intel come standard. In the box you will find the processor itself, the cooling system, a branded sticker for the PC case, and installation instructions. You will also find a paper stack of recommendations in the box, which is not informative enough for the user.
Intel Core i5-3570K: briefly about the cooling system
The company has a rather interesting policy regarding the price of BOX and OEM versions. The devices have almost the same cost. Many buyers are interested in questions related to the effectiveness of the branded system Intel cooling Core i5-3570K. The cost of the device ranges from 15 to 20 thousand rubles. There is no need to worry about this for those users who do not plan to overclock the system in the future. The standard Intel cooling system is practically in no way inferior to its competitors in terms of operating efficiency, including noise indicators. There are no comments on the efficiency of the cooler when overclocking the central processor. However, high fan speeds may cause some discomfort. The cooling system makes a loud noise, which causes dissatisfaction among many users. For overclocking, experts recommend using radiators that are mounted into the motherboard using spring-loaded screw systems, rather than standard plastic latches.
Intel Core i5-3570K: main competitors
There is no point in comparing inexpensive solutions from the competing company AMD, which are built on the basis of 4 and 6 cores, with the Intel Core i5-3570K. The opponent simply has no chance of winning in synthetic tests and games. In this case we are talking about standard frequency values without overclocking. Therefore, in all tests, the Intel Core i5-3570K is compared with AMD products with 8 cores. The logic of this comparison is quite simple - the difference in price. It turns out that the solution from Intel is in the same price category as a competing device. In this case, the determining factor will be the greed of the buyer, who believes that twice larger number cores should also increase system performance. Whether this is actually the case can be found out by conducting tests that can be found in the media.
Comparison of powerful processors
The test involved an Intel Core i5-3570K processor and an AMD FX-8350. The cost of both devices ranges from 15 to 20 thousand rubles. Both crystals have an unlocked multiplier, which makes it possible to compare them based on the overclocked frequencies of the platform. The competitor with 8 cores shows rather low performance in all synthetic tests when compared with Intel's results. User reviews often mention support for Hyper-Threading technology. It creates 4 virtual cores on the Core i5 processor, which stretch the system during tests. However, when disabling Hyper support Threading competitor AMD is approaching Intel's results.
The situation is different with the overclocking of the AMD representative to 4 GHz. When conducting synthetic tests, it gives performance much higher than the same Core i5. It even comes close to the flagship of the line - the Core i7 processor. The quad-core representative from Intel can also be overclocked to 4 GHz. As a result, he will get all the standing ovations. Few people pay attention to such a parameter as heat dissipation when conducting tests. It does not exceed 100 W for a Core i5 processor. For a product from a competing company, this figure is 125 W. You will have to spend money not only on the cooling system, but also on electricity bills.
Mind games with Intel Core i7
Due to the great similarity of production technology with flagship processor Intel Core i7-3770K line, the idea arises to compare this processor with the Intel Core i5-3570K we are considering. A review of both new products has already been carried out previously. It is only worth adding that the difference in the cost of these devices is quite significant. Core i7 is 5-7 thousand rubles more expensive. It is clear that the Intel Core i5 has no chance in normal operating modes, but if you remember about the presence of an unlocked multiplier, you can try testing with an overclocked processor. Interactive work in 3D packages, including dynamic scenes, demonstrates high performance Intel work Core i5-3570K. However, the processor performance is still not enough to beat the flagship model. This is a little strange, because, if you follow the logic, the Core i7 operates at lower frequencies, which means it should have given primacy.
Intel: unspoken policy
Many owners of Intel Core i5-3570K processors in their reviews quite often put forward the assumption that the manufacturer specifically made it so that during overclocking the performance did not exceed the values of the top-end device. Looking at the test results budget processors Core i3, then you can find that when overclocked, the maximum performance does not exceed the values that were achieved by the top line Core i5. Of course, with pricing policy It's hard for the company to argue. Many buyers will have to live with what they have. If you want a powerful processor, buy the flagship Core i7 model. For rival company AMD, this is a little simpler. Tests demonstrate this. Overclocked representatives of the budget class easily outperform their counterparts who operate at standard frequencies.
Intel Core i5-3570K: gaming performance
At first glance, everything is clear. When conducting synthetic tests, the leadership remains with the Intel Core i5-3570K. Overclocking is remarkable, which leaves no chance for competing solutions from AMD. Practice shows that if you study the performance of tests in games and calculate the effectiveness of a given purchase in terms of price-quality ratio, then it will be more profitable to buy a competitor’s product. The whole point here is that synthetic tests relate only to the performance of the central processor. Gaming requires the power of the entire platform. In games such as Far Cry, Metro 2032, GTA, World of Tanks and others, the difference between processors from AMD and Intel is insignificant.
In normal mode, the Intel Core i5-3570K outperforms the FX-8350 core by several frames per second. Users also note in their reviews that one of the problems Intel products is collaboration with memory modules DDR3-1866 MHz and higher.
Choosing a decent processor
Before buying a new Intel Core i5-3570K processor, the user must understand for what purposes he plans to use it. If you need a powerful platform that can be used for video processing and editing, a cost-effective system for office applications and games, then the solution from Intel should be quite a good buy. However, if you only need a processor for games, then you should give preference to an inexpensive device. So, having saved on the cost of the processor, you can invest additional money in purchasing an SSD drive, video adapter and random access memory. This solution will contribute to increased performance in dynamic computer games. If we are talking about multi-core processors is not carried out, then you can completely get by with top-end solutions built on the basis of 2 or 4 cores. Today this is the ceiling for all high-performance gaming systems.
Conclusion
The result turned out to be quite interesting. The buyer will always have to make a choice. There are no questions about the performance of the Intel Core i5-3570K processor with an unlocked multiplier. He is one of the best representatives in his class. However, when it comes to gaming platforms, the cost of the product spoils the whole picture. Savings when purchasing a gaming solution from AMD average about 30%. If you need to use an inexpensive solution from a competing company with fairly demanding applications, then the high performance of platforms from Intel immediately comes to mind. There is no middle ground for processors in this class. They either have high performance for a lot of money, or the price is quite reasonable, but this is reflected in the power of the system.
Before purchasing a new processor, the user needs to clearly understand for what purposes it will be used computer hardware. For gaming systems, a silicon crystal with two to four cores and a good clock speed is quite suitable. For versatile applications, a more universal chip is needed. An example of this is the Intel Core i5-3570K. It is perfect for video editing, 3D modeling, and office applications. It is this model that the current review is devoted to.
Equipment Intel Core i5-3570K
The Core i5-3570K package can be called traditional. The package contains:
- the processor itself;
- installation instructions;
- warranty obligations;
- complete cooling system;
- traditional manufacturer's sticker (where would we be without it!).
To cool the microchip, the developers suggest using the E97378-001 cooler. This model is most often used for the Core i5 and Core i7 families. The fan blades have a fairly large angle of attack and provide good heat dissipation to the copper tubes and aluminum fins of the radiator. The device is quite quiet and does its job well.
All technologies and assembly features are described on the front and side edges of the box.
Technical characteristics of the Intel Core i5-3570K processor
The Core i5-3570K processor contains four cores with a base frequency of 3.4 GHz. It was created based on the Ivy Bridge architecture. However, the only thing fundamental difference from Sandy, which the manufacturer highlighted, is the presence of an integrated graphics adapter Intel HD Graphics 4000. The video card belongs to the third generation and operates with a clock frequency from 350 to 1350 MHz. So wide range possible thanks to Turbo Boost technology.
Most users liked the presence of a built-in video core with good graphics. It will handle medium-demand games and most applications. Avid gamers, of course, will have to separately purchase a powerful discrete video processor.
Technical process – 22 nanometers. The L1 cache is divided into two parts:
- four blocks of instructions, 32 KB each;
- four data blocks of 32 KB each.
Four 256 KB modules are allocated for the intermediate buffer of the second level, and 6 MB for L3. The LGA 1155 socket pins are located on the back of the processor.
Supply voltage of this processor quite high - 1.38 Volts. At the same time, the chip dissipates 77 watts of power. We will demonstrate other characteristics in the general table.
| Processor model | Intel Core i5-3570K |
| Cipher | SR0PM |
| Processor socket type | LGA1155 |
| Operating frequency | 3400 MHz |
| Turbo Boost Frequency | 3800 MHz |
| Kernel architecture | Ivy Bridge |
| Number of cores/threads | 4/4 |
| Thermal energy consumption | 77W |
| Technical process | 22 nm |
| Memory type | DDR3-1333/ DDR3-1600 |
| Maximum throughput | 25.6 Gb/sec |
| Integrated graphics card | GMA HD4000 |
Synthetic test results
To find out how the Intel Core i5-3570K will perform in games and popular programs, let's do some tests. To analyze performance we use special benchmarks. In the table below we show how the model in question performed in comparison with its closest Intel competitors.
| Benchmark and application | Processor model | Result |
| Futermark 3D-Mark Vantage, iGPU, Score | Intel Core i5-3570K + Turbo | 4253 |
| Intel Core i7-3770K + Turbo | 4370 | |
| Intel Core i5-2500K + Turbo | 1870 | |
| Intel Core i3-3220 | 1800 | |
| Resident Evil 5, iGPU, FPS | Intel Core i5-3570K + Turbo | 25,5 |
| Intel Core i7-3770K + Turbo | 25,8 | |
| Intel Core i5-2500K + Turbo | 16,9 | |
| Intel Core i3-3220 | 15,5 | |
| Warhammer 40,000: Dawn of War II - Retribution, iGPU, FPS | Intel Core i5-3570K + Turbo | 31 |
| Intel Core i7-3770K + Turbo | 31 | |
| Intel Core i3-3220 | 10,6 | |
| Fritz Chess Benchmark 4.2, knodes/sec | Intel Core i5-3570K + Turbo | 11573 |
| Intel Core i7-3770K + Turbo | 13998 | |
| Intel Core i5-2500K + Turbo | 9517 | |
| Intel Core i3-3220 | 6457 |
As you can see, the Core i5-3570K microchip confidently holds its mark and competes worthy with the representative of the older Core i7 family.
Almost every day, like reports from the front, we read with bitterness the news that the market desktop computers continues to lose its loyal supporters. It’s not just the army of users who suffer losses. One after another, hardware manufacturers are dropping out of the ranks of classic desktops. But it is especially offensive when, among companies that have made a name for themselves and earned enormous capital precisely in the desktop systems market, traitors and saboteurs are discovered, in words declaring unshakable loyalty to the old ideals, but in reality - not only looking, but also actively going “on the side” » ( mobile devices, naturally). A blatant example of such treacherous infidelity, which has not yet been eclipsed in memory by some new terrible betrayal, has recently shown us Intel company.
Yes, yes, we are talking about Haswell. About that very processor, which was initially presented as another cycle of development of a high-performance microarchitecture, but in fact turned out to be purposefully and deeply adapted for use in low-power portable computing systems. The same Haswell that users of desktop systems eventually received, wits called Hasfail not empty space. The fourth generation Core desktop processors, based on a new microprocessor design, became a byproduct for Intel with all the ensuing consequences. Our review of the Core i7-4770K revealed the main shortcomings: the lack of clear progress in computing performance and deteriorating overclocking potential. The conclusion from all this was then made unequivocally: there is no point in upgrading existing systems and switching to the new LGA1150 platform.
However, several weeks have passed since Haswell's announcement, and the former indignation has subsided a little. Thoughts began to creep into my head about whether we were too eager to stigmatize the new processor design? Maybe desktop Haswells can be interesting after all, because these processors still have certain improvements. In other words, there is a need for a fresh look.
But, of course, we will not repeat tests already done a second time. Today we'll look at Haswell from a different angle. Namely, we will try to understand which Intel processor should be purchased by an enthusiast who has a budget of about 200-250 dollars for this purpose. That is, let’s try to answer the question of which of the overclocking Core i5s available in stores has the greatest practical value today. Since Sandy Bridge, in each new generation of desktop CPUs we have seen small steps towards improving performance, on the one hand, but a systematic rollback in overclocking potential, on the other. Therefore, choosing modern platform, advanced users today actually face a trilemma: Sandy Bridge, Ivy Bridge or Haswell. And in this material we decided to directly compare all three available options: Core i5-2550K, Core i5-3570K and Core i5-4670K.
⇡ Excursion into processor microarchitectures
We are all accustomed to the fact that the newer the processor, the better it is. And until recently it really worked. Production processes were improved. This resulted in an increase in frequency potential and an increase in the complexity of processor semiconductor crystals. The increased transistor budget was spent either on microarchitectural innovations, or on increasing the number of cores or increasing the amount of cache memory.
However, since the advent of Sandy Bridge generation processors, the usual pace of progress began to slow down. Even though Sandy Bridge uses 32nm technology, and the newer Ivy Bridge and Haswell use 22nm technology, all three generations of desktop processors have a similar multi-core structure, operate at very similar clock speeds and have the same cache memory volumes. Virtually all differences affecting performance are now buried in the depths of the microarchitecture.
In principle, there is nothing wrong with the fact that the formal specifications of desktop processors have stopped growing since 2011. As we know from previous experience, microarchitectural improvements can do a lot. Moreover, both Ivy Bridge and Haswell are not simple “tics” in Intel terminology. Even about Ivy Bridge, the release of which was associated with a change in the technical process, Intel spoke as a “tick+” cycle, emphasizing that we are not talking about a simple transfer of Sandy Bridge to a new technological framework, but about a comprehensive refinement of the old design. Haswell generally refers to the development cycle “this way”, that is, it represents new version microarchitecture without any reservations. Therefore, an increase in performance could be expected from the current development of Intel processors, even if it is not accompanied by a change in numbers in the list of formal characteristics.
However, there is actually no rapid growth in the performance of desktop processors. The reason is that the main efforts of Intel developers are not aimed at improving computing power - it is more than enough to leave competitors far behind - but at improving parameters that are critical for mobile market. Wanting to simultaneously outshine both AMD hybrid processors and mobile processors with ARM architecture, Intel is systematically optimizing heat dissipation and power consumption, and is also improving its own graphics core. For desktop processors, these parameters are of little significance, therefore, from the point of view of desktop computer users, the development of Sandy Bridge → Ivy Bridge → Haswell looks like a manifestation of technological infantilism.
Let's try to remember what has happened to the computing cores of processors since 2011, when the first Sandy Bridge appeared on the market with a truly innovative microarchitecture with a completely redesigned out-of-order command execution scheme. The original Sandy Bridge design became a solid basis for all subsequent generations of microarchitecture. It was then that such key and still relevant elements appeared as a ring bus, a “zero-level” cache of decoded instructions, fundamentally new block branch predictions, execution scheme for 256-bit vector instructions and much more. After Sandy Bridge, Intel engineers limited themselves to only minor changes and additions, without affecting the foundation laid in this microarchitecture.
In the Ivy Bridge family of processors released a year later, progress affected the computing cores to a very small extent. As a conveyor front designed for processing four instructions per clock, and the entire scheme of out-of-order execution of commands has been preserved in its completely original form. However, the performance of Ivy Bridge is still slightly higher than that of its predecessors. This was achieved in three small steps. First, a long-overdue opportunity has arisen dynamic distribution resources of internal data structures between threads, while previously all queues and buffers calculated for Hyper-Threading were divided into two threads strictly in half. Secondly, the unit for executing integer and real division was optimized, as a result of which the rate of execution of these operations doubled. And thirdly, the task of processing data transfer operations between registers was removed from actuators, and the corresponding commands began to be translated into simple register dereferences.
With the advent of Haswell, computing performance has increased slightly again. And although there is no reason to talk about a qualitative leap, the set of innovations does not look nonsense. In this processor design, engineers dug deep into the mid-pipeline, resulting in Haswell increasing the number of execution ports (by the way, for the first time since 2006). Instead of six, there are eight, so in theory, Haswell's pipeline throughput has increased by a third. At the same time, a number of steps were taken to ensure that all these ports work, that is, to improve the processor’s ability to execute instructions in parallel. For this purpose, branch prediction algorithms were optimized and the volume of internal buffers was increased: first of all, the windows for out-of-order execution of commands. At the same time, Intel engineers expanded the instruction set by adding a subset of AVX2 instructions. The main asset of this set is FMA commands, which combine a couple of operations on floating point numbers at once. Thanks to them, Haswell's theoretical performance for single- and double-precision floating-point operations has doubled. The subsystem for working with data also did not go unnoticed. The expansion of the internal parallelism of the processor, as well as the emergence of new instructions that move large amounts of data, required developers to speed up the operation of the cache memory. Therefore, the L1 and L2 cache bandwidth in Haswell has been doubled compared to previous generation processor designs.
However, when new generations of processors are released, enthusiasts want to see not so much extensive lists of changes made, but rather increased bars on application performance charts. Therefore, we will supplement our theoretical calculations with the results of practical tests. Moreover, for better illustrativeness, first of all we will resort to a synthetic benchmark, which allows us to see changes in various aspects of performance isolated from the overall picture. The popular test utility SiSoftware Sandra 2013 is excellent for this purpose, using which we compared three quad-core processors (Sandy Bridge, Ivy Bridge and Haswell), the clock frequency of which was adjusted to a single and constant value of 3.6 GHz. Please note that Haswell indicators are shown twice in the graphs. Once - when the testing algorithms do not use the new instruction sets introduced in this processor design, and the second time - with AVX2 instructions activated.
A simple arithmetic test reveals that Haswell has seen a noticeable increase in integer performance. The increase in speed is obviously associated with the appearance in this microarchitecture of a port specially designated for an additional integer arithmetic-logical device. As for speed standard operations floating point, it does not change with the release of new generations of processors. This is understandable, because the emphasis now is on introducing new sets of instructions with a higher bit capacity into everyday use.
When evaluating multimedia performance, the speed of execution of vector instructions comes first. Therefore, Haswell's advantage here is especially strong when using the AVX2 kit. If we exclude new instructions from consideration, we will see only a 7% increase in performance compared to Ivy Bridge. Which, in turn, is only 1-2 percent faster than Sandy Bridge.
The situation is similar with the speed of cryptographic algorithms. The introduction of new generations of microarchitectures increases productivity by only a few percent. A significant increase in speed can only be obtained if you use Haswell and its new commands. However, make no mistake: taking advantage of AVX2 in real life requires rewriting the program code, and this, as you know, is far from a quick process.
What happened to cache memory latency doesn’t look too optimistic either.
| Latency, clock cycles | |||
|---|---|---|---|
| Sandy Bridge | Ivy Bridge | Haswell | |
| L1D cache | 4 | 4 | 4 |
| L2 cache | 12 | 12 | 12 |
| L3 cache | 18 | 19 | 21 |
Haswell's L3 cache actually works with b O higher delays than in processors of the previous generation, since the Uncore part of this processor received asynchronous clocking relative to the computing cores.
However, the increase in delays is more than compensated by a twofold increase in bandwidth, which occurred not only in theory, but also in practice.
| Bandwidth, GB/s | |||
|---|---|---|---|
| Sandy Bridge | Ivy Bridge | Haswell | |
| L1D cache | 510,68 | 507,64 | 980,79 |
| L2 cache | 377,37 | 381,63 | 596,7 |
| L3 cache | 188,5 | 193,38 | 206,12 |
But in general, the Haswell microarchitecture, compared to Sandy Bridge, still does not look like a noticeable advance. The fundamental advantage is observed only when using the AVX2 command set, and so far it can only be observed in synthetic tests, since the real software still has to go through a long path of optimization and adaptation. If the new instructions are not taken into consideration, then average level Haswell's superiority over Sandy Bridge is about 10 percent. And the old Sandy Bridge guys should be able to overcome such a gap through overclocking. Especially considering the fact that the frequency potential of older processors is higher than that of their modern successors.
⇡ Three generations of Core i5 for overclockers
If you go to the store and look at what overclocking processors from the Core i5 family you can buy, the choice will come down to three options belonging to different generations: Core i5-2550K, Core i5-3570K and Core i5-4670K. For clarity, let’s compare their characteristics:
| Core i5-2550K | Core i5-3570K | Core i5-4670K | |
|---|---|---|---|
| Microarchitecture | Sandy Bridge | Ivy Bridge | Haswell |
| Cores/threads | 4/4 | 4/4 | 4/4 |
| Hyper-Threading Technology | No | No | No |
| Clock frequency | 3.4 GHz | 3.4 GHz | 3.4 GHz |
| Maximum frequency in turbo mode | 3.8 GHz | 3.8 GHz | 3.8 GHz |
| TDP | 95 W | 77 W | 84 W |
| Production technology | 32 nm | 22 nm | 22 nm |
| HD Graphics | No | 4000 | 4600 |
| Graphics core frequency | - | 1150 MHz | 1200 MHz |
| L3 cache | 6 MB | 6 MB | 6 MB |
| DDR3 support | 1333 | 1333/1600 | 1333/1600 |
| Instruction Set Extensions | AVX | AVX | AVX 2.0 |
| Package | LGA1155 | LGA1155 | LGA1150 |
| Price | No data | No data | No data |
Three Core i5 different generations look almost like twin brothers in this table. However, a more detailed acquaintance with each of these three processors reveals interesting nuances.
Corei5-2550K. This is one of the latest Sandy Bridge models. It was released a year after the main announcement and was discontinued only recently, and therefore is still widely available in retail sales. But if you are seriously thinking about building a system based on the Core i5-2550K processor, then we consider it our duty to remind you of a number of important points.
Firstly, despite the fact that in the formal specifications the operating frequencies of all older Core i5 models are designated the same: from 3.4 to 3.8 GHz, in reality the Core i5-2550K in normal mode operates at a slightly lower frequency than processors with later versions of the microarchitecture. The fact is that the Turbo Boost technology in Sandy Bridge is not as aggressive as in Ivy Bridge and Haswell, and at full load the frequency exceeds the rated frequency by 100 rather than 200 MHz.
Secondly, Sandy Bridge processors - and the Core i5-2550K among them - have a slightly less flexible memory controller than Ivy Bridge and Haswell. It supports overclocking memory with frequencies up to DDR3-2400, but the step of changing this frequency is 266 MHz. That is, the choice of memory modes when using the Core i5-2550K is somewhat limited.
And thirdly, the Core i5-2550K is the only Intel overclocking processor that lacks a graphics core. In fact, there is a core on the semiconductor chip, but it is strictly disabled at the processor assembly stage. This, by the way, is one of the reasons why the Core i5-2550K overclocks well.
However, the main reason for the attractiveness of the Core i5-2550K as an object for overclocking is that Sandy Bridge is the last of the families of Intel desktop CPUs in the mid-price category, where a special flux-free solder is used as a thermal interface between the semiconductor crystal and the processor cover, rather than plastic material with questionable thermal conductivity. Intel considered the subsequent transfer of semiconductor production to 22-nm technology and the accompanying reduction in the heat emission of crystals to be a sufficient argument for simplifying the CPU assembly technique by eliminating soldering. However, overclockers suffered seriously from this, since the thermal interface between the processor chip and its cover unexpectedly became a significant obstacle to the transfer of heat flow and the organization of good cooling.
Corei5-3570K. A typical carrier of the Ivy Bridge design - the first generation of Intel processors produced using the 22 nm process technology. Using something more advanced than before, technological process allowed Intel to significantly reduce processor heat generation and power consumption. Systems built on the Core i5-3570K are obviously more economical than similar configurations on Sandy Bridge. However, Intel did not convert this advantage into an increase in clock frequencies. The operating frequencies of the older third-generation Core i5, Core i5-3570K, are almost the same as those of the Core i5-2550K.
What's worse, despite the lower nominal voltage and heat dissipation in nominal mode, Ivy Bridge generation processors are much less willing to overclock than their predecessors. The problem is that due to the reduction in the physical dimensions of the crystal that accompanies the introduction of a more subtle technical process, the density of the heat flux emitted by it has increased. At the same time, the removal of this heat is artificially hampered by the sabotage committed by Intel technologists to remove the highly efficient thermal interface, proven over the years, from under the processor cover. Therefore, without using extreme methods Ivy Bridge cooling does not reach the same high frequencies as Sandy Bridge when overclocked.
So, if you close your eyes to minor microarchitectural improvements and decreased energy appetites, the only thing that the Core i5-3570K can be better than the Core i5-2550K in an overclocking system is a more flexible DDR3 SDRAM controller, which allows you to set higher memory levels than before , frequencies and vary them in smaller steps.
Corei5-4670K. Latest processor Based on the Haswell microarchitecture for the new LGA1150 platform, it again has almost the same formal characteristics as its predecessors. In other words, we haven’t seen an increase in nominal clock speeds in the Core i5 series for a very long time. At the same time, the Core i5-4670K, compared to Ivy Bridge, is surprising in the increase in calculated heat dissipation, which occurred against the background of the unchanged semiconductor process technology.
But everything is quite understandable. The increase in heat dissipation is due to fundamental changes in the design of the platform: in LGA1150, a significant part of the power converter has been moved from the motherboard to the processor. On the one hand, this has significantly simplified the design of the platform, since the processor now generates all the voltages necessary for its operation independently. On the other hand, it gave the processor a full set of means to monitor and manage its own power consumption.
As for overclocking, the built-in power controller brings certain benefits here too. It is very accurate, and the voltages it produces are virtually undistorted as current or temperature increases. When setting a fixed voltage on processor cores, this allows you to forget about the horrors of Loadline Calibration, that is, it simplifies the selection of parameters in overclocker configurations. However, it should be borne in mind that when dynamically setting processor voltages in the offset and adaptive modes, the built-in controller goes crazy during overclocking and very zealously increases the voltage as the load increases. Therefore, the use of such modes is undesirable; it does not allow Haswell to fully reveal its overclocking potential.
However, all this is not so important, since the final assembly scheme for Haswell desktops has not changed. Thermal paste that is not of the best quality is laid between the semiconductor crystal and the processor cover, so overclocking the Core i5-4670K, like the Core i5-3570K, in the vast majority of cases rests on overheating of the processor crystal, which cannot be eliminated by conventional means.
For the same reason, the changes made to the LGA1150 platform, which allow overclocking the Core i5-4670K not only by a multiplier, but also by the frequency of the base clock generator, do not inspire optimism. Of course, all this adds a certain flexibility when choosing options, but, unfortunately, bringing the maximum achievable frequencies in overclocking closer to the bar, installed by processors Sandy Bridge does not allow it without the use of extreme cooling methods. Moreover, as practice shows, due to their higher heat dissipation, Haswell overclocks even worse than their Ivy Bridge generation predecessors.
Intel traditionally makes users confused about its products, wandering between the numerical designations of processors in the same line. Intel Core i5 was no exception, although in the case of its two flagships 3570k and 3570 there should be no confusion. Both models represent a productive, relatively inexpensive and modern solution, and multimedia and gaming systems are assembled on their basis.
Definition
Intel Core i5 3570 is a quad-core i5 processor from Intel, built on the Ivy Bridge microarchitecture and designed for desktop systems. Positioned as a reasonable compromise in terms of cost and performance.
Intel Core i5 3570k— a quad-core i5 processor from Intel with an unlocked multiplier, built on the Ivy Bridge microarchitecture and designed for systems that require overclocking.
Comparison
As evidenced by the letter k in the model name, the 3570k is a processor with an unlocked multiplier, and it can be overclocked not only on the bus, but by increasing this very multiplier, thereby increasing performance. The 3570 model does not have an unlocked multiplier, so it is of little interest to overclockers. Overclocking significantly increases temperatures, so using the 3570k entails the expense of a good cooling system.
Intel Core 3570 processor on board graphics core HD Graphics 2500, 3570k includes HD Graphics 4000 graphics. This difference may form the basis of the choice, since the performance of the first core when performing a wide range of tasks is almost two times lower than that of the older model. Technically, HD Graphics 2500 graphics are functionally identical to HD Graphics 4000, only fewer execution units are involved.
Intel Core i5 3570k
Oddly enough, the older model 3570k lacks support for many technologies implemented in the 3570. These are, for example, the Intel vPro security tool, Intel VT-d directed I/O virtualization technology, and the Intel Trusted Execution isolated program launch environment.
Intel Core i5 3570k will cost more than Intel Core i5 3570 by about 20 dollars, or about 700-800 rubles.
Conclusions website
- The Intel Core i5 3570k processor has an unlocked multiplier, is overclockable and is designed for overclocking.
- The Intel Core i5 3570k processor includes an HD Graphics 4000 graphics core, the Intel Core i5 3570 has an HD Graphics 2500, respectively, the former’s graphics performance is higher.
- The Intel Core i5 3570 processor supports more proprietary technologies.
- The Intel Core i5 3570k processor is more expensive.
This year, Intel decided to release new processors not in January, but in April. Which, in principle, is quite logical - for once there was no need for haste. Sandy Bridge, produced according to 32 nm standards, does not have such a high cost, and in terms of performance they compete only with themselves, so the partners could be given a break - let them get rid of warehouse stocks calmly. Moreover, the practice of announcing new platforms at the beginning of January in past years slightly undermined Christmas sales of old ones - some buyers considered it necessary to wait a month.
This time the situation is completely different. Firstly, it was necessary to wait for a whole block, which was psychologically more difficult. Secondly, there is simply no new platform. That is, of course, new chipsets have appeared, but not for all market segments: interesting to many budget systems, as before, will continue to be used Intel chipset H61, for which there is simply no replacement. And older chipset models are, in principle, not necessary - from a consumer point of view, they are not too much different from their predecessors. A truly noticeable improvement, namely the built-in USB controller 3.0, is greatly offset by the fact that on most boards of average and (even more so) high level support of this standard has been around for a long time too. Even if using a discrete controller, this is enough for external hard drives or, especially, mass-produced flash drives. And the compatibility of processors and boards is for once complete - you can rearrange Sandy Bridge into a board based on the “seventh” series chipset, or you can buy Ivy Bridge for an existing board on some H67, and automatically get both a new architecture and PCIe 3.0, for example.
And architecture, too by and large, not too new - “tick-tock” in all its glory. The previous step, namely Sandy Bridge, was precisely a step in the architectural sense (i.e. “so”). It was also aggravated by the fact that from the point of view of buyers of mass-produced quad-core processors, this step also concerned production technology - processors of this type for LGA1156 did not migrate to 32 nm. And now - just a change in production standards (“tick”). Of course, with certain internal improvements, but nothing more: there were no radical alterations. In any case, in the processor part - the video core (what Intel was kicked for for a long time) new and more powerful. And it occupies a slightly larger chip area than before.
More precisely, not the video core, but the video cores - all Core i7 will receive GMA HD 4000, as well as mobile processors, but desktop i5 and i3 can be equipped with both it and GMA HD 2500. What is the difference? As before, in the number of conveyors: the GMA HD 2500 supports as many of them as the GMA HD 2000 (both six each), so the main differences between these solutions seem to be in functionality, not performance . But in the GMA HD 4000 there are already 16 conveyors versus 12 in the HD 3000 version, which, even with the same architecture, would necessarily have an impact on performance. Moreover, we are talking not only about computing power, but also (finally) about full support DirectX 11. In general, there should be even fewer reasons to pay attention to low-end discrete video cards.
However, the video part - separate question, deserving a separate article. Or even several articles, because new architecture graphics is not only suitable for games or video transcoding (which promises as much as a twofold increase), but also supports OpenCL. Today we will focus on something else - we will test the performance of the processor part of older models of both lines: Core i5 and Core i7. Both models belong to the K-family, i.e. they can also offer the buyer unlocked multipliers. And since this year, the concept has changed a little - now the K does not have a cheaper twin with the same clock frequency, i.e. models with an unlocked multiplier from all points of view are the eldest in the family. (Doesn’t remind you of anything? Well, yes - the Black Edition in its purest form.) More precisely, for the i5 this is completely true: the senior model of the “regular” line is the 3550, which differs from the 3570K at 100 MHz in both base and maximum frequency(and even GMA HD 2500 versus 4000). But the i7-3770 and i7-3770K differ only in the base frequency with the same maximum frequency and video core. But how this will affect productivity (you can be sure that it will), how poor sales consultants will be able to explain the difference to customers, and other philosophical questions are not of interest to us today. Since there are only two processors in the laboratory, let us remind you that so far they have received exactly two and both are “improved”.
Test bench configuration
| CPU | Core i5-2550K | Core i5-3570K | Core i7-2700K | Core i7-3770K |
| Kernel name | Sandy Bridge QC | Ivy Bridge QC | Sandy Bridge QC | Ivy Bridge QC |
| Production technology | 32 nm | 22 nm | 32 nm | 22 nm |
| Core frequency (std/max), GHz | 3,4/3,8 | 3,4/3,8 | 3,5/3,9 | 3,5/3,9 |
| 34 | 34 | 35 | 35 | |
| Scheme Turbo works Boost | 4-3-2-1 | 4-3-2-2 | 4-3-2-1 | 4-3-2-2 |
| 4/4 | 4/4 | 4/8 | 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 | 8 | 8 |
| UnCore frequency, GHz | 3,4 | 3,4 | 3,5 | 3,5 |
| RAM | 2×DDR3-1333 | 2×DDR3-1600 | 2×DDR3-1333 | 2×DDR3-1600 |
| Video core | - | GMA HD 4000 | GMA HD 3000 | GMA HD 4000 |
| Socket | LGA1155 | LGA1155 | LGA1155 | LGA1155 |
| TDP | 95 W | 77 W | 95 W | 77 W |
| Price | N/A() | $284() | $316() | $431() |
We don't have today new platform, But new line processors, so, despite the fact that there are only two newcomers, there will be a lot of competitors. Let's start with the two main ones. The new 3570K and the not-so-new 2550K have similar parameters and close (but not equal!) prices, so we absolutely need them. At the same time, they differ radically in terms of graphics, which are simply blocked on the 2550K, and the TDP of these models is different. Nothing unexpected - there have long been rumors about how Intel will use the advantages of the new technological process. In general, now all quad-core models fit into a thermal package of 77 W, which is comparable to dual-core Core i3/i5 for LGA1156. Frequencies will increase significantly only for energy-efficient processors, which will make families closer to each other: for example, the i5-3570S with a TDP of 65 W has base frequency 3.1 GHz (like the “regular” i5-2400), and the maximum is the same 3.8 GHz as the 3570K. Well, for manufacturers, for example, this will give the opportunity to use “regular” Ivy Bridge even in all-in-one computers and other compact systems (if desired, of course - as shown above, the current S-family is already comparable in performance to the models of the main line).
Fans of traditional systems have the opportunity to get by with standard cooling systems even with a small overclock, and even with a large overclock, they may be able to squeeze out something extra (especially since the maximum multiplier in the K-series has been increased from 57 to 63; although the old value is there were almost no restrictions in practice). Also, support for high-frequency memory has been slightly improved, but this is for everyone else - overclockers previously preferred systems on chipsets that support overclocking, so they could at least use DDR3-2133. They, however, were also pleased that the upper limit of the memory frequency during overclocking was moved to 2667 MHz, but here it is more important that 1600 MHz is now available even on the simplest H61, since this standard feature processor, not overclocking. True, on our board based on the H67 chipset, a multiplier of 16 was available only when using a discrete video card, but it is not a fact that users of integrated video will chase high-frequency modules. Moreover, this may well turn out to be a feature of a specific board or even version UEFI firmware. But in the future we will try to look for a more precise answer to this question.
| CPU | Core i7-2600 | Core i7-3820 | Core i7-3930K | Phenom II X6 1100T | FX-8150 |
| Kernel name | Sandy Bridge QC | Sandy Bridge-E | Sandy Bridge-E | Thuban | Zambezi |
| Production technology | 32 nm | 32 nm | 32 nm | 45 nm | 32 nm |
| Core frequency (std/max), GHz | 3,4/3,8 | 3,6/3,8 | 3,2/3,8 | 3,3/3,7 | 3,6/4,2 |
| Starting multiplication factor | 34 | 36 | 32 | 33 | 18 |
| How Turbo Boost works | 4-3-2-1 | 2-2-1-1 | 6-6-5-4-3-3 | - | - |
| Number of cores/threads | 4/8 | 4/8 | 6/12 | 6/6 | 8/8 |
| L1 cache, I/D, KB | 4×32/4×32 | 4×32/4×32 | 6×32/6×32 | 6x64/6x64 | 4x64/8x16 |
| L2 cache, KB | 4×256 | 4×256 | 6×256 | 6×512 | 4×2048 |
| L3 cache, MiB | 8 | 10 | 12 | 6 | 8 |
| UnCore frequency, GHz | 3,4 | 3,6 | 3,2 | 2 | 2,2 |
| RAM | 2×DDR3-1333 | 4×DDR3-1600 | 2×DDR3-1333 | 2×DDR3-1866 | |
| Video core | GMA HD 2000 | - | - | - | - |
| Socket | LGA1155 | LGA2011 | AM3 | AM3+ | |
| TDP | 95 W | 130 W | 130 W | 125 W | 125 W |
| Price | $340() | $318() | $546() | N/A(0) | N/A(0) |
So, the first pair is Core i5-2550K and 3570K. The second is Core i7-2700K and 3770K, for which everything said above about the first is true (except that the video core of the 2700K remains in place and works, unlike the 2550K). Who else can you compare this four with? Firstly, it's time to dust off the Core i7-2600 again: frequencies processor cores it has exactly the same as the Core i5-2550K/3570K. But there is more cache memory and support for Hyper-Threading: we’ll see how this affects it. Secondly, we will need two processors for LGA2011: a quad-core Core i7-3820 (which in the overall standings was slightly ahead of 2700K, so it’s very interesting to compare it with 3770K) and a six-core Core i7-3930K (out of competition, but... again - it will be very interesting and useful to compare the 3770K with it). Well, since we have as many as seven Intel processors, and two of them belong to the Core i5 line, it’s worth adding a couple of AMD models to the list: the fastest from the new and from the old (but still popular) lines - FX-8150 and Phenom II X6 1100T. Moreover, after the previous article about AMD products, we received complaints that, supposedly, comparison with old Intel processors is a form of mockery of the company. Well - get a comparison with the new and latest :)
| Motherboard | RAM | |
| LGA1155 | Biostar TH67XE (H67) | Corsair Vengeance CMZ8GX3M2A1600C9B (2×1333; 9-9-9-24) |
| LGA2011 | ASUS P9X79 Pro (X79) | 16 GB 4x1333; 9-9-9-24 |
| AM3 | ASUS M4A78T-E (790GX) | Corsair Vengeance CMZ8GX3M2A1600C9B (2×1333; 9-9-9-24-2T, Unganged Mode) |
| AM3+ | ASUS Crosshair V Formula (990FX) | G.Skill F3-14900CL9D-8GBXL (2×1866; 9-10-9-28) |
Despite what we said above about memory support, regular DDR3-1333 was again used in testing - this makes it easier to compare the processors themselves. And if we increase the frequency, then not from 1333 to 1600 MHz, but higher - at least to the FX level. For the same reason, we have not yet started using motherboard on a new chipset: since this time there is an opportunity to achieve the most equal testing conditions, it would be a sin not to take advantage of it.
Testing
Traditionally, we divide all tests into a number of groups, and show the average result for a group of tests/applications in diagrams (you can find out more about the testing methodology in a separate article). The results in the diagrams are given in points; the performance of the reference test system from the 2011 sample site is taken as 100 points. It is based on the AMD Athlon II X4 620 processor, but the amount of memory (8 GB) and video card () are standard for all tests of the “main line” and can only be changed within the framework of special studies. For those who are interested in more detailed information, again, it is traditionally proposed to download a table in Microsoft Excel format, in which all the results are presented both converted into points and in “natural” form.
Interactive work in 3D packages
Traditionally a low-threaded group in which the “Bulldozer” modules turn into pumpkins, and most of the “full-fledged” Phenom cores are idle. However, in Core, the kernels turn out to be no less “superfluous”, and if there is anything from Hyper-Threading, then useless, and therefore Core i5 is faster than Core i7 at the same clock speed. With the same cores, internal optimizations in Ivy Bridge allow processors on the new core to run a little faster. Quite a bit, but this is already enough for the Core i7-3770K to become the leader, and the Core i5-3570K is only behind it and the Core i7-2700K, which was recently the flagship of the line.
Final rendering of 3D scenes
“Undocking” the modules does not work for the reasons already described more than once - there are just four vector blocks. The Phenom II X6 has a slightly better position, but we can only talk about fighting with the old Core i5. The latter finds it difficult to fully compete with the Core i7 - in this group the effect of Hyper-Threading is clearly visible to the naked eye. Therefore, the Core i5-3570K is slower than the Core i7-2600. But 5% faster than Core i5-2550K. A little? Yes a little. However, no one promised much. But the TDP is only slightly higher than that of dual core processors first and even second Core generation, and their performance in such tasks is much lower. And the Core i7-3770K, where the new architecture is combined with Hyper-Threading, has become the fastest quad-core processor. Naturally, it is fundamentally behind the six-core Core i7-3930K, but ahead of the former leaders of its class by the mentioned 5%. And the remark about TDP, of course, also applies to him.
Packing and Unpacking
The FX-8150 stops looking like a whipping boy, reaching a level comparable to the Core i5. In archiving with 7-Zip, it is even faster than these processors, but only one subtest out of four can boast such good multithreading support, so overall result at the Core i5 level, but lower models. The older ones have only one competitor - the modern Core i7. At the same time, among the three processors operating at the same frequency, the i5-3570K is exactly average, i.e., it is again 5% faster than the 2550K. And the situation is similar for the Core i7-3770K - even the larger cache or the four-channel memory controller did not help the Core i7-3820 “fight” with it, which was enough to overtake the 2700K. The Core i7-3930K, of course, is still ahead of everyone, but not with such a noticeable margin as it was at the time of its announcement last year.
Audio encoding
We finally get to the application where the FX-8150 justifies its positioning between the Core i5 and i7, due to the very high dependence of the test on the number of cores and computation threads. Within reasonable limits, of course, that’s why Core i5 has long been able to easily outperform six-core Phenom IIs. Ivy Bridge does not demonstrate miracles, and no one promised them. True, the i5-3570K outperformed its predecessor by 6.5% - not much, but not bad. And the gap between the i7-3770K and 2700K has already exceeded 7%. And even more importantly, the six-core 3930K is only 10% faster than the 3770K, i.e., thanks to the update, the cheaper platform “caught up” half the gap from the more expensive one.
Compilation
But in compilers the growth has decreased. However, again, in other applications it is only a pleasant bonus - after all, it is Core i5 and i7, i.e. four cores and four or eight computing threads with the same cache memory capacity. Perhaps we can “squeeze out” a little more due to a slightly higher memory frequency, but we have not yet taken advantage of this advantage of the new core. And it’s obvious that there won’t be a fundamental change in the situation - the Core i7-3930K is still much faster. They bypassed 3820 - and that’s it.
Mathematical and engineering calculations
And again a low-flow group and a slight advantage over the previous generation models. This turns into a victory, since the Core i7-3770K has become the absolute leader, and the i5-3570K is second only to it and the extreme Core i7-3960X - a normal result. Although here, in general, all processors have it Intel high class “normal”: the difference between them does not deserve to be discussed in detail. Especially considering the fact that AMD’s single-threaded performance is not doing great (and that’s putting it mildly), so there is simply no competition in tasks of this kind.
Raster graphics
Partial optimization for multi-threading of some of the programs included in the group previously allowed the Core i7 to work slightly faster than the equivalently frequency Core i5. In the new generation, apparently, this trend continues, but when comparing devices of different generations, representatives of the new one are slightly faster. Even if they are less advanced in terms of supporting various improving technologies. And if they exist, then the situation is even more beautiful from the point of view of LGA1155 fans: the Core i7-3770K is second only to the extreme one for LGA2011 and... that’s all. Again, there is simply no intercompany competition in this situation.
Vector graphics
These two programs are the best illustration of the fact that even professional packages in 2012 continue to successfully ignore the “multi-core race”. Therefore, the Core i5-3570K is again better than absolutely all processors, with the exception of its “brother” Core i7-3770K. It is possible that the i7-3770 will also be faster, although this is not too important: it will simply turn out that the three older Ivy Bridge models will be located in all three “prizing places”
Video encoding
It would seem that this is the very case where Hyper-Threading can give a lot, since most of the programs included in this group are very good at increasing the number of computational threads. However, the reality, as often happens, turned out to be a little more complicated than naive ideas about it - MS Expression Encoder and XviD are positive about increasing the number of cores, but not NT, so the Core i7 in them is slower than the Core i5 at the same frequency. In other programs it was faster, which gave greater overall performance. But when switching to a new core, the Core i5 accelerated so much that it lags behind the “old” Core i7-2600 only by Adobe Premiere, ahead of him in all other programs. And not only that, but also the Core i7-2700K! In general, the only “old” Core i7 that performs slightly faster than the new Core i5 is the relatively new Core i7-3820. And the Core i7-3770K, naturally, is even faster - at least a Core i7-3930K can beat it, and even then with a margin of less than 10%. A very illustrative example of the usefulness of new technologies.
By the way, it is indicative in another sense. After the first reviews of AMD FX, saddened fans often complained that there was no point in testing it with old programs - new ones would come out, and then it would destroy everyone. There is a certain amount of truth in this opinion, but only a fraction. Since practice has shown for the second time that new processors can be developed so that they are faster than old ones in any programs. Naturally, the maximum gain can only be obtained in optimized software, but... At the time of our first testing of Sandy Bridge, naturally, the methodology did not have applications well optimized for it. That didn’t stop the then-new Core i5 and i7 from easily crushing their predecessors, and more high class. It cannot be said that the situation has now been repeated exactly, since the increase is much more modest (as planned - this step is primarily dedicated to updating the production process, and not at all to a radical reworking of the microarchitecture), but it is there again.
Office software
Even here, improvements were found, albeit cheap, and, in general, unnecessary - as we have said more than once, in the performance of older models modern processors too much for this application. Some differences can only be found with the help of tests, so the easiest way is not to pay attention to them at all. Well, or in order to improve your general education, remember that this is not the first time that the extremely extreme Core i7-3960X is faster than the Core i7-3770K.
Java
But the JVM is a program for which no matter how much you give it, it won’t be enough. And in it one observes almost maximum effect from improvements in Ivy Bridge, which allowed the i5-3570K to fall exactly in the middle of the gap between the i5-2550K and i7-2600. Even a little closer to the second, although this is not important. But what’s interesting is a slight change in the situation in inter-company competition - after all, the older FX and Phenom II X6 used to outperform any Core i5. After the appearance of 2550K, the difference became minimal, and even if only a little, the FX-8150 remained faster. And taking into account the fact that, frankly speaking, there are few such groups, each one was all the more valuable. And now their number has decreased. Well, the Core i7-3770K is simply the fastest quad-core processor, which has already become common.
Games
We have long been haunted by a strong desire to get another video card similar to the one we are using from the bins of the Motherland and test it in these applications in SLI mode. Simply because there is no point in changing the environment between changes to the version of the methodology for obvious reasons - then it is better to redo and update everything. And, in fact, there’s not much to change - until recently, the GTX 570 continued to be among the top ten most productive solutions in our rating in the corresponding section, despite the fact that the first two lines in it were occupied by dual-chip solutions. But with a couple of cards, perhaps, you can measure at least something more interesting. If you want, of course, since the diagnosis is already clear, you just need to look at the detailed test results to come to the conclusion that a processor in this price segment will be enough for games with any video card. And if anything is not enough, then it’s probably a video card. In general, everyone is fast. Although in terms of points the fastest of all tested Core processors i7-3770K, which, however, was expected. As well as the fact that the Core i5-3570K will take third place, leaving behind only the Core i7-3960X Extreme Edition among the “oldies”.
Multitasking environment
This experimental test has recently demonstrated good stability and predictability, so we once again decided to use it to look at the subjects from this point of view.
But we didn’t find anything fundamentally new - the new i5 and i7 are faster than the old ones of the same family and at the same frequency (who would doubt it after all the above results), but this is not enough for any fundamental change in the situation on the market (LGA1155 it and there is LGA1155 in all its manifestations - no one doubted it either). What does it take for her to change? Obviously, “so”, not “tick”. However, even a big step is not always so big - the Core i5-2550K in this test is only capable of overtaking Junior Core i7 of the previous generation, like 860/920/930, but it’s capable. And a small step is unable to provide such an increase - the Core i7-870 was added to the list of defeated Core i5-3570K, but nothing more.
Total
We have a feeling that Intel came up with the new TDP class for a reason. More precisely, it may very well be that the company generally wanted to push 65 W and quad-core processors into the framework. But it was not possible to do this while maintaining the exact level of performance. And spending thermal reserves on increasing clock frequencies is also not the best solution: as we have seen more than once, great amount applications still cannot boast of any high-quality support for multi-threading, so even a Core i5 with a nominal frequency of 4 GHz (and a maximum frequency of 4.5) would begin to compete even with six-core SB-E. Well, maybe it’s not too obvious to compete, but the Core i7 is doing this even now, losing in the overall standings only due to the presence modern life(but not for all users) applications with “severe multithreading”.
At the same time, processors for LGA2011 have appeared quite recently, so everything would have turned out something like during the confrontation between LGA1155 and LGA1366 and even worse. Well, so - they showed a nominal increase compared to the previous generation, and did not spoil the sales of more expensive devices, and received additional profit (since the area of the new chip is only 160 mm², which is closer to 131 mm² of dual-core Sandy Bridge than to 216 mm² of quad-core , and almost half the size of the 315 mm² AMD Zambezi), and partners were pleased with the relaxation of cooling requirements in ready-made systems, as well as a slight reduction in prices (i5-3570K, for example, is cheaper in bulk not only than the i5-2550K model, but also the i5-2500K). What to users? Another opportunity speak out about the stagnation in the processor market and complain about the lack of competition :) And, naturally, those who purchased a system with one of the lower processors under LGA1155 with the expectation of an upgrade will be most upset - in fact, they could immediately buy the processor that was needed, and don't expect anything.
Because Intel managed to do what automakers do every year - release a new family of cars that are not radically different from last year's. No different from a consumer point of view, of course - for the company itself, the development of a new technological process is economically justified. Especially with an eye on the laptop market, where for the first time buyers will be pleased with the QM series Core i7 (i.e., quad-core) with a TDP of 35 W. GMA HD 4000 will be more than welcome there in all its manifestations. But from the point of view of the desktop market... The Ford Focus has been improved and refined several times, but it is a Ford Focus. LGA1155 also underwent a certain redesign, but remained the same LGA1155. The only difference from the car market is that there is no deterioration in the new year’s lineup at all: everything is either a little better, or simply no changes.
