I7 2nd generation processors. From Sandy Bridge to Coffee Lake: comparing seven generations of Intel Core i7. Best Intel processor with Skylake architecture

The first processors under the brand Intel Core i7 appeared nine years ago, but the LGA1366 platform did not pretend to be widely used outside the server segment. Actually, all “consumer” processors for it fell in the price range from ≈$300 to full-fledged “stubs,” so there is nothing surprising in this. However, modern i7s also live in it, so they are devices of limited demand: for the most demanding buyers (the appearance of the Core i9 this year has changed the disposition a little, but just a little). And already the first models of the family received the formula “four cores - eight threads - 8 MiB of third-level cache.”

Later, it was inherited by models for the mass market-oriented LGA1156. Later, without changes, it migrated to LGA1155. Even later, it appeared in LGA1150 and even LGA1151, although many users initially expected six-core processor models from the latter. But this did not happen in the first version of the platform - the corresponding Core i7 and i5 appeared only this year as part of the “eighth” generation, with the “sixth” and “seventh” incompatible. According to some of our readers (which we partially share), it’s a little late: it could have been earlier. However, the claim “good, but not enough” applies not only to processor performance, but in general to any evolutionary changes in any market. The reason for this lies not in a technical, but in a psychological plane, which goes far beyond the scope of interests of our site. Here's a test computer systems different generations to determine their performance and energy consumption (even if only on a limited sample of tasks) we can. That's what we'll do today.

Test bench configuration

CPU	Intel Core i7-880	Intel Core i7-2700K	Intel Core i7-3770K
Kernel name	Lynnfield	Sandy Bridge	Ivy Bridge
Production technology	45 nm	32 nm	22 nm
Core frequency, GHz	3,06/3,73	3,5/3,9	3,5/3,9
Number of cores/threads	4/8	4/8	4/8
L1 cache (total), I/D, KB	128/128	128/128	128/128
L2 cache, KB	4×256	4×256	4×256
L3 cache, MiB	8	8	8
RAM	2×DDR3-1333	2×DDR3-1333	2×DDR3-1600
TDP, W	95	95	77

Our parade is opened by the three oldest processors - one for LGA1156 and two for LGA1155. Note that the first two models are unique in their own way. For example, the Core i7-880 (appeared in 2010 - in the second wave of devices for this platform) was the most expensive processor of all participants in today's testing: its recommended price was $562. In the future, not a single desktop quad-core Core i7 cost that much. And the quad-core processors of the Sandy Bridge family (as in the previous case, we have here a representative of the second wave, and not the “starter” i7-2600K) are the only models for LGA115x that use solder as a thermal interface. In principle, no one noticed its introduction then, as well as the earlier transitions from solder to paste and vice versa: it was later that narrow but noisy circles began to endow the thermal interface with truly magical properties. Somewhere starting with the Core i7-3770K just (mid-2012), after which the noise did not subside.

CPU	Intel Core i7-4790K	Intel Core i7-5775C
Kernel name	Haswell	Broadwell
Production technology	22 nm	14 nm
Core frequency std/max, GHz	4,0/4,4	3,3/3,7
Number of cores/threads	4/8	4/8
L1 cache (total), I/D, KB	128/128	128/128
L2 cache, KB	4×256	4×256
L3 (L4) cache, MiB	8	6 (128)
RAM	2×DDR3-1600	2×DDR3-1600
TDP, W	88	65

What we will be somewhat missing today is the original Haswell in the form of the i7-4770K. As a result, we skip 2013 and go straight to 2014: formally, 4790K is already Haswell Refresh. Some people were already waiting for Broadwell, but the company released processors of this family exclusively for the tablet and laptop market: where they were most in demand. As for desktops, plans changed several times, but in 2015 a couple of processors (plus three Xeons) appeared on the market. Very specific: like Haswell and Haswell Refresh, they were installed in the LGA1150 socket, but were compatible only with a couple of 2014 chipsets, and most importantly, they turned out to be the only “socket” models with a four-level cache memory. Formally, for the needs of the graphics core, although in practice all programs can use L4. There were similar processors earlier and later - but only in BGA design (i.e., they were soldered directly to system board). These are unique in their own way. Enthusiasts, naturally, were not inspired due to low clock speeds and limited overclockability, but we will check how this “side shoot” relates to the main line in modern software.

CPU	Intel Core i7-6700K	Intel Core i7-7700K	Intel Core i7-8700K
Kernel name	Skylake	Kaby Lake	Coffee Lake
Production technology	14 nm	14 nm	14 nm
Core frequency, GHz	4,0/4,2	4,2/4,5	3,7/4,7
Number of cores/threads	4/8	4/8	6/12
L1 cache (total), I/D, KB	128/128	128/128	192/192
L2 cache, KB	4×256	4×256	6×256
L3 cache, MiB	8	8	12
RAM	2×DDR3-1600 / 2×DDR4-2133	2×DDR3-1600 / 2×DDR4-2400	2×DDR4-2666
TDP, W	91	91	95

And the most “fresh” trio of processors, formally using the same LGA1151 socket, but in two versions that are incompatible with each other. However, we wrote about the difficult path of mainstream six-core processors to the market quite recently: when they were tested for the first time. So we won't repeat ourselves. Let us only note that we tested the i7-8700K again: using not a preliminary, but a “release” copy, and even installing it on a “normal” board with debugged firmware. The results changed little, but in several programs they became somewhat more adequate.

CPU	Intel Core i3-7350K	Intel Core i5-7600K	Intel Core i5-8400
Kernel name	Kaby Lake	Kaby Lake	Coffee Lake
Production technology	14 nm	14 nm	14 nm
Core frequency, GHz	4,2	3,8/4,2	2,8/4,0
Number of cores/threads	2/4	4/4	6/6
L1 cache (total), I/D, KB	64/64	128/128	192/192
L2 cache, KB	2×256	4×256	6×256
L3 cache, MiB	4	6	9
RAM	2×DDR4-2400	2×DDR4-2400	2×DDR4-2666
TDP, W	60	91	65

With whom to compare the results? It seems to us that you should definitely take a couple of the fastest modern dual- and quad-core processors from the Core i3 and Core i5 lines, fortunately they have already been tested, and it will be interesting to see which of the old guys they will catch up with and where (and whether they will catch up with them). In addition, we managed to get our hands on a completely new six-core Core i5-8400, so we took the opportunity to test that too.

CPU	AMD FX-8350	AMD Ryzen 5 1400	AMD Ryzen 5 1600
Kernel name	Vishera	Ryzen	Ryzen
Production technology	32 nm	14 nm	14 nm
Core frequency, GHz	4,0/4,2	3,2/3,4	3,2/3,6
Number of cores/threads	4/8	4/8	6/12
L1 cache (total), I/D, KB	256/128	256/128	384/192
L2 cache, KB	4×2048	4×512	6×512
L3 cache, MiB	8	8	16
RAM	2×DDR3-1866	2×DDR4-2666	2×DDR4-2666
TDP, W	125	65	65

There is no way to do without AMD processors, and there is no need to. Including the “historical” FX-8350, which is the same age as the Core i7-3770K. Fans of this line have always claimed that it is not only cheaper, but generally better - just few people know how to cook it. But if you use the “correct programs”, it will immediately overtake everyone. Since this year we have just at the request of workers we reworked the testing methodology towards “severe multi-threading”, so there is a reason to test this hypothesis - the testing is still historical. And modern models will require at least two. The Ryzen 5 1500X would be very suitable for us, very similar to the old Core i7, but it was not tested. Ryzen 5 1400 is formally also suitable... but in fact, in this model (and in modern Ryzen 3), along with the halving of the cache memory, the connections between the CCX have also suffered. Therefore, I also had to take the Ryzen 5 1600, which does not have this problem - as a result, it often outperforms the 1400 by more than one and a half times. And a pair of six-core Intel processors are also present in today's testing. The others are clearly too slow to compare to this one inexpensive processor, Well, okay - let him dominate.

Testing methodology

Methodology. Let us briefly recall here that it is based on the following four pillars:

Methodology for measuring power consumption when testing processors
Methodology for monitoring power, temperature and processor load during testing
Methodology for measuring performance in games 2017

Detailed results of all tests are available in the form of a complete table with results (in Microsoft Excel 97-2003 format). In our articles, we use already processed data. This especially applies to application tests, where everything is normalized relative to the reference system (AMD FX-8350 with 16 GB of memory, GeForce GTX 1070 video card and Corsair Force LE 960 GB SSD) and grouped by computer application.

iXBT Application Benchmark 2017

In principle, the claims of AMD fans that in “severe multi-threaded” FX were not so bad, if we consider only performance, are justified: as we see, the 8350, in principle, could compete on equal terms with the Core i7 of the same year. However, here it looks good against the background of the younger Ryzen, but between these two families the company has produced practically nothing for this market segment. Intel, on the other hand, has a uniform line-up that has made it possible to double performance within the framework of the “quad-core” concept. Although the cores are of great importance here - the best dual-core of 2017 still did not catch up with the quad-core Core of the “previous” generation (let us remember that this is still officially called in the company’s materials, clearly distinguished from those numbered starting from the second). And six-core models are good - that’s all. So the accusations against Intel that the company delayed their entry into the market too much can be considered fair to some extent.

The only difference from the previous group is that the code here is not so primitive, so, in addition to cores, threads and gigahertz, the architectural features of the processors executing it are also important. Although the overall result for Intel products “offhand” is quite comparable: there is still a twofold difference between 880 and 7700K, the i5-8400 is still inferior only to the latter, and the i3-7350K still hasn’t caught up with anyone. And this happened within the same seven years. We can count it as eight - after all, LGA1156 entered the market in the fall of 2009, and the Core i7-880 differed from the 860 and 870 that appeared in the first wave only in frequencies, and even then only slightly.

All you have to do is slightly “weaken” the utilization of multithreading, and the position of newer processors immediately improves, albeit quantitatively weaker. However, the traditional “two ends”, all other things (relatively) being equal, comparison of the “previous” and “seventh” generations Core gives us. Although it is easy to notice that “revolutionary” is to the maximum extent drawn to “second” and... “eighth”. But this is more than understandable: the latter increased the number of cores, and in the “second” the microarchitecture and technical process radically changed, and at the same time.

As we already know, Adobe Photoshop is somewhat weird (the bad news is that the problem is not fixed in the latest version of the package; the very bad news is that now it will be relevant for the new Core i3), so we are not considering processors without HT. But our main characters have support for this technology, so no one bothers them all to work normally. As a result, in general, the situation is similar to other groups, but there is a nuance: the fastest processor for LGA1150 was not the high-frequency i7-4790K, but the i7-5775C. Well, in some places intensive methods of increasing productivity are very effective. It’s a pity that it’s not always the case: it’s easier to “work” with frequency. And cheaper: you don’t need an additional eDRAM chip, which also needs to be somehow placed on the same substrate as the “main one”.

The number of cores as a “driver” for increasing performance is also suitable - even more than the frequency. Although in our first testing the Core i7-8700K looked worse, this was due to the results of the same Adobe Photoshop: they turned out to be almost the same as for the i7-7700K. Switching to a “release” processor and board solved the problem in this case: the performance turned out to be similar to other six-core Intel processors. With a corresponding improvement in the overall result in the group. The behavior of other programs has not changed - they were previously positive about increasing the number of supported computation threads while maintaining a similar level of frequency.

Moreover, sometimes only it, and the number of computation threads, “decides”. Basically, of course, there are certain nuances here too, but “ there is no remedy against scrap" The entire revolutionary Ryzen architecture, for example, allowed the 1400 to only demonstrate performance at the level of the FX-8350 or Core i7-3770K, which entered the market in 2012. Considering that its frequency is lower than both, and in general it is a special budget model that actually uses only half of the semiconductor crystal, this is not so bad. But it doesn’t inspire reverence. Especially compared to another (and also inexpensive) representative of the Ryzen 5 line, which easily and noticeably outperformed any quad-core Core i7 of any year of production :)

Although we abandoned the single-threaded decompression test, this program still cannot be considered too “greedy” for cores and their frequencies. It’s clear why - the performance of the memory system is very important here, so the Core i7-5775C managed to outperform only the i7-8700K, and even then by less than 10%. It’s a pity that there are no products yet where L4 is combined with six cores and memory with high bandwidth: such a processor “without bottlenecks” could be used in such tasks show a miracle. Theoretically, at least, it is obvious that in desktop computers We won’t see anything like this in the near future for sure.

It is characteristic that this branch from the “main line” of desktop processors demonstrates (still!) high results in this group of programs. However, what unites them is mainly their intended purpose, and not the optimization methods chosen by programmers. But the latter are not ignored either - unlike some more “primitive” tasks, such as video encoding.

What do we end up with? The effect of “evolutionary development” has decreased somewhat: the Core i7-7700K outperforms the i7-880 by less than two times, and its superiority over the i7-2700K is only one and a half times. In general, it’s not bad: this was achieved by intensive means in comparable “quantitative” conditions, i.e., it can be extended to almost any software. However, in relation to the interests of the most demanding users - not enough. Especially if you compare the gains at each annual step, adding a Core i7-4770K (which is why we regretted above that this processor was not found).

At the same time, the company has long had the opportunity to dramatically increase productivity, at least in multi-threaded software (and there has long been a lot of this among resource-intensive programs). Yes, and it was also implemented - but within the framework of completely different platforms with their own characteristics. It’s not for nothing that many people have been waiting for six-core models for LGA115x since 2014... But many didn’t expect any breakthroughs from AMD in those years - all the more impressive were the first Ryzen tests. It’s not surprising - as we see, even the inexpensive Ryzen 5 1600 can compete in performance with the Core i7-7700K, which just a couple of months ago was the fastest processor for LGA1151. Now a similar level of performance is quite accessible with Core i5, but it would be better if this happened earlier :) In any case, there would be fewer reasons for complaints.

Energy consumption and energy efficiency

However, this diagram once again demonstrates why the performance of mass-produced central processors in the second decade of the 21st century grew at a much slower pace than in the first: in this case, all development occurred against the backdrop of a “non-increase” in energy consumption. If possible, even reduce it. It was possible to reduce it using architectural or some other methods - users of mobile and compact systems (of which much more have been sold for a long time than “standard desktop ones”) will be satisfied. And in the desktop market there is a small step forward, since you can increase the frequencies a little more, which was done in the Core i7-4790K at one time, and then became established in the “regular” Core i7, and even in the Core i5.

This is especially clearly seen when assessing the power consumption of the processors themselves (unfortunately, for LGA1155, measure it separately from the platform by simple means impossible). At the same time, it becomes clear why the company does not need to somehow change the cooling requirements for processors within the LGA115x line. Also why more and more products in the (formally) desktop assortment are starting to fit into the traditional thermal packages for laptop processors: this happens naturally without any effort. In principle, it would be possible to install all quad-core processors under LGA1151 TDP=65 W and not suffer :) Just for the so-called. For overclocking processors, the company considers it necessary to tighten the requirements for the cooling system, since there is a small (but not zero) probability that the buyer of a computer with one will overclock it and use all sorts of “stability tests”. But mass-produced products do not cause such concerns, and are initially more economical. Even six-core ones, although the power consumption of the older i7-8700K has grown - but only to the level of processors for LGA1150. In normal mode, of course - with overclocking you can inadvertently return to 2010 :)

But, at the same time, modern economical processors are not necessarily slow - three to five years ago, the performance of “energy efficient” models compared to the top ones in the line often left much to be desired, since they had to reduce the frequency too much, or even reduce the number of cores. Therefore, in general, “energy efficiency” increased at a much faster rate than pure performance: here, when comparing the Core i7-7700K and i7-880, not two times, but two and a half times. However... the first “big leap”, by one and a half times, came with the introduction of LGA1155, so it is not surprising that complaints about the further evolution of the platform were also heard from this direction.

iXBT Game Benchmark 2017

Of course, the results of the oldest processors, such as the Core i7-880 and i7-2700K, are of greatest interest. Unfortunately, nothing good happened with the first of them: apparently, none of the GPU manufacturers seriously dealt with the issues of compatibility of new video cards with the platform of the end of the last decade. And it’s clear why: many people skipped LGA1156 altogether, or have already managed to migrate from it to other solutions over so many years. But with the Core i7-2700K there is another problem: its performance (remember, in normal mode) is still often enough to work at the level of the new Core i7. In general, this is an indestructible legend: which (together with the older Core i5 for LGA1155) was first made a good gaming processor by its high single-threaded performance (in those years, Intel strongly “pressed” the Core i3 and Pentium in frequency), and then they began to work more or less effectively all eight supported computation threads will be disposed of. Although the same level of performance in games is often achieved by “simpler” solutions for new platforms, sometimes there is a feeling that this is connected not only and not so much with performance “in its pure form”. Therefore, for those who are to some extent interested in the results in games, we recommend that you familiarize yourself with them using the full table, and here we will present only a couple of the most interesting and indicative diagrams.

Here, for example, is Far Cry Primal. We immediately discard the results of the Core i7-880: obvious incorrect work GTX 1070 video cards with this platform. Perhaps, by the way, this also applies to LGA1155, although in general the frame rate here cannot be called low: in practice it is enough. But clearly lower than it could be. And LGA1151 too somehow doesn't shine, and the best platform seems to be LGA1150. Now we remember that a modified version of the Dunia Engine 2 (here it is used) was developed between 2013 and 2014, so they could just further optimize. An indirect confirmation of this is the low (relative to expected) frame rate on Ryzen 5: there is a feeling that there should be more and that's it.

But games on the EGO 4.0 engine began to appear in 2015 - and here we no longer see such artifacts. With the exception of the Core i7-880, which once again amused us with its “brakes,” but this correlates well with other games. And the best look is not just multi-core processors, but also released since 2015, i.e. LGA1151 and AM4 platforms. The exact opposite of the previous case, although in general both games were released in 2016. And both within the same family of processors always “vote” for the model that has more computing cores. But within one- different (especially significantly different architecturally) they need to be compared very carefully. If you want to compare, of course: in general, you can play both (and not only them) on a system with a five-year-old processor and a “good” video card with much greater comfort than with any processor, but on a budget video card for $200 In general, whether games require processors to grow or not, a gaming computer needs to be assembled “from the video card.” However, it would be strange if something changed in this industry - especially considering that the performance of video cards over the past eight years has not doubled or even tripled;)

Total

Actually, all we wanted to do was compare several processors from different years at once when working with a modern software. Moreover, some characteristics of older Core i7 models have remained virtually unchanged during this time, especially if we take the interval from the winter of 2011 to the same period in 2017. But productivity grew at the same time - slowly, but slightly more than the often discussed “5% per year.” And taking into account the fact that a normal user does not buy computers every year, but usually focuses on 3-5 years - over such a period, there has been an increase in productivity, efficiency, and functionality of the platform. But could have been better. At the same time, some “weak points” are clearly visible: for example, increasing the clock frequency in 2014 did not allow achieving significantly higher performance either in 2015 or even at the beginning of 2017. It was possible to “break away” noticeably from LGA1155 (as the software was optimized for processors starting with Haswell - at the start the results were more modest), and that’s all. And then (suddenly) +30% productivity, which hasn’t happened for a long time. In general, from a historical point of view, a smoother implementation of this process would have looked better. But what has happened has already happened.

An advanced gamer knows that buying a powerful video card without a modern and powerful processor is a waste of money. That is why it is worth purchasing a modern multi-core CPU for GeForce 20 series video adapters. Looking for a ready-made computer with intel i7? Then be sure to check out the models presented in our catalog.

Key advantages of the intel core i7 processor line

from six physical cores;
multithreading;
high operating frequency;
large amount of third-level cache memory.

Computers with Intel 7 series are able to offer gamers Turbo boost technology, which increases the operating clock frequency. The performance of Core i7 is enough to unleash the potential of any video card. It is worth noting that there are games that place a significant load on the processor. To have stable 60 FPS in such projects, you need to choose an i7 gaming computer.

Don't forget that Intel Core i7 models with the "K" index can be overclocked. Thanks to this, you can significantly improve system performance. Particularly relevant for clients working in graphics applications. Some programs use the computing power of the CPU, floating point operations, complex engineering calculations, and object modeling.

IntroductionThis summer, Intel did something strange: it managed to change as many as two generations of processors aimed at commonly used personal computers. At first, Haswell was replaced by processors with the Broadwell microarchitecture, but then within just a couple of months they lost their status as new products and gave way to Skylake processors, which will remain the most progressive CPUs for at least another year and a half. This leapfrog with the change of generations occurred mainly in connection with the problems Intel encountered when introducing the new 14-nm process technology, which is used in the production of both Broadwell and Skylake. Productive carriers of the Broadwell microarchitecture were greatly delayed on their way to desktop systems, and their successors were released according to a pre-planned schedule, which led to a crumpled announcement of the fifth generation Core processors and a serious reduction in their life cycle. As a result of all these upheavals, in the desktop segment Broadwell occupied a very narrow niche of economical processors with a powerful graphics core and are now content with only a small level of sales typical of highly specialized products. The attention of the advanced part of users switched to the followers of Broadwell - Skylake processors.

It should be noted that over the past few years, Intel has not been pleasing its fans with the growth in performance of its products. Each new generation of processors adds only a few percent in specific performance, which ultimately leads to a lack of clear incentives for users to upgrade older systems. But the release of Skylake - a generation of CPUs along the way to which Intel actually jumped over a step - inspired certain hopes that we would get a truly worthwhile update to the most common computing platform. However, nothing like this happened: Intel performed in its usual repertoire. Broadwell was introduced to the public as a sort of offshoot from the main line of desktop processors, and Skylake turned out to be marginally faster than Haswell in most applications.

Therefore, despite all expectations, the appearance of Skylake on sale aroused skepticism among many. After reviewing the results of real tests, many buyers simply did not see the real point in switching to sixth-generation Core processors. Indeed, the main trump card of the new CPUs is primarily a new platform with accelerated internal interfaces, but not a new processor microarchitecture. And this means that Skylake offers few real incentives to update legacy systems.

However, we still would not dissuade all users without exception from switching to Skylake. The fact is that even though Intel is increasing the performance of its processors at a very restrained pace, four generations of microarchitecture have already passed since the advent of Sandy Bridge, which are still working in many systems. Each step along the path of progress has contributed to an increase in performance, and today Skylake is able to offer quite a significant increase in performance compared to its earlier predecessors. Just to see this, you need to compare it not with Haswell, but with earlier representatives of the Core family that appeared before it.

Actually, this is exactly the comparison we will do today. Considering all that has been said, we decided to see how much the performance of Core i7 processors has increased since 2011, and we collected older Core i7s belonging to the Sandy Bridge, Ivy Bridge, Haswell, Broadwell and Skylake generations in a single test. Having received the results of such testing, we will try to understand which processor owners should start upgrading older systems, and which of them can wait until subsequent generations of CPUs appear. Along the way, we will look at the performance level of the new Core i7-5775C and Core i7-6700K processors of the Broadwell and Skylake generations, which have not yet been tested in our laboratory.

Comparative characteristics of the tested CPUs

From Sandy Bridge to Skylake: Specific Performance Comparison

In order to remember how the specific performance of Intel processors has changed over the last five years, we decided to start with a simple test in which we compared the operating speed of Sandy Bridge, Ivy Bridge, Haswell, Broadwell and Skylake, reduced to the same frequency 4 .0 GHz. In this comparison, we used processors from the Core i7 line, that is, quad-core processors with Hyper-Threading technology.

The complex test SYSmark 2014 1.5 was taken as the main testing tool, which is good because it reproduces typical user activity in common office applications, when creating and processing multimedia content and when solving computing problems. The following graphs display the results obtained. For ease of perception, they are normalized; the performance of Sandy Bridge is taken as 100 percent.

The integral indicator SYSmark 2014 1.5 allows us to make the following observations. The transition from Sandy Bridge to Ivy Bridge increased specific productivity only slightly - by about 3-4 percent. The next step to Haswell was much more effective, resulting in a 12 percent improvement in performance. And this is the maximum increase that can be observed in the above graph. After all, Broadwell is ahead of Haswell by only 7 percent, and the transition from Broadwell to Skylake even increases specific productivity by only 1-2 percent. All the progress from Sandy Bridge to Skylake results in a 26 percent increase in performance at constant clock speeds.

A more detailed explanation of the obtained SYSmark 2014 1.5 indicators can be found in the following three graphs, where the integral performance index is broken down into components by application type.

Please note that with the introduction of new versions of microarchitectures, multimedia applications increase execution speed most noticeably. In them, the Skylake microarchitecture outperforms Sandy Bridge by as much as 33 percent. But in counting problems, on the contrary, progress is least evident. Moreover, with such a load, the step from Broadwell to Skylake even results in a slight decrease in specific performance.

Now that we have an idea of what has happened to the specific performance of Intel processors over the past few years, let's try to figure out what caused the observed changes.

From Sandy Bridge to Skylake: what has changed in Intel processors

We decided to make the representative of the Sandy Bridge generation the starting point for comparing different Core i7s for a reason. It was this design that laid a strong foundation for all further improvements in high-performance Intel processors up to today's Skylake. Thus, representatives of the Sandy Bridge family became the first highly integrated CPUs, in which both computing and graphics cores were assembled in one semiconductor chip, as well as north bridge with L3 cache and memory controller. In addition, they were the first to use an internal ring bus, through which the problem of highly efficient interaction of all structural units that make up such a complex processor was solved. These universal design principles embedded in the Sandy Bridge microarchitecture continue to be followed by all subsequent generations of CPUs without any major adjustments.

The internal microarchitecture of the computing cores has undergone significant changes in Sandy Bridge. It not only implemented support for the new AES-NI and AVX instruction sets, but also found numerous major improvements in the bowels of the execution pipeline. It was in Sandy Bridge that a separate level-0 cache was added for decoded instructions; a completely new instruction reordering unit has appeared, based on the use of a physical register file; Branch prediction algorithms have been significantly improved; and in addition, two of the three execution ports for working with data have become unified. Such diverse reforms, carried out simultaneously at all stages of the pipeline, made it possible to significantly increase the specific productivity of Sandy Bridge, which immediately increased by almost 15 percent compared to the previous generation Nehalem processors. Added to this was a 15% increase in nominal clock frequencies and excellent overclocking potential, resulting in a family of processors that is still held up by Intel as an exemplary embodiment of the “so” phase in the company’s pendulum development concept.

Indeed, we have not seen improvements in microarchitecture similar in scale and effectiveness since Sandy Bridge. All subsequent generations of processor designs make much smaller improvements in the computing cores. Perhaps this is a reflection of the lack of real competition in the processor market, perhaps the reason for the slowdown in progress lies in Intel's desire to focus on improving the graphics cores, or perhaps Sandy Bridge simply turned out to be such a successful project that its further development requires too much effort.

The transition from Sandy Bridge to Ivy Bridge perfectly illustrates the decline in innovation intensity. Despite the fact that the next generation of processors after Sandy Bridge was transferred to a new production technology with 22 nm standards, its clock speeds did not increase at all. The improvements made in the design mainly affected the more flexible memory controller and bus controller PCI Express, which received compatibility with the third version this standard. As for the microarchitecture of the computing cores itself, some cosmetic changes made it possible to speed up the execution of division operations and slightly increase the efficiency of Hyper-Threading technology, and that’s all. As a result, the increase in specific productivity was no more than 5 percent.

At the same time, the introduction of Ivy Bridge also brought something that the million-strong army of overclockers now bitterly regrets. Starting with processors of this generation, Intel abandoned the pairing of the semiconductor chip of the CPU and the cover that covers it using flux-free soldering and switched to filling the space between them with a polymer thermal interface material with very dubious thermal conductive properties. This artificially worsened the frequency potential and made Ivy Bridge processors, like all their successors, noticeably less overclockable compared to the very vigorous “oldies” Sandy Bridge in this regard.

However, Ivy Bridge is just a “tick”, and therefore no one promised any special breakthroughs in these processors. However, the next generation, Haswell, which, unlike Ivy Bridge, already belongs to the “so” phase, did not bring any encouraging growth in productivity. And this is actually a little strange, since a lot of various improvements have been made in the Haswell microarchitecture, and they are dispersed across different parts of the execution pipeline, which in total could well increase the overall speed of command execution.

For example, in the input part of the pipeline, the performance of branch prediction was improved, and the queue of decoded instructions began to be dynamically divided between parallel threads coexisting within the Hyper-Threading technology. At the same time, there was an increase in the window for out-of-order execution of commands, which in total should have increased the share of code executed in parallel by the processor. Two additional functional ports were added directly to the execution unit, aimed at processing integer commands, servicing branches and storing data. Thanks to this, Haswell became capable of processing up to eight micro-operations per clock cycle - a third more than its predecessors. Moreover, the new microarchitecture has doubled the bandwidth of the first and second level cache memory.

Thus, improvements in the Haswell microarchitecture did not affect only the speed of the decoder, which seems to have become the biggest bottleneck in modern Core processors at the moment. Indeed, despite the impressive list of improvements, the increase in specific productivity for Haswell compared to Ivy Bridge was only about 5-10 percent. But in fairness, it must be noted that in vector operations the acceleration is noticeably much stronger. And the greatest gains can be seen in applications that use the new AVX2 and FMA commands, support for which also appeared in this microarchitecture.

Haswell processors, like Ivy Bridge, were also not particularly liked by enthusiasts at first. Especially considering the fact that in the original version they did not offer any increase in clock frequencies. However, a year after its debut, Haswell began to seem noticeably more attractive. First, there has been an increase in the number of applications that take advantage of the architecture's greatest strengths and use vector instructions. Secondly, Intel was able to correct the situation with frequencies. Later Haswell modifications, codenamed Devil's Canyon, were able to increase their advantage over their predecessors by increasing the clock speed, which finally broke through the 4-GHz ceiling. In addition, following the lead of overclockers, Intel has improved the polymer thermal interface under the processor cover, which makes Devil's Canyon more suitable for overclocking. Of course, not as pliable as Sandy Bridge, but still.

And with such baggage, Intel approached Broadwell. Since the main key feature These processors were supposed to be a new production technology with 14 nm standards; no significant innovations in their microarchitecture were planned - it was supposed to be almost the most banal “tick”. Everything necessary for the success of new products could well be provided by just one thin technical process with second-generation FinFET transistors, which in theory allows reducing power consumption and raising frequencies. However, the practical implementation of the new technology resulted in a series of failures, as a result of which Broadwell only gained efficiency, but not high frequencies. As a result, those processors of this generation that Intel introduced for desktop systems came out more like mobile CPUs than successors to Devil’s Canyon. Moreover, in addition to reduced thermal packages and rolled back frequencies, they differ from their predecessors in having a smaller L3 cache, which, however, is somewhat compensated by the appearance of a fourth-level cache located on a separate chip.

At the same frequency as Haswell, Broadwell processors demonstrate approximately a 7 percent advantage, provided by both the addition of an additional level of data caching and another improvement in the branch prediction algorithm along with an increase in the main internal buffers. In addition, Broadwell implements new and faster schemes for executing multiply and divide instructions. However, all these small improvements are negated by the clock speed fiasco, which takes us back to the pre-Sandy Bridge era. For example, the older overclocker Core i7-5775C of the Broadwell generation is inferior in frequency to the Core i7-4790K by as much as 700 MHz. It is clear that it is pointless to expect any increase in productivity against this background, as long as there is no serious drop in productivity.

Largely because of this, Broadwell turned out to be unattractive to the majority of users. Yes, processors of this family are highly economical and even fit into a thermal package with a 65-watt frame, but who really cares about that? The overclocking potential of the first generation 14nm CPU turned out to be quite restrained. There is no talk of any operation at frequencies approaching the 5-GHz bar. The maximum that can be achieved from Broadwell when used air cooling lies in the vicinity of 4.2 GHz. In other words, Intel's fifth generation Core turned out to be, at least, strange. Which, by the way, the microprocessor giant eventually regretted: Intel representatives note that the late release of Broadwell for desktop computers, its shortened life cycle and atypical characteristics had a negative impact on sales, and the company does not plan to undertake any more such experiments.

Against this background, the newest Skylake appears not so much as a further development of Intel microarchitecture, but as a kind of work on mistakes. Despite the fact that this generation of CPU uses the same 14nm process technology as Broadwell, Skylake does not have any problems with operating at high frequencies. The nominal frequencies of the sixth generation Core processors have returned to those that were characteristic of their 22-nm predecessors, and the overclocking potential has even increased slightly. The fact that in Skylake the processor power converter again migrated to motherboard and thereby reduced the total heat generation of the CPU during overclocking. The only pity is that Intel never returned to using an effective thermal interface between the die and the processor cover.

But as for the basic microarchitecture of computing cores, despite the fact that Skylake, like Haswell, is the embodiment of the “so” phase, there are very few innovations in it. Moreover, most of them are aimed at expanding the input part of the executive pipeline, while the remaining parts of the pipeline remained without any significant changes. The changes relate to improving the performance of branch prediction and increasing the efficiency of the prefetch unit, and that’s all. At the same time, some of the optimizations serve not so much to improve performance, but are aimed at further increasing energy efficiency. Therefore, one should not be surprised that Skylake is almost no different from Broadwell in its specific performance.

However, there are exceptions: in some cases, Skylake can outperform its predecessors in performance and more noticeably. The fact is that the memory subsystem has been improved in this microarchitecture. The on-chip ring bus became faster, and this ultimately increased the bandwidth of the L3 cache. Plus, the memory controller received support for high-frequency DDR4 SDRAM memory.

But in the end, it turns out that no matter what Intel says about the progressiveness of Skylake, from the point of view of ordinary users this is a rather weak update. The main improvements in Skylake are made in the graphics core and in energy efficiency, which opens the way for such CPUs to fanless systems of the tablet form factor. Desktop representatives of this generation do not differ too noticeably from those of Haswell. Even if we close our eyes to the existence of the intermediate generation Broadwell, and compare Skylake directly with Haswell, the observed increase in specific productivity will be about 7-8 percent, which can hardly be called an impressive manifestation of technical progress.

Along the way, it is worth noting that the improvement of technological production processes does not live up to expectations. On the way from Sandy Bridge to Skylake, Intel changed two semiconductor technologies and reduced the thickness of transistor gates by more than half. However, the modern 14-nm process technology, compared to the 32-nm technology of five years ago, has not made it possible to increase the operating frequencies of processors. All Core processors of the last five generations have very similar clock speeds, which, if they exceed the 4-GHz mark, are very small.

To clearly illustrate this fact, you can look at the following graph, which displays the clock speed of older overclocking Core i7 processors of different generations.

Moreover, the peak clock speed does not even occur on Skylake. Haswell processors belonging to the Devil’s Canyon subgroup can boast the maximum frequency. Their nominal frequency is 4.0 GHz, but thanks to the turbo mode in real conditions they are capable of accelerating to 4.4 GHz. For modern Skylake, the maximum frequency is only 4.2 GHz.

All this, naturally, affects the final performance of real representatives of various CPU families. And then we propose to see how all this is reflected in the performance of platforms built on the basis of flagship processors from each of the Sandy Bridge, Ivy Bridge, Haswell, Broadwell and Skylake families.

How we tested

The comparison involved five Core i7 processors of different generations: Core i7-2700K, Core i7-3770K, Core i7-4790K, Core i7-5775C and Core i7-6700K. Therefore, the list of components involved in testing turned out to be quite extensive:

Processors:

Intel Core i7-2600K (Sandy Bridge, 4 cores + HT, 3.4-3.8 GHz, 8 MB L3);
Intel Core i7-3770K (Ivy Bridge, 4 cores + HT, 3.5-3.9 GHz, 8 MB L3);
Intel Core i7-4790K (Haswell Refresh, 4 cores + HT, 4.0-4.4 GHz, 8 MB L3);
Intel Core i7-5775C (Broadwell, 4 cores, 3.3-3.7 GHz, 6 MB L3, 128 MB L4).
Intel Core i7-6700K (Skylake, 4 cores, 4.0-4.2 GHz, 8 MB L3).

CPU cooler: Noctua NH-U14S.
Motherboards:

ASUS Z170 Pro Gaming (LGA 1151, Intel Z170);
ASUS Z97-Pro (LGA 1150, Intel Z97);
ASUS P8Z77-V Deluxe (LGA1155, Intel Z77).

Memory:

2x8 GB DDR3-2133 SDRAM, 9-11-11-31 (G.Skill F3-2133C9D-16GTX);
2x8 GB DDR4-2666 SDRAM, 15-15-15-35 (Corsair Vengeance LPX CMK16GX4M2A2666C16R).

Video card: NVIDIA GeForce GTX 980 Ti (6 GB/384-bit GDDR5, 1000-1076/7010 MHz).
Disk subsystem: Kingston HyperX Savage 480 GB (SHSS37A/480G).
Power supply: Corsair RM850i (80 Plus Gold, 850 W).

Testing was performed on the operating system Microsoft Windows 10 Enterprise Build 10240 using the following driver set:

Intel Chipset Driver 10.1.1.8;
Intel Management Engine Interface Driver 11.0.0.1157;
NVIDIA GeForce 358.50 Driver.

Performance

Overall Performance

To evaluate processor performance in common tasks, we traditionally use the Bapco SYSmark test package, which simulates user work in real common modern office programs and applications for creating and processing digital content. The idea of the test is very simple: it produces a single metric characterizing the weighted average speed of the computer during everyday use. After release operating system Windows 10 this benchmark has been updated once again, and now we use the most latest version– SYSmark 2014 1.5.

When comparing Core i7 of different generations, when they work in their nominal modes, the results are not at all the same as when compared at a single clock frequency. Still, the actual frequency and operating features of the turbo mode have a fairly significant impact on performance. For example, according to the data obtained, the Core i7-6700K is faster than the Core i7-5775C by as much as 11 percent, but its advantage over the Core i7-4790K is very insignificant - it is only about 3 percent. At the same time, we cannot ignore the fact that the newest Skylake turns out to be significantly faster than processors of the Sandy Bridge and Ivy Bridge generations. Its advantage over the Core i7-2700K and Core i7-3770K reaches 33 and 28 percent, respectively.

A deeper understanding of the SYSmark 2014 1.5 results can be provided by familiarizing yourself with the performance estimates obtained in various system usage scenarios. The Office Productivity scenario simulates typical office work: word preparation, processing spreadsheets, work with by email and visiting Internet sites. The script uses the following set of applications: Adobe Acrobat XI Pro, Google Chrome 32, Microsoft Excel 2013, Microsoft OneNote 2013, Microsoft Outlook 2013, Microsoft PowerPoint 2013, Microsoft Word 2013, WinZip Pro 17.5 Pro.

The Media Creation scenario simulates the creation of a commercial using pre-shot digital images and videos. For this purpose, the popular packages Adobe Photoshop CS6 Extended, Adobe Premiere Pro CS6 and Trimble SketchUp Pro 2013 are used.

The Data/Financial Analysis scenario is devoted to statistical analysis and investment forecasting based on a certain financial model. The scenario uses large amounts of numerical data and two applications: Microsoft Excel 2013 and WinZip Pro 17.5 Pro.

The results we obtained under various load scenarios qualitatively repeat the general indicators of SYSmark 2014 1.5. The only noteworthy fact is that the Core i7-4790K processor does not look outdated at all. It noticeably loses to the latest Core i7-6700K only in the Data/Financial Analysis calculation scenario, and in other cases it is either inferior to its successor by a very insignificant amount, or is generally faster. For example, a member of the Haswell family is ahead of the new Skylake in office applications. But the older processors, Core i7-2700K and Core i7-3770K, already look like somewhat outdated offerings. They lose to the new product in different types tasks from 25 to 40 percent, and this is perhaps quite sufficient reason for the Core i7-6700K to be considered as a worthy replacement.

Gaming Performance

As is known, the performance of platforms equipped with high-performance processors is overwhelmingly modern games determined by the power of the graphics subsystem. That is why, when testing processors, we select the most processor-dependent games, and measure the number of frames twice. The first pass tests are carried out without turning on anti-aliasing and with settings that are far from the highest. Such settings allow you to evaluate how well processors perform with a gaming load in principle, and therefore allow you to speculate about how the tested computing platforms will behave in the future, when more quick options graphics accelerators. The second pass is performed with realistic settings - when selecting FullHD resolution and the maximum level of full-screen anti-aliasing. In our opinion, such results are no less interesting, since they answer the frequently asked question about what level of gaming performance processors can provide right now - in modern conditions.

However, in this testing we assembled a powerful graphics subsystem based on the flagship NVIDIA GeForce GTX 980 Ti video card. And as a result, in some games the frame rate showed a dependence on processor performance, even in FullHD resolution.

Results in FullHD resolution with maximum quality settings

Typically, the impact of processors on gaming performance, especially when it comes to powerful representatives of the Core i7 series, is insignificant. However, when comparing five Core i7s of different generations, the results are not at all uniform. Even when installed maximum settings graphics quality Core i7-6700K and Core i7-5775C demonstrate the highest gaming performance, while the older Core i7 lags behind them. Thus, the frame rate obtained in a system with a Core i7-6700K exceeds the performance of a system based on a Core i7-4770K by an unnoticeable one percent, but the Core i7-2700K and Core i7-3770K processors already seem to be a noticeably worse basis for a gaming system. Switching from a Core i7-2700K or Core i7-3770K to the latest Core i7-6700K gives an increase in fps of 5-7 percent, which can have a quite noticeable impact on the quality of the gameplay.

You can see all this much more clearly if you look at the gaming performance of processors at a reduced image quality, when the frame rate does not depend on the power of the graphics subsystem.

Results at reduced resolution

The latest Core i7-6700K processor once again manages to show the highest performance among all Core i7s of the latest generations. Its superiority over the Core i7-5775C is about 5 percent, and over the Core i7-4690K – about 10 percent. There is nothing strange about this: games are quite sensitive to the speed of the memory subsystem, and it is in this area that serious improvements have been made in Skylake. But the superiority of the Core i7-6700K over the Core i7-2700K and Core i7-3770K is much more noticeable. The older Sandy Bridge lags behind the new product by 30-35 percent, and Ivy Bridge loses to it by about 20-30 percent. In other words, no matter how much Intel is criticized for improving its own processors too slowly, the company has been able to increase the speed of its CPUs by a third over the past five years, and this is a very tangible result.

Testing in real games complete the results of the popular synthetic benchmark Futuremark 3DMark.

The results produced by Futuremark 3DMark echo the gaming indicators. When the microarchitecture of Core i7 processors was transferred from Sandy Bridge to Ivy Bridge, 3DMark scores increased by 2 to 7 percent. The introduction of the Haswell design and the release of Devil’s Canyon processors added an additional 7-14 percent to the performance of older Core i7s. However, then the appearance of the Core i7-5775C, which has a relatively low clock frequency, somewhat rolled back the performance. And the newest Core i7-6700K, in fact, had to take the rap for two generations of microarchitecture at once. The increase in the final 3DMark rating for the new Skylake family processor compared to the Core i7-4790K was up to 7 percent. And in fact, this is not so much: after all, Haswell processors have been able to bring the most noticeable improvement in performance over the past five years. The latest generations of desktop processors are indeed somewhat disappointing.

Tests in applications

In Autodesk 3ds max 2016 we test the final rendering speed. Measures the time it takes to render a single frame of a standard Hummer scene at 1920x1080 resolution using the mental ray renderer.

Another test of the final rendering is carried out by us using a popular free rendering package 3D graphics Blender 2.75a. In it we measure the time it takes to build the final model from Blender Cycles Benchmark rev4.

To measure the speed of photorealistic 3D rendering, we used the Cinebench R15 test. Maxon recently updated its benchmark, and now it again allows you to evaluate the speed of various platforms when rendering in current versions of the Cinema 4D animation package.

We measure the performance of websites and Internet applications built using modern technologies in the new browser Microsoft Edge 20.10240.16384.0. For this purpose, a specialized test, WebXPRT 2015, is used, which implements algorithms actually used in Internet applications in HTML5 and JavaScript.

Processing Performance Testing graphic images takes place in Adobe Photoshop CC 2015. Measures the average execution time of a test script that is a creative reworking of the Retouch Artists Photoshop Speed Test, which involves typical processing of four 24-megapixel images taken with a digital camera.

Due to numerous requests from amateur photographers, we conducted performance testing in graphics program Adobe Photoshop Lightroom 6.1. The test scenario involves post-processing and exporting to JPEG at 1920x1080 resolution and maximum quality of two hundred 12-megapixel RAW images taken with a Nikon D300 digital camera.

Adobe Premiere Pro CC 2015 tests performance for non-linear video editing. The time for rendering a Blu-Ray project containing HDV 1080p25 video with various effects applied is measured.

To measure the speed of processors when compressing information, we use the WinRAR 5.3 archiver, with which we archive a folder with various files with a total volume of 1.7 GB with the maximum compression ratio.

To evaluate the speed of video transcoding into the H.264 format, the x264 FHD Benchmark 1.0.1 (64bit) test is used, based on measuring the time the x264 encoder encodes the source video into MPEG-4/AVC format with a resolution of 1920x1080@50fps and default settings. It should be noted that the results of this benchmark have a huge practical significance, since the x264 encoder underlies numerous popular transcoding utilities, for example, HandBrake, MeGUI, VirtualDub, etc. We periodically update the encoder used for performance measurements, and this testing involved version r2538, which supports all modern instruction sets, including AVX2.

In addition, we have added to the list of test applications a new x265 encoder designed for transcoding video into the promising H.265/HEVC format, which is a logical continuation of H.264 and is characterized by more efficient compression algorithms. To evaluate performance, a source 1080p@50FPS Y4M video file is used, which is transcoded into H.265 format with a medium profile. The release of the encoder version 1.7 took part in this testing.

Advantage of Core i7-6700K over earlier predecessors in various applications there is no doubt. However, two types of problems have benefited most from the evolution that has occurred. Firstly, related to the processing of multimedia content, be it video or images. Secondly, the final rendering in 3D modeling and design packages. In general, in such cases, the Core i7-6700K outperforms the Core i7-2700K by at least 40-50 percent. And sometimes you can see a much more impressive improvement in speed. So, when transcoding video with the x265 codec, the latest Core i7-6700K produces exactly twice as much high performance than the old Core i7-2700K.

If we talk about the increase in the speed of performing resource-intensive tasks that the Core i7-6700K can provide compared to the Core i7-4790K, then there are no such impressive illustrations of the results of the work of Intel engineers. The maximum advantage of the new product is observed in Lightroom; here Skylake turned out to be one and a half times better. But this is rather an exception to the rule. In most multimedia tasks, the Core i7-6700K offers only a 10 percent improvement in performance compared to the Core i7-4790K. And under loads of a different nature, the difference in performance is even smaller or absent altogether.

Separately, I need to say a few words about the result shown by the Core i7-5775C. Due to its low clock speed, this processor is slower than the Core i7-4790K and Core i7-6700K. But do not forget that its key characteristic is efficiency. And he is quite capable of becoming one of best options in terms of specific productivity per watt of electricity consumed. We can easily verify this in the next section.

Energy consumption

Skylake processors are manufactured using a modern 14-nm process technology with second-generation 3D transistors, however, despite this, their thermal package has increased to 91 W. In other words, the new CPUs are not only “hotter” than the 65-watt Broadwell, but also exceed the calculated heat dissipation of Haswell, produced using 22-nm technology and coexisting within the 88-watt thermal package. The reason, obviously, is that the Skylake architecture was initially optimized not for high frequencies, but for energy efficiency and the ability to be used in mobile devices. Therefore, in order for desktop Skylake to receive acceptable clock frequencies lying in the vicinity of the 4-GHz mark, it was necessary to raise the supply voltage, which inevitably affected power consumption and heat dissipation.

However, Broadwell processors also did not have low operating voltages, so there is hope that the Skylake 91-watt thermal package was obtained due to some formal circumstances and, in fact, they will turn out to be no more voracious than their predecessors. Let's check!

The new Corsair RM850i digital power supply we use in our test system allows us to monitor the consumed and output electrical power, which is what we use for measurements. The following graph shows the total system consumption (without monitor), measured “after” the power supply and representing the sum of the power consumption of all components involved in the system. The efficiency of the power supply itself is not taken into account in this case. To correctly assess energy consumption, we have activated turbo mode and all available energy-saving technologies.

At idle, a quantum leap in the efficiency of desktop platforms occurred with the release of Broadwell. The Core i7-5775C and Core i7-6700K feature noticeably lower idle consumption.

But under the load of video transcoding, the most economical CPU options are the Core i7-5775C and Core i7-3770K. The latest Core i7-6700K consumes more. His energy appetite is at the level of the older Sandy Bridge. True, the new product, unlike Sandy Bridge, has support for AVX2 instructions, which require quite significant energy costs.

The following diagram shows the maximum consumption under load created by the 64-bit version of the LinX 0.6.5 utility with support for the AVX2 instruction set, which is based on the Linpack package, which is distinguished by its exorbitant energy appetites.

Once again, the Broadwell generation processor shows miracles of energy efficiency. However, if you look at how much power the Core i7-6700K consumes, it becomes clear that progress in microarchitectures has bypassed the energy efficiency of desktop CPUs. Yes, in the mobile segment, with the release of Skylake, new offerings have emerged with extremely tempting performance-to-power ratios, but the latest desktop processors continue to consume about the same amount as their predecessors consumed five years before today.

conclusions

Having tested the latest Core i7-6700K and compared it with several generations of previous CPUs, we again come to the disappointing conclusion that Intel continues to follow its unspoken principles and is not too keen on increasing the performance of desktop processors aimed at high-performance systems. And if, compared to the older Broadwell, the new product offers approximately a 15% improvement in performance due to significantly better clock frequencies, then in comparison with the older, but faster Haswell, it no longer seems as progressive. The difference in performance between the Core i7-6700K and Core i7-4790K, despite the fact that these processors are separated by two generations of microarchitecture, does not exceed 5-10 percent. And this is very little for the older desktop Skylake to be unambiguously recommended for updating existing LGA 1150 systems.

However, it would take a long time to get used to such minor steps by Intel in increasing the speed of processors for desktop systems. The increase in performance of new solutions, which lies approximately within these limits, is a long-established tradition. There have been no revolutionary changes in the computing performance of Intel CPUs aimed at desktop PCs for a very long time. And the reasons for this are quite clear: the company’s engineers are busy optimizing the microarchitectures being developed for mobile applications and, first of all, think about energy efficiency. Intel's success in adapting its own architectures for use in thin and light devices is undeniable, but adherents of classic desktops can only be content with small increases in performance, which, fortunately, have not yet completely disappeared.

However, this does not mean that the Core i7-6700K can only be recommended for new systems. Owners of configurations based on the LGA 1155 platform with processors of the Sandy Bridge and Ivy Bridge generations may well be thinking about upgrading their computers. In comparison with the Core i7-2700K and Core i7-3770K, the new Core i7-6700K looks very good - its weighted average superiority over such predecessors is estimated at 30-40 percent. In addition, processors with the Skylake microarchitecture can boast support for the AVX2 instruction set, which has now found widespread use in multimedia applications, and thanks to this, in some cases the Core i7-6700K turns out to be much faster. So, when transcoding video, we even saw cases where the Core i7-6700K was more than twice as fast as the Core i7-2700K!

Skylake processors also have a number of other advantages associated with the introduction of the new LGA 1151 platform accompanying them. And the point is not so much in the support for DDR4 memory that appeared in it, but in the fact that the new logic sets of the hundredth series finally received really high-speed connection to the processor and support for a large number of PCI Express 3.0 lanes. As a result, advanced LGA 1151 systems boast numerous fast interfaces for connecting drives and external devices that are not subject to any artificial bandwidth limitations.

Plus, when assessing the prospects of the LGA 1151 platform and Skylake processors, you need to keep one more thing in mind. Intel will not rush to bring the next generation of processors, known as Kaby Lake, to market. If you believe the available information, representatives of this series of processors in versions for desktop computers will appear on the market only in 2017. So Skylake will be with us for a long time, and the system built on it will be able to remain relevant for a very long period of time.

In the process of assembling or purchasing a new computer, users are always faced with a question. In this article we will look at Intel Core i3, i5 and i7 processors, and also tell you the difference between these chips and what is better to choose for your computer.

Difference No. 1. Number of cores and support for Hyper-threading.

Perhaps, The main difference between Intel Core i3, i5 and i7 processors is the number of physical cores and support for Hyper-threading technology, which creates two threads of computation for each actually existing physical core. Creating two computation threads per core allows for more efficient use of the processing power of the processor core. Therefore, processors with Hyper-threading support have some performance benefits.

The number of cores and support for Hyper-threading technology for most Intel Core i3, i5 and i7 processors can be summarized in the following table.

	Number of physical cores	Hyper-threading technology support	Number of threads
Intel Core i3	2	Yes	4
Intel Core i5	4	No	4
Intel Core i7	4	Yes	8

But there are exceptions to this table. Firstly, these are Intel Core i7 processors from their “Extreme” line. These processors can have 6 or 8 physical computing cores. At the same time, they, like all Core i7 processors, have support for Hyper-threading technology, which means the number of threads is doubled more quantity cores. Secondly, some exceptions include mobile processors(laptop processors). So, some Intel Core i5 mobile processors have only 2 physical cores, but at the same time have support for Hyper-threading.

It should also be noted that Intel has already planned to increase the number of cores in its processors. According to the latest news, Intel Core i5 and i7 processors with Coffee Lake architecture, scheduled for release in 2018, will each have 6 physical cores and 12 threads.

Therefore, you should not completely trust the table provided. If you are interested in the number of cores in a particular Intel processor, then it is better to check the official information on the website.

Difference No. 2. Cache memory size.

Also, Intel Core i3, i5 and i7 processors differ in cache memory size. The higher the processor class, the larger the cache memory it receives. Intel processors Core i7s get the most cache, Intel Core i5s have a little less, and Intel Core i3s have even less. Specific values should be looked at in the characteristics of the processors. But as an example, you can compare several processors from the 6th generation.

	Level 1 cache	Level 2 cache	Level 3 cache
Intel Core i7-6700	4 x 32 KB	4 x 256 KB	8 MB
Intel Core i5-6500	4 x 32 KB	4 x 256 KB	6 MB
Intel Core i3-6100	2 x 32 KB	2 x 256 KB	3 MB

You need to understand that a decrease in cache memory is associated with a decrease in the number of cores and threads. But, nevertheless, there is such a difference.

Difference number 3. Clock frequencies.

Typically, higher-end processors come with higher clock speeds. But, not everything is so simple here. It is not uncommon for Intel Core i3 to have higher frequencies than Intel Core i7. For example, let's take 3 processors from the 6th generation line.

	Clock frequency
Intel Core i7-6700	3.4 GHz
Intel Core i5-6500	3.2 GHz
Intel Core i3-6100	3.7 GHz

In this way, Intel is trying to maintain the performance of Intel Core i3 processors at the desired level.

Difference No. 4. Heat dissipation.

Another important difference between Intel Core i3, i5 and i7 processors is the level of heat dissipation. The characteristic known as TDP or thermal design power is responsible for this. This characteristic tells you how much heat the processor cooling system should dissipate. As an example, let's take the TDP of three 6th generation Intel processors. As can be seen from the table, the higher the processor class, the more heat it produces and the more powerful the cooling system is needed.

	TDP
Intel Core i7-6700	65 W
Intel Core i5-6500	65 W
Intel Core i3-6100	51 W

It should be noted that TDP tends to decrease. With each generation of processors, the TDP becomes lower. For example, the TDP of the 2nd generation Intel Core i5 processor was 95 W. Now, as we see, only 65 W.

Which is better Intel Core i3, i5 or i7?

The answer to this question depends on what kind of performance you need. The difference is in the number of cores, threads, cache and clock speeds creates a noticeable difference in performance between Core i3, i5 and i7.

Intel Core i3 processor - an excellent option for office or budget home computer. If you have a video card of the appropriate level, you can play computer games on a computer with an Intel Core i3 processor.
Intel Core i5 processor – suitable for a powerful work or gaming computer. A modern Intel Core i5 can handle any video card without any problems, so on a computer with such a processor you can play any games even at maximum settings.
The Intel Core i7 processor is an option for those who know exactly why they need such performance. A computer with such a processor is suitable, for example, for editing videos or conducting game streams.