Intel will soon begin shipping a new family of notebook processors. Code-named processors Kaby lake The 7th generation is of particular interest to those who are preparing to change the platform to a more productive one in the near future. Video coding enthusiasts will notice a significant difference in the benefits of the new processor. Movie fans will be truly satisfied when watching videos with a high bitrate. Gamers will be able to enjoy video games right on their laptops. All this is quite achievable with Intel 7th generation processors.
This month conference Intel Developer Forum gave you a taste of all the delights of the 7th generation processors. On the forum during a demo, the Dell XPS 13 laptop was able to handle super graphics in heavy video games using Intel's standard integrated graphics on the new platform. This is just an amazing achievement.
Thus, the announced debut of Intel, which took place on August 30, 2016, clearly demonstrated to us how these processors will be more productive than the entire processor market that exists now.
Here's what became known after the forum about the 7th generation Intel multi-core processors:
100 projects by the end of the year
At its Developer Forum, Intel announced that the entire line of 7th Gen processors is now available to leading PC makers and Intel partners, which means very promising new processor-based notebooks will be released by the end of the year. Chris Walker, Intel's general manager for mobile client platforms, said the new processors, ranging from 4.5 watts to 15 watts, will be the first to appear in notebooks, namely ultra-thin laptops. As previously reported, when 7th Gen processors first surfaced, there are already 100 projects underway with 7th Gen processors that will be available in Q4 2016.
The new processor family will expand to other markets as early as next year. So, in particular, in January, the 7th generation Intel processors are expected to appear in workstations, gaming systems and virtual reality.
The chips have a familiar architecture
Intel built the 7th generation processors on the same Skylake architecture as the 6th generation processors unveiled last year. So Intel didn't revolutionize by inventing a new architecture. Skylake was just a little tweaked to perfection.
In particular, Intel announced that it has improved the voltage of transistors on processors. The result is that the microarchitecture is more energy efficient and therefore 7th generation processors can offer performance gains over previous generations of Intel processors.
M5 and m7 kernels go off
Intel is changing the naming convention for low-power chips, eliminating the 4.5 watt Core m5 and m7 processors and converting them to the Core i5 and Core i7. The company hopes this change will help consumers, many of whom don't understand the difference between a Core i5 and a Core m5. However, 4.5-watt processors, also known as series chips Kaby lake, with the letter Y are similar in power. If you see Y at the end of the SKU, it is one of the chips previously known as m5 or m7 cores.
What's even more interesting is that Intel won't change the core brand for its entry-level Core m3 processors, which is the slowest and least expensive of the lineup. m... So, in order of performance, the 4.5-watt chips are named Core m3, Core i5 Y series, and Core i7 Y series.
Increase in productivity
You probably shouldn't throw away your 6th gen processor if you upgraded this year or last winter. Skylake definitely should not be changed in favor of one of the 7th generation processors of the same line. The replacement is justified only by increasing the processor index. But Intel says that if you decide to replace, you will get a tangible performance boost. Using the SYSmark benchmark suite to measure performance, Intel unveiled a 7th Gen Core i7-7500U processor that saw 12 percent faster performance gains than the 6th Gen Core i7-6500U. Testing WebXPRT 2015 showed a 19 percent performance improvement.
I don't think that even a 19% advantage will spur buyers to change their not so old and kind Skylake for Kaby Lake. Obviously, the increase in performance looks more significant when the comparison is made with processors of the 5th and 4th generations, which Intel is relying on to replace the processors. The new Core i5-7200U is 1.7x faster than its 5-year-old Core i5-2467M in SYSmark. On the 3DMark benchmark, the new processor is three times faster than the five-year-old processor.
Intel officials said the 7th Gen CPUs will be able to play demanding games at medium settings at 720p with integrated graphics or at 4K with a compatible graphics amplifier.
These chips are for video
Intel has taken notice of all 4K and 360-degree video we're consuming. In response, the chipmaker introduced a new video engine for its 7-Gen processor core that aims to handle whatever content requirements you might throw at it.
The new chips support HEVC hardware decoding of 10-bit color profile, which allows you to play 4K and UltraHD video without any lag. Intel has also added VP9 decoding for 7th Gen cores to help you work more efficiently when you watch 4K video and do other things at the same time.
The 7th generation cores will also be able to perform video conversion operations much faster than other processors. For example, according to Intel, you can transcode 1 hour of 4K video in just 12 minutes.
More energy efficiency
In terms of improving battery energy efficiency for laptops, Intel said the laptop with a 7th generation processor can run for 7 hours while streaming 4K or 4K 360 degree YouTube videos. Compared to the 6th generation cores, the performance advantage will average 4 hours in favor of the 7th generation. In terms of 4K video streaming, Intel promises all-day uptime, which is 9 and a half hours.
7th generation offers a range of other features
7th generation processors offer several other features designed to help your notebooks run more efficiently. For example, Intel Turbo Boost 2.0 technology. It is a feature that controls the performance of the processor and its power, such as automatically overclocking the processor when the CPU clock speed exceeds the rated performance.
Hyper-Threading Technology helps the processor complete tasks faster by providing two processing threads for each core.
7th generation processors also include technology Speed Shift which should make the applications run faster. This technology allows the processor to be more responsive to application requests to increase or decrease frequency for the best performance, thereby optimizing performance and efficiency. This is especially effective when applications require very short bursts of activity, such as browsing the web or retouching photos with numerous brush strokes in an image editor.
We've covered the "top" worst gaming graphics cards. Now, after the release of Coffee Lake, we can make a list of the worst processors, since nothing very important is observed on the CPU market until the end of the year. Of course, I will only consider the relevance of buying such processors now: if you already own one of the "stones" below, then it means that you clearly had your own reasons for taking it.
Intel Core i7-7740X and Core-i5 7640X (Kaby Lake-X) - welcome to 2010
It's mid-2017. AMD unveils the first honest eight-core desktop processor - Ryzen 7. Intel is unveiling new processors for its high-performance platform, now called Skylake-X and Kaby Lake-X. It can include solutions with 16 or even 18 cores, and the simplest representatives have ... stop, 4 cores ?! Hmm, how do they differ from the simple i5-7600K and i7-7700K? The frequencies are the same, the number of memory channels and PCIe lines is the same as the instruction sets. Unless the X-line has a built-in video core, but this is more a minus than a plus. Considering the fact that these processors are more expensive than non-xix counterparts, and motherboards based on the X299 chipset are expensive - there is absolutely no point in buying these "stones", and it is difficult to explain the meaning of their release - well, except that Intel has a lot of unnecessary 4-core crystals.
AMD FX - Goodbye Gaming Bulldozer 
The FX line, which was top-notch before Ryzen for almost seven years, can now safely retire. To tell the truth, even at the time of its release it was not top-end: and although the programs showed that the FX-8000 line had as many as 8 cores, in fact it was 4 APUs, and according to tests, the top FX turned out to be at the level of the best i5s, while the i7s were not achievable - that is why Intel then did not "get it out", continuing to release new processors with an increase of 5% in performance per generation. Before the release of the 4-thread Pentiums earlier this year, it made sense to buy the FX-4000 line - they were extremely cheap, but at the same time they made it possible to create a basic gaming system with video cards of the level of GTX 750 Ti and even GTX 950. But, alas, the new Pentiums turned out to be so are good at leaving junior FX out of work. Well, the senior representatives, FX-8000, AMD "finished off" themselves, releasing the younger Ryzen 3 at the same price and with higher performance and lower heat dissipation. So the FX line, which was once a good choice for mid-budget gaming builds, is now finally going to rest.
But still, these processors can be taken in one case - for the sake of an upgrade: for example, if you have an FX-4000 line, now is the time to upgrade to the FX-8000 - you will get a doubling of performance for a fairly small amount of money. Considering that the 8000 line pulls out video cards of the level of GTX 1060 or RX 580, you can easily play comfortably for another couple of years.
Most of the representatives of the Skylake and Kaby Lake lines - Intel smothers "old" 
Rumors that Intel should release desktop processors with a large number of cores have been floating around for a long time, and this happened, and since October 5, their tests have flooded the Internet. And, alas, they clearly show that the previous lines no longer have a place under the Sun: why take an 8-thread processor for 19 thousand rubles, if the younger 12-thread processor costs only 20.5 thousand, and even in overclocking the previous generation is worse by at least 20%? Similarly, with the i5, and even more so with the i3 of the 6th and 7th generations - the latter were already meaningless processors on the market after the appearance of the new Pentium, but now, after the release of the 4-core i3 of the 8th generation, the i3 Skylake and Kaby Lake can definitely be written off for scrap.
By the way, now the line of processors from Intel looks quite logical: the lowest-level are 2-core Celerons: they are quite enough for comfortable surfing the Internet and watching movies, and even simple games like Dota, WoT and CS: GO. The next step is Pentium, which has all the same 2 cores, but already 4 threads, and somewhat higher frequencies - on their basis, you can already assemble a low-middle-level gaming system. Core i3, which are now 4-core, take it one step higher, allowing you to build a middle-level build. Well, for the top, there are 6-core i5 and i7 - for those who want the best gaming solution on the market.
But, however, there is one reason why "old" processors should be taken, and it is still the same - an upgrade. For example, a couple of years ago you took a younger i5-6400. And now there is a good opportunity to upgrade it to the i7-7700K, and get a twofold increase in performance, and even not very expensive (especially if you sell an i5).
Haswell-E and Broadwell-E line - oldies at top prices 
Let's see how much the 8-core processor of the new Skylake-X line - Intel Core i7-7820X - costs. In Moscow retail, the price tag for it is about 40 thousand rubles. Expensive, you say? Well, here for this price we get 8 cores on a new architecture with a frequency of 4 GHz - quite good for a high-performance PC. Is it expensive anyway? Hmm, okay, let's take a look at the previous generation processors - they should be cheaper, right? So, the analog from Broadwell-E is the i7-6900X: also 8 cores, but on the previous architecture, and the frequencies are about 3.5 GHz. And the price ... 70 thousand rubles ?! Where? Why? Let's look for the pluses of the old processor. And yet, yes, we find one - this is a solder under the lid, which allows it to be overclocked better than representatives of Skylake-X with "mayonnaise" instead of solder. But even if you are very lucky and you overclock the i7-6900X so that it is on the same level as the i7-7820X, this will not remove the almost twofold difference in price.
As a result, Intel this year killed two old lines at once - Broadwell-E and Kaby Lake, and the last one is not even a year old. This is how it is, monopoly ...
AMD Ryzen with X - the company steps on the same rake 
Those who remember AMD FX processors know that there was no point in overpaying for older processors in the line - all processors could be overclocked, so the younger "stone" turned into an older one with one light wave of the hand. And for some reason, AMD continued this in Ryzen, and here it comes to the point of absurdity: for example, the younger Ryzen 7 1700 costs about 20 thousand rubles. The older 7 1800X already costs 30 thousand - one and a half times more. And they have the same overclocking potential - about 4 GHz. Should you overpay for 1800X? I think the answer is obvious. And so in all Ryzen lines - 3, 5 and 7 - it makes sense to take a younger processor, without the X index, and overclock it to the level of the older one.
AMD Bristol Ridge - For Those Who Don't Have Ryzen Money 
AMD continues to develop its APUs with the same persistence - two-in-one systems, where the average CPU level includes full-fledged graphics from AMD, only with a smaller number of computing units and frequency than in full-fledged video cards. In principle, it is quite a good solution for those who need a simple home PC - the processor's performance is enough for the OS, browser and movies to work quickly, and the GPU will allow you to play even new games, albeit in HD resolution and with low graphics settings. Well, and most importantly, the new APUs are compatible with AM4, that is, in the future, no one bothers to replace such a processor with some Ryzen 7, which is well suited for those who assemble a PC for themselves in stages.
But, on the other hand, yes, this is a budget solution, but why is it based on the Excavator architecture, which is 7 years old at lunchtime, and even 28 nm ?! Was it really so difficult to make these "stones" on Zen, which would also allow reducing the heat dissipation from 65 to 30 watts acceptable for such a system? In general, APUs are strange - on the one hand, new, on the other - ancient. But, in principle, they can find their buyers.
But something we started talking about desktops, it would be time to switch to mobile processors, because there are also a lot of strange "features" here.
Intel Celeron N3050 and N3350 - worse than Atom for the same money
For some reason, branded laptop manufacturers have one trick - we put Celeron and Pentium in netbooks / laptops, and Atom in tablets. It would seem that everything is correct, Celeron should be better than Atom, but no - Intel thinks differently: the architecture of these processors is similar, but Atom has 4 computing cores, when Celeron has only 2. Taking into account that we are considering the lowest -level (10-15 thousand rubles), a couple of cores will not be superfluous here, and if notebooks on Celeron may well start to freeze with 3-4 tabs in Chrome, Atom will be quite capable of simultaneously surfing and watching a PiP movie. And given the fact that for $ 150 you can simply remain silent about the quality of branded netbooks - it makes sense to take a solution from any Digma or iRu, but with Atom, and get seriously better performance for the same money.
Intel Core i3-6006U and Pentium 4405U - i3 worse than Pentium 
After Atom, which is better than Celeron, it would seem that it is much worse. However, they knocked on the bottom - the i3-6006U, which is quite massive in the segment of 18-25 thousand rubles ... is worse than its counterpart in the same segment, but from the Pentium mill! Let's take a closer look at these processors: both have 2 cores and 4 threads, the same instruction set, but the Pentium has a 100 MHz higher frequency, but the integrated graphics are twice as bad: HD 510 versus HD 520 in i3. It would seem that 100 MHz frequency (+ 5%) will definitely not outweigh twice the worst graphics, but there are two nuances:
- If a laptop has discrete graphics (and often it is - this is an Nvidia GT 920M), then it makes no difference to integrated graphics - it will be discrete graphics that will work in games, so a slightly higher-frequency Pentium is better here.
- If a person chooses a laptop without discrete graphics, then he does not need games, and both integrated video cards cope with GUI rendering and playback, including 1080p60, that is, again, there is no point in taking i3.
AMD mobile processors - Intel won the war
The fact that AMD did not really update its mobile processors for a couple of years, and Intel, even in low-voltage solutions, increased the number of cores to 4, led to the fact that it simply does not make sense to buy laptops with AMD processors - analogs on Intel processors will be more productive. and more autonomous. Yes, the "reds" do not want to lose the mobile market, and they are actively making mobile Ryzen, but so far the only thing on the Internet is a couple of tests, where processors from AMD again do not perform in the best light. Of course, when they come out, everything may change, but so far Intel reigns in the mobile segment. You can read more about this.
What's the bottom line? And in the end, the same confusion and hesitation as with video cards - there are excellent solutions, there are good ones, and there are those at the sight of which you think - and what was the manufacturer guided by when releasing of this?! But, what pleases me is that the processor market has seriously moved lately, and mainly thanks to AMD: Intel rolled out 6-core desktop processors in response to 8-core Ryzen, the number of cores in the same lines has also increased in the mobile segment. So those who wanted to upgrade or build a new PC - IMHO, it's time to get started.
Thanks to the improvement of the technical process, it was possible to achieve a significant increase in productivity, which will amount to more than 15% according to the SysMark test. Thus, this year the performance of the Core i7 processors will increase more than in the past. This is shown in the slide from the presentation above under the heading "Moving Moore's Law at 14nm".
The new generation of processors on the improved 14 nm platform is scheduled for release in the second half of 2017. These will be designated the Core i7 / i5 / i3-8000 family and will replace the existing 7th generation family.
At the presentation to investors, Intel did not say anything about plans to release the Cannonlake family (formerly called Skymont) - microprocessors based on a 10nm technological process. They are supposed to be out in late 2017, and a working 10nm Cannonlake prototype was shown recently at CES. It was the Cannonlake family that was previously positioned as the 8th generation of the processor architecture, which will replace Skylake as part of the tick-tock strategy. Now another family has appeared that has nothing to do with Cannonlake. Perhaps this is an attempt to sell an old product in a new package.
Cancellation of the "tick-tock" strategy
Intel has consistently followed a tick-tock strategy since 2006. Since then, it has released processors every two years using a new process technology, significantly increasing the number of transistors on a chip. Each transition to a new technical process was designated as a "tick", and the subsequent improvement of the microarchitecture with the same technical process - "so". The semiconductor giant has worked like a clock for ten years, delivering new architectures without a hitch.It seems that in 2016 Intel's "clock" shorted a little at 14nm, and the company announced about.
In principle, there is nothing wrong with that. Again, this year's chip performance (over 15%) will be even greater than last year (15%), Intel said. Maybe it's really better to squeeze all the reserves out of the existing technical process, optimizing it, and only then move on. We cannot criticize Intel for deviating from a strategy that it voluntarily set for itself.
One way or another, but now the "tick-tock" strategy has been modified in a different form.
Instead of a measured metronome, a new procedure with a greater emphasis on optimization is now implemented. Perhaps the new architecture will not be released every two years, as it was before.
Why is Intel not pushing for 10nm? She does not need to do this, because she believes that she is already far ahead in her technological superiority from competitors in the semiconductor industry (Samsung, TSMC and others). The company estimates this gap to be about three years.
Such a reserve allows you to feel quite confident.
New plant for 7 nm
The bright future of Moore's Law should be provided by the new Intel Fab 42 plant, which will be able to support production using the 7 nm process technology.
Construction and equipment will take another three to four years and will require significant investment. The plant in Chandler, Arizona will reduce the number of local unemployed by about 3,000 people (+ another 10,000 jobs will be added indirectly).
Construction of the Chandler plant began in 2011. It should become the most advanced and innovative semiconductor enterprise in the world. The building itself was completed in 2013, but instead of installing 14nm hardware in early 2014, Intel decided to postpone the launch of the pipeline. At the moment, the plant is ready: air conditioning, heating systems and others - everything is functioning, all that remains is to install and adjust the equipment. Intel does not plan to use this factory for production according to the 10 nm process technology, so in a few years, it is likely that they will master the production at the next 7 nm rate.
According to Intel, the equipment will cost about $ 7 billion. This is the cost of a modern industrial enterprise. It is not yet known what specific equipment will be needed. Perhaps Intel will start using deep ultraviolet photolithography (EUV) there.
At the dawn of the 2000s, Intel hoped that by 2005 processor frequencies would rise to 10 GHz, and they would operate at voltages below a volt. As we know, this did not happen. About ten years ago, Dennard's scaling law ceased to work, claiming that as the size of transistors decreased, the voltage applied to the gate could be reduced and the switching speed increased. Since then, rarely any processor gets a nominal operating frequency above 4 GHz, but there are more cores, the north bridge has migrated to the crystal from the motherboard, other optimizations and accelerations have appeared. Now Moore's law is also slowing down, an empirical observation that speaks of a constant increase in the number of transistors on a chip due to a decrease in their size.
In early January Intel officially introduced a new generation of processors Intel Core on architecture Kaby Lake... The update turned out to be rather strange, so today we will do without lengthy reasoning and will tell only about what you really need to know.
Fact one: no "Tick-tock"
For a long time, Intel followed a simple "tick-tock" processor upgrade scheme. In one year, the technical process was updated, and in the next, a new architecture was released. For the first few years, the rhythm was maintained almost flawlessly, but in recent years the circuit began to noticeably break down. And with Kaby Lake, the manufacturer officially admitted that it is no longer possible to live with "tick-tock" and another stage is added to it, called "optimization", at which the already created crystals will be finished. Unfortunately, it was on this new stage that Kaby Lake got.
Why Intel decided to change itself is hard to say. According to the company itself, the high cost of switching to new technical processes is to blame for everything. However, we believe that the general decline in sales in the computer market is more likely to blame - it is becoming more and more difficult to beat off money with such short production cycles.
Fact two: architecture
Despite the new name and the solid word "optimization", technically and structurally, Kaby Lake is an exact copy of last year's Skylake. The structure of chips, memory structure, logic of work, sets of instructions - everything remains the same. Even the numerical indicators have not changed: a maximum of four cores, 8 MB of cache and 16 PCIe lanes for communicating with the video card. In general, apart from the name, there is no innovation.
Fact three: process technology
The technical process also remained unchanged. Kaby Lake is manufactured at the same 14nm rate. Only now a plus sign (14 nm +) is attributed to their name, behind which there are indeed some updates. In Kaby Lake, the transistors have slightly increased the height of the fins and the distance between them. As a result, the leakage currents and heat dissipation slightly decreased, and this made it possible to increase the frequency of the crystals.
Fact four: frequency of operation
 |  |
| The official frequency record for the Core i7-7700K is 7383 MHz. Installed, by the way, by a Russian team on the ASUS Maximus IX Apex motherboard. | |
Compared to the previous generation processors, the frequency of the new crystals increased by an average of 200-300 MHz. At the same time, the TDP of the models remained the same. That is, with the same 90 W, a new Core i7-7700K takes the bar at 4.5 GHz, while the i7-6700K only went up to 4.2 GHz.
Moreover, the processors are also better overclocked. If from Skylake it was possible to squeeze out 4.4-4.5 GHz on average, then 4.8 GHz is considered the norm for Kaby Lake, and with a good combination of circumstances, 5 GHz. And yes, now we are talking about working under conventional air coolers.
We also note that, as before, all Intel Core and Pentium crystals can be overclocked via the bus, and models with the “K” index are also driven by a multiplier. By the way, unlocked crystals are now available not only in the Core i5 and Core i7 series, but also in the Core i3. And the family Pentium, the cheapest Kaby Lake, now supports Hyper-Threading.
Fact five: embedded kernel
Remained in Kaby Lake and integrated graphics. But if earlier it was Intel HD Graphics 530, now it is HD Graphics 630 ... Evolution? Far from it, the same 24 units with a frequency of 1150 MHz are on board. The new number in the title was registered thanks to the updated media engine Quick Sync... It can now decode H.265 and VP.9 video on the fly. In other words, if you are a keen connoisseur of 4K movies or are going to stream in this resolution, know that with Kaby Lake the processor will no longer be 100% loaded.
As for the performance of the graphics itself, it's a sin to complain about it. She copes with drawing Windows without problems, and as a bonus, she also pulls not particularly demanding toys. You can also a village in Rim world build, and a prison in Prison Architect grow off, and even in DOTA 2 drive. The latter in Full HD and at medium settings produces quite decent 62 fps.
 |  |
Fact six: chipsets
Along with Kaby Lake, Intel also introduced new 200-series chipsets. However, there are as few changes in them as in the processors. Older models, the Z270, received an additional four PCIe lanes, to which motherboard manufacturers can tie up extra USB or M.2 ports. Frankly, the list is not particularly intriguing, but the scarcity is compensated to some extent by the motherboard manufacturers.
So, for example, the top-end ASUS Apex motherboards have DIMM.2 technology, which allows you to install two M.2 drives in the slot for the RAM. And to our test Maximus IX Formula, you could easily connect a custom "dropsy" to remove heat from the power circuits.
However, if none of these new products appeals to you, we have a pleasant fact in stock. They did not change the socket for Kaby Lake, leaving the already familiar LGA 1151. That is, the new processors work fine on old Z170 Express motherboards, but Skylake feels good on the Z270.
Fact seven: performance
| Test results | ||
| CPU | Intel Core i7-7700K | Intel Core i7-6700K |
| Cinebench R15 | ||
| One core | 196 | 175 |
| All cores | 988 | 897 |
| Multiplier | 5,05 | 5,11 |
| WinRar (KB / s) | ||
| One core | 2061 | 1946 |
| All cores | 11258 | 10711 |
| TrueCrypt (MB / s) | ||
| AES-Twofish-Serpent | 336 | 295 |
| PCMark (Work) | ||
| Work | 5429 | 5281 |
| Rise of the Tomb Raider | ||
| 1920x1080, VeryHigh | 118,1 | 119 |
| Tom Clancy's Rainbow Six: Siege | ||
| 1920x1080, Ultra | 115,7 | 114,9 |
| Tom clancy "s the division | ||
| 1920x1080, Max | 93 | 92,6 |
And finally, about the most important thing: performance. The test was attended by a senior representative of the line - the Core i7-7700K, which replaced the Core i7-6600K. As we already said, technically the crystals differ only in frequency: under Turbo Boost, the new product produces 300 MHz more, and in the standard it keeps the speed 200 MHz higher. Actually, this difference in frequency is where the performance gains fit. In all tasks the i7-7700K is about 5-6% faster than its predecessor. And when comparing at the same frequency, the difference fits into the measurement error.
As for the processor temperature, nothing has changed here. At the limit, the processor easily reaches 80 ° C. But our processor was scalped and even at a frequency of 4.8 GHz it did not heat up above 70 ° C.
* * *
The seventh generation Intel Core i7 can hardly be called "new". In fact, we have before us the same Skylake, but at slightly higher frequencies. Good or bad, decide for yourself, our opinion is that. If you are on a relatively recent Intel architecture (Skylake or Haswell), there is no point in upgrading to Kaby Lake. But if you are building a computer from scratch, then the seventh Core is the only correct option until AMD Ryzen comes out.
We are grateful to ASUS for the provided equipment.
| Test bench | |
| Cooling | Thermalright Macho HR-02 |
| Motherboard | ASUS ROG Maximus IX Formula |
| Memory | 2x 4 GB DDR4-2666 MHz Kingston HyperX Fury |
| Video card | NVIDIA GeForce GTX 1070 |
| Storage devices | Toshiba OCZ RD400 (512GB) |
| Power Supply | Hiper K900 |
| Additionally | Windows 10 64-bit |
| NVIDIA Drivers | 378.41 |
| Core i7 specifications | ||
| CPU | Intel Core i7-7700K | Intel Core i7-7700 |
| Architecture | Kaby lake | Kaby lake |
| Technological process | 14 nm | 14 nm |
| Socket | LGA1151 | LGA1151 |
| Number of cores / threads | 4/8 pcs. | 4/8 pcs. |
| L3 cache size | 8 MB | 8 MB |
| Nominal clock frequency | 4.2 GHz | 3.6 GHz |
| 4.5 GHz | 4.2 GHz | |
| Number of memory channels | 2 pcs. | 2 pcs. |
| Supported memory type | DDR4-2400 / DDR3L-1600 | DDR4-2400 / DDR3L-1600 |
| 16 | 16 | |
| Thermal package (TDP) | 91 watts | 65 watts |
| Price for January 2017 | 20,700 rubles ($ 345) | 18 600 rubles ($ 310) |
| Core i5 Specifications | ||||
| CPU | Core i5-7600K | Core i5-7600 | Core i5-7500 | Core i5-7400 |
| Architecture | Kaby lake | Kaby lake | Kaby lake | Kaby lake |
| Technological process | 14 nm | 14 nm | 14 nm | 14 nm |
| Socket | LGA1151 | LGA1151 | LGA1151 | LGA1151 |
| Number of cores / threads | 4/4 pcs. | 4/4 pcs. | 4/4 pcs. | 4/4 pcs. |
| L3 cache size | 6 MB | 6 MB | 6 MB | 6 MB |
| Nominal clock frequency | 3.8 GHz | 3.5 GHz | 3.4 GHz | 3.0 GHz |
| Maximum Turbo Boost Frequency | 4.2 GHz | 4.1 GHz | 3.8 GHz | 3.5 GHz |
| Number of memory channels | 2 pcs. | 2 pcs. | 2 pcs. | 2 pcs. |
| Supported memory type | DDR4-2400 / DDR3L-1600 | DDR4-2400 / DDR3L-1600 | DDR4-2400 / DDR3L-1600 | DDR4-2400 / DDR3L-1600 |
| # Of PCI Express 3.0 Lines Supported | 16 | 16 | 16 | 16 |
| Thermal package (TDP) | 91 watts | 65 watts | 65 watts | 65 watts |
| Price for January 2017 | 14,500 rubles ($ 242) | 13,200 rubles ($ 220) | 12,000 rubles ($ 200) | 11,100 rubles ($ 185) |
| Core i3 specifications | ||||
| CPU | Core i3-7350K | Core i3-7320 | Core i3-7300 | Core i3-7100 |
| Architecture | Kaby lake | Kaby lake | Kaby lake | Kaby lake |
| Technological process | 14 nm | 14 nm | 14 nm | 14 nm |
| Socket | LGA1151 | LGA1151 | LGA1151 | LGA1151 |
| Number of cores / threads | 2/4 pcs. | 2/4 pcs. | 2/4 pcs. | 2/4 pcs. |
| L3 cache size | 4 MB | 4 MB | 4 MB | 3 MB |
| Nominal clock frequency | 4.2 GHz | 4.1 GHz | 4.0 GHz | 3.9 GHz |
| Maximum Turbo Boost Frequency | - | |||
| Number of memory channels | 2 pcs. | 2 pcs. | 2 pcs. | 2 pcs. |
| Supported memory type | DDR4-2400 / DDR3L-1600 | DDR4-2400 / DDR3L-1600 | DDR4-2400 / DDR3L-1600 | DDR4-2400 / DDR3L-1600 |
| # Of PCI Express 3.0 Lines Supported | 16 | 16 | 16 | 16 |
| Thermal package (TDP) | 60 watts | 51 watts | 51 watts | 51 watts |
| Price for January 2017 | 10,500 rubles ($ 175) | 9300 rubles ($ 155) | 8700 rubles ($ 145) | 7000 rubles ($ 117) |
Previously, when choosing a processor for their computer, users mainly paid attention to the brand and the clock speed. Today the situation has changed a bit. No, even today you will need to make a choice between two manufacturers - Intel and AMD, but this will not end there. Times have changed and both companies produce a good quality product that can satisfy the needs of almost any demanding user.
However, each manufacturer's product has its own strengths and weaknesses, which are manifested in the performance of various software applications, as well as in the range of prices and performance. Plus, today a processor with a much lower clock frequency can easily bypass its faster brother, and a multi-core processor may turn out to be slower than a processor based on the old architecture, with a certain load on the system.
We will tell you how modern processors differ from each other, and the choice is yours.
Characteristics of modern processors
1. CPU clock speed
This indicator, which determines the number of clock cycles (operations) that the processor can do per second of time. Previously, this indicator was decisive when choosing a computer and a subjective assessment of processor performance.
Now, the times have come when this figure is sufficient for the vast majority of modern processors to perform standard tasks, therefore, when working with many applications, there will not be a significant increase in performance due to a higher clock frequency. Performance is now determined by other parameters.
2. Number of Cores
Most modern computer processors have two or more cores, the exception can be made only by the most budget models. Everything seems to be logical here - more cores, higher performance, but in reality it turns out that everything is not so simple. In some applications, the performance gain may indeed be due to the number of cores, but in other applications the multi-core processor may be inferior to its predecessor with fewer cores.
3 The amount of cache memory for processors
In order to increase the speed of data exchange with the computer's RAM, additional high-speed memory blocks are installed on the manufactured processors (the so-called caches of the first, second, third levels, or LI, L2, L3 cache). Again, everything seems logical - the larger the cache memory in the processor, the higher its performance.
But here again different processor models emerge, which, as a rule, differ from each other by several technical parameters at once, so it is practically impossible to reveal the direct dependence of performance on the size of the chip's cache memory.
Moreover, a lot also depends on the specifics of the software application code. Some applications, with a large cache, give a noticeable increase, while others, on the contrary, start to work worse because of the program code.
4 Core
The core is the basis of any processor, from which other characteristics are based. You can find two processors with at first glance similar technical characteristics (number of cores, clock frequency), but with different architectures and they will show completely different results in performance tests and software applications.
Traditionally, processors based on newer cores are much better at handling various programs and therefore perform better than models based on outdated technologies (even if the clock speeds are the same).
5 Technical process
This is the scale of modern technologies, which actually determine the size of the semiconductor elements that serve in the internal circuits of the processor. The smaller these elements are, the more perfect the applied technology. This does not mean at all that a modern processor, created on the basis of a modern technical process, will be faster than a representative of the old series. It's just that he can, for example, heat up less, and therefore work more efficiently.
6 Front Side Bus (FSB)
The system bus frequency is the rate at which the processor core communicates with the RAM, discrete graphics card, and peripheral controllers on the computer's motherboard. Everything is simple here. The higher the bandwidth, the correspondingly higher the performance of the computer (all other things being equal technical characteristics of the computers in question).
Deciphering the names of Intel processors
Learning to navigate the huge range of different names of Intel processors is quite simple. First you need to figure out the positioning of the processors themselves:
Core i7- currently the top line of the company
Core i5- are distinguished by high performance
Core i3- low price, high / medium performance
All Core i series processors are built on the Sandy Bridge core and belong to the second generation of Intel Core processors. Most models start with 2, while more recent modifications based on the latest Ivy Bridge core are marked with 3.
Now it is very easy to determine what generation this or that processor, and on the basis of which core it was created. For example, the Core i5-3450 belongs to the third generation based on the Ivy Bridge core, and the Core i5-2310 is, respectively, the second generation based on the Sandy Bridge core.
When you know the type of processor core, you can already roughly judge not only its capabilities, but also about the potential heat release during loading. Representatives of the third generation heat up much less than their predecessors thanks to a more modern technical process.
In addition to numbers, suffixes are sometimes used in processor names:
TO- for processors with an unlocked multiplier (this allows experienced computer users to overclock the processor themselves)
S- for products with increased energy efficiency, T - for the most economical processors.
Intel Core 2 Quad
The line of popular four-core processors based on the outdated Yorkfield core (45 nm process technology), due to the attractive low price and fairly high performance, the line of these processors is relevant today.
Intel Pentium and Celeron
When labeling budget Pentium and Celeron processors, they use the designations G860, G620 and some others. The higher the number after the letter, the correspondingly the processor is more productive. If the marking numbers differ insignificantly, then, most likely, we are talking about various modifications of chips in the same production line, usually they are small and consists of only a few hundred megahertz of the core clock frequency. Sometimes the size of the cache memory, and even the number of cores, differ, and this already has a much stronger effect on the differences in power and performance. Therefore, it will be better if you do not rely on the marking of chips, but check all the technical specifications on the official website of the seller or manufacturer, because it will take little time, but it will help save nerves and money.
An illustrative example is that the Celeron G440 and Celeron G530 processors, differing in price by only 200 rubles, actually have a different number of cores (Celeron G440 - one, Celeron G530 - two), different clock frequency of the core (the G530 has 800 MHz more ), the G530 also has twice the cache. However, the heat dissipation of the latter processor is almost twice as much, although both processors are based on the same Sandy Bridge core.
Intel processor technologies
Intel processors are considered the most productive today thanks to the Core i7 Extreme Edition family. Depending on the model, they can have up to 6 cores simultaneously, clock speeds up to 3300 MHz and up to 15 MB L3 cache. The most popular cores in the desktop processor segment are based on Intel - Ivy Bridge and Sandy Bridge.
Just like the competitor, Intel processors use proprietary technologies of their own design to improve system efficiency.
1. Hyper Threading- Due to this technology, each physical core of the processor is capable of processing two threads of calculations simultaneously, it turns out that the number of logical cores actually doubles.
2. Turbo Boost- Allows the user to automatically overclock the processor, while not exceeding the maximum allowable operating temperature of the cores.
3. Intel QuickPath Interconnect (QPI)- The QPI ring bus connects all components of the processor, due to this, all possible delays in the exchange of information are minimized.
4. Visualization Technology- Hardware support for virtualization solutions.
5. Intel Execute Disable Bit- In practice, it provides hardware protection against possible virus attacks, which are based on buffer overflow technology.
6. Intel SpeedStep-A tool that allows you to change the voltage level and frequency depending on the load on the processor.
Deciphering the names of AMD processors
AMD FX
The top line of multi-core computer processors with a specially removed multiplier limit (for the sake of self-overclocking) to ensure high performance when working with demanding applications. Based on the first digit of the name, we can say how many cores are installed in the processor: FX-4100 has four cores, FX-6100 has six cores, and FX-8150 has eight cores. In the line of these processors, there are also several modifications with slightly different clock speeds (the FX-8150 processor has it 500 MHz higher than that of the FX-8120 processor). AMD A
A line with a graphics core built into the processor. The number designation in the name indicates that it belongs to a specific performance class: AC - performance sufficient for the vast majority of standard daily tasks, A6 - performance sufficient for creating HD video conferencing, A8 - performance sufficient for confident viewing of Blu-ray movies with the effect of 3D or launching modern 3D games in multi-display mode (with the ability to simultaneously connect four monitors).
AMD Phenom II and Athlon II
The earliest processors from the AMD Phenom II line were officially released back in 2010, but due to their low price and rather high performance, they still enjoy some popularity today.
The number of cores in the processor is indicated by the number in the name immediately after the X. For example, the marking of the AMD Phenom II X4 Deneb processor tells us that it belongs to the Phenom II processor family, has four cores and is based on the Deneb core. Completely similar marking rules can be seen in the Athlon series.
AMD Sempron
Under this name, the manufacturer produces budget processors designed for desktop office computers.
AMD processor technologies
The top-of-the-line AMD FX processors, based on the new Zambezi core, can offer the discerning user eight cores, 8MB L3 cache and a processor clock speed of up to 4200MHz.
Most modern processors made by AMD support the following technologies by default:
1. AMD Turbo CORE- This technology is designed to automatically adjust the performance of all processor cores through controlled overclocking (Intel has a similar technology called TurboBoost).
2.AVX (Advanced Vector Extensions), ХОР and FMA4- A tool that has an extended set of commands specifically designed for working with floating point numbers. Definitely a toolkit.
3. AES (Advanced Encryption Standard)- In software applications using data encryption, improves performance.
4. AMD Visualization (AMD-V)- This virtualization technology helps to share the resources of one computer among several virtual machines.
5. AMD PowcrNow!- Power management technology. They help the user achieve performance gains by dynamically activating and deactivating a portion of the processor.
6. NX Bit- Unique anti-virus technology that helps prevent infection of a personal computer with certain types of malware.
Comparison of processor performance
Looking through price lists with prices and characteristics of modern processors, you can come to real confusion. Surprisingly, a processor with more cores on board and with a higher clock speed can cost less than copies with fewer cores and lower clock speeds. The thing is that the real performance of the processor depends not only on the main characteristics, but also on the efficiency of the core itself, the support of modern technologies and, of course, on the capabilities of the platform itself for which the processor was created (you can recall the logic of the motherboard, the capabilities of the video system, about bus bandwidth and much more).
That is why, one cannot judge the performance of a processor based only on the characteristics written on paper, you need to have data on the results of independent performance tests (preferably with those applications with which you plan to constantly work). Depending on the type of workload created, similar processors can produce completely different results when working with the same programs. How can an untrained person figure out which type of processor is right for him? Let's try to figure it out by conducting a comparative test of processors with the same retail price in various software applications.
1. Working with office software. When using familiar office applications and browsers, the performance gain can be achieved due to the higher clock speed of the processor. A large amount of cache memory or a large number of cores will not give the expected performance gain for this type of application. For example, the AMD Sempron 145 processor, which is cheaper compared to the Intel Celeron G440, based on the 45nm Sargas core, shows better performance in benchmarks with office applications, while the Intel product is based on the more modern 32nm Sandy Bridge core. The clock speed is the key to success when working with office applications.
2. Computer games. Modern 3D games with settings set to maximum are some of the most demanding for computer components. Processors show performance gains in modern computer games as the number of cores grows and the amount of cache memory increases (of course, if at the same time, the RAM and video system meet all modern requirements)... Take the AMD FX-8150 processor with 8 cores and 8 megabytes of L3 cache. When tested, it gives a better result in computer games than the almost identical Phenom II X6 Black Thuban 1100T with 6 cores, but with 6 megabytes of L3 cache. As noted above, when testing office programs, the performance picture is exactly the opposite.
If you start testing the performance in modern games of two similarly priced processors brands FX-8150 and Core i5-2550K, it turns out that the latter shows better results, despite the fact that it has fewer cores, and it has a lower clock speed and even volume it has less cache memory. Most likely, here, in terms of efficiency, the more successful architecture of the kernel itself played the main role.
3. Raster graphics. Popular graphics applications such as Adobe Photoshop, ACDSee and Image-Magick were originally created by developers with excellent multi-threaded optimization, which means that with constant work with these programs, additional kernels will not be superfluous. There are also a large number of software packages that do not use multicore at all (Painishop or GIMP). It turns out that it is impossible to say unequivocally which technical parameter of modern processors more than others affects the increase in the speed of raster editors.... Different programs working with raster graphics are demanding on a variety of parameters, such as clock frequency, number of cores (especially related to the real performance of one core), and even the amount of cache memory. Nevertheless, the inexpensive Core 13-2100 in tests shows much better performance in this kind of applications than, for example, the same FX-6100, and this even despite the fact that the basic characteristics of Intel are slightly inferior.
4. Vector graphics. Nowadays, processors show themselves in a very strange way when working with such popular software packages as CorelDraw and Illustrator. The total number of processor cores has practically no effect on application performance, which indicates that this type of software does not have multi-threaded optimization. In theory, there will even be a lot of dual-core processor for normal work with vector editors, since here the clock frequency comes to the fore.
An example is AMD Ab-3650, which with four cores, but with a low clock speed cannot compete in vector editors with the budget dual-core Pentium G860, which has a slightly higher clock speed (while the cost of processors is practically the same).
5. Audio encoding. When working with audio data, you can observe completely opposite results. When encoding audio files, performance improves as the number of processor cores increases and as the clock frequency increases. In general, even 512 megabytes of cache memory is quite enough to perform operations of this kind, since this type of memory is practically not used when processing streaming data. A good example is the eight-core FX-8150 processor, which, when converting audio files into different formats, shows the result much better than the more expensive quad-core Core 15-2500K, thanks to the larger number of cores.
6. Video encoding. The kernel architecture plays a big role in software packages such as Premier, Expression Encoder or Vegas Pro. Here, the emphasis is on fast ALU / FPU - these are the hardware computational units of the kernel responsible for logical and arithmetic operations in data processing. Kernels with different architectures (even if these are different lines of the same manufacturer), depending on the type of load, provide a different level of performance
Intel's Sandy Bridge Core i3-2120 processor, with a lower clock speed, less cache and fewer cores, outperforms the AMD FX-4100 processor based on the Zambezi core, which costs almost the same money. This unusual result can be explained by differences in kernel architecture and better optimization for specific software applications.
7. Archiving. If you are often involved in archiving and unpacking large files at your computer in programs such as WinRAR or 7-Zip, then pay attention to the amount of cache memory of your processor. In such cases, the cache memory has a direct proportion: the larger it is, the more computer performance when working with archivers... The indicator is the AMD FX-6100 processor with 8 MB of level 3 cache installed on board. It manages the archiving task much faster than comparable processors Core i3-2120 with 3 MB of L3 cache and Core 2 Quad Q8400 with 4 megabytes of L2 cache.
8. Extreme multitasking mode. Some users work with several resource-intensive software applications at once with background operations activated in parallel. Just think, you are unpacking a huge RAR archive on your computer, simultaneously listening to music, editing several documents and spreadsheets, while you are running Skype and an Internet browser with several open tabs. With such active use of the computer, the ability of the processor to perform several threads of operations in parallel plays a very important role. It turns out that the number of cores of the processor is of paramount importance for this use.
Multitasking is handled by the AMD Phenom II Xb and FX-8xxx multi-core processors. It is worth noting here that the AMD FX-8150 with eight cores on board, while running several applications at the same time, has a slightly larger performance margin than, for example, the more expensive Core i5-2500K processor with only four cores. Of course, if maximum speed is required, then it is better to look towards the Core i7 processors, which can easily overtake the FX-8150.
Output
In conclusion, we can say that a huge number of different factors affect the overall performance of the system. Of course, it is good to have a processor with a high clock speed, a large number of cores and a large cache memory, plus it would be nice to have the most modern architecture, but all these parameters have different meanings for different types of tasks.
The conclusion suggests itself: if you want to invest money sensibly in upgrading your computer, then identify the highest priority tasks and imagine scenarios for everyday use. Knowing the specific goals and objectives, you can easily choose the optimal model that best suits your needs, work and, most importantly, budget.
