Intel divides its microprocessors into two main groups. On the one hand, its Celeron and Pentium family for users who do not require high performance, and on the other hand, i3, i5 and i7, for advanced users.
I5 is a processor that can be called an SUV. If , is sufficient for 80% of users, the i5 processor is suitable for almost anyone.
The differences between an i5 processor and an i7 processor are small and not worth the extra expense in most cases. It might make more sense, depending on how you intend to use your computer, to invest in SSDs, RAM, or a good graphics card.
Of course, the i7 processor is no worse than the i5, just the applications for which it is needed are rather narrowly specific.
Nuclei . In desktop PCs - 4 cores, except for i5-6xx models and 2 cores in laptops. All 2 core i5 processors support HyperThread technology.
turbo boost . The fundamental difference from i3. Turbo boost, if needed, allows the processor to run at higher speeds. The benefits of this additional technology are especially noticeable in applications that use a single thread. And such applications, by the way, the majority.
Integrated graphics . Some of the i5 processor models have an integrated graphics card. A computer with such a processor is of course cheaper, but then you should keep in mind that the processor is discrete, that is, less powerful, and will be used to start the computer.
Memory controller . As with the graphics card, the memory controller is integrated into the processor. Such a processor determines the type of RAM that can be installed. That is, with an i5 processor, only DDR3 can be used.
PCI Express . The PCI Express controller is also integrated into the i5 processor. Thus, if you are using a discrete graphics card, the connection to the processor will be direct.
Versions of the i5 processor.
The first generation of i5 processors. It has several types of processors. I5-7xx, 7xxS - on the Lynnfield core. i5-6xx - on the Clarkdale core. i5-5xxM, 4xxM, 5xxUM, 4xxUM - on the Arrandale core for portable devices. The first models of processors have 4 cores, the other 2 cores with Hyperthread technology.
The production technology allows the creation of transistors 45 nanometers in Lynnfield, against 32 nanometers in Arrandale and Clarkdale.
As a set of instructions, they support SSE 4.1/4.2 and MMX. The i5 6xx series processor and Arrandale already have a graphics card integrated.
Second generation i5 processors. Also known by his own name Sandy Bridge. Support for AVX instructions has been added to the processor, which allows you to speed up scientific, financial calculations, signal processing, etc.
In desktop versions of the computer, all i5 processors have 4 cores, except for the 2390T, which has 2 cores and Hyperthread technology. The laptop has everything as in the last version.
Another distinguishing feature of these i5 processors is the inclusion of Quicksync, which increases the speed of video processing and encoding.
Third generation i5 processors. Also known as Ivy Bridge. In these processors, Intel has improved the manufacturing technology itself. The corporation managed to create transistors of 22 nanometers. Thus, in the same area, they managed to place twice as many. What added energy efficiency and increased the speed of data processing.
Like Sandy Bridge, desktop PCs have i5 processors with four cores. Except i5 3470T series processor which has 2 cores and Hyperthread technology. Everything in the laptop is like in the i5 processor of the 3470T series.
For whom i5 processor.
As mentioned above, the i5 processor will suit almost any user. If your budget is still limited, this processor is the best choice for you. Add to this that the actual applications for which it is better to take advantage of the i7 processor are quite specific and you have an almost perfect processor.
Part two: "The most important features of each Intel Core i3/i5/i7 processor family. Which of these chips are of particular interest"
Introduction
First, we will highlight the most important features of each Intel Core i3/i5/i7 processor family, and then we will talk about which of these chips are of particular interest. For the convenience of readers, we considered it appropriate to present the information in the form of a kind of reference book, and to bring all the data on the current models of the model into small tables. The prices given by us are Russian retail prices, fixed at the time of publication of this material, for processors in a "boxed" configuration (that is, with a branded cooler).
Core i3
Core i3 (Clarkdale) is the latest generation dual-core processor designed for entry-level desktops. First introduced January 7, 2010. It is installed in the LGA1156 connector. Produced using 32nm technology.
Equipped with an integrated PCI Express 2.0 x16 controller, thanks to which the graphics accelerator can be connected directly to the processor. To connect to the system logic set, a DMI (Digital Media Interface) bus with a bandwidth of 2 GB/s is used.
The Core i3 processors have a 12-pipeline GMA HD graphics core clocked at 733MHz.
The base clock frequency for all Core i3 models is 133 MHz, the nominal frequencies are achieved using multipliers.
Compatible chipsets: Intel H55 Express, H57 Express, P55 Express, Q57 Express
Main technical parameters of Core i3
- Nehalem microarchitecture
- Two cores
- L3 cache - 4 MB, shared by all cores
- Integrated PCI Express 2.0 x16 controller
- Integrated 733 MHz graphics adapter
- SSE 4.2 instruction set
- AES-NIS instruction set
Core i5
Core i5 (Clarkdale or Lynnfield) is the latest generation dual or quad-core processor designed for mid-range desktops. First introduced September 8, 2009. It is installed in the LGA1156 connector. The dual-core Clarkdale is manufactured using 32nm technology, the quad-core Lynnfield is manufactured using 45nm technology.
Equipped with a built-in dual-channel DDR3-1066 / 1333 RAM controller with a voltage of up to 1.6 V. Modules designed for a higher voltage will not work with this chip and may even damage it.
Equipped with an integrated PCI Express 2.0 x16 controller, thanks to which the graphics accelerator can be connected directly to the processor. In models with an integrated GMA HD graphics core, one video card in x16 mode can be connected to the chip, in models without integrated graphics, two video cards in x8 mode each.
To connect to the system logic set, a DMI (Digital Media Interface) bus with a bandwidth of 2 GB/s is used.
The dual-core models (6xx series) have a built-in GMA HD graphics adapter and Hyper-Threading technology, the quad-core models (7xx series) do not have graphics and Hyper-Threading. Models ending in 1 have a graphics clock speed of 900 MHz, models ending in 0 have a graphics core clocked at 733 MHz.
All Core i5s feature Turbo Boost technology for automatic overclocking in resource-intensive tasks.
The base clock frequency for all Core i5 models is 133 MHz, the nominal frequencies are achieved using multipliers.
Compatible chipsets: Intel H55 Express, H57 Express, P55 Express, Q57 Express.
Main technical parameters of Core i5
- Nehalem microarchitecture
- two or four cores
- L1 cache - 64 KB (32 KB data and 32 KB instructions) for each core
- L2 cache - 256 KB per core
- L3 cache - 4 or 8 MB, shared by all cores
- Built-in dual-channel DDR3-1066/1333 MHz RAM controller
- Integrated PCI Express 2.0 controller (one x16 lane or two x8 lanes on non-integrated graphics models)
- Integrated 733 or 900 MHz graphics adapter
- Support for VT virtualization technology
- Support for 64-bit Intel EM64T instructions
- Support for Hyper-Threading technology in dual-core models
- SSE 4.2 instruction set
- AES-NIS instruction set
- Antivirus Technology Execute Disable Bit
- Enhanced SpeedStep Dynamic Frequency Technology
Core i7
Core i7 (Bloomfield, Lynnfield, or Gulftown) is the latest generation of four or six-core processors designed for high-end desktops. First introduced in November 2008. Quad-core Bloomfield and Lynnfield are manufactured using 45nm technology, six-core Lynnfield - using 32nm technology.
Available in two versions: 9xx series (for LGA1366 connector) with built-in three-channel memory controller and QPI bus and 8xx series (for LGA1156 connector) with dual-channel memory controller, built-in PCI Express 2.0 controller and DMI bus) DDR3-1066/1333 RAM is supported with voltages up to 1.6 V. Modules rated for higher voltages will not work with this chip and may even damage it.
Processors for the LGA1366 socket are equipped with a high-speed QPI bus operating at a frequency of 2.4 GHz (up to 4.8 GB / s) in regular i7 and at a frequency of 3.2 GHz (6.4 GB / s) in Extreme modifications (these include i7-965, i7-975 and i7-980X.
Chips for the LGA1156 connector are equipped with an integrated PCI Express 2.0 x16 controller, thanks to which the graphics accelerator can be connected directly to the processor. To connect with a set of system logic, a DMI (Digital Media Interface) bus with a bandwidth of 2 GB / s is used here.
All Core i7s feature Turbo Boost automatic overclocking technology for resource-intensive tasks, as well as Hyper-Threading technology.
The base clock frequency for all Core i7 models is 133 MHz, the nominal frequencies are achieved using multipliers. In modifications of Core i7 Extreme, the multiplier is unlocked, which allows you to freely increase the clock speed of the processor.
Compatible chipsets: 8xx series - Intel H55 Express, H57 Express, P55 Express, Q57 Express, 9xx series - Intel X58 Express.
Main technical parameters of Core i7
- Nehalem microarchitecture
- Four or six cores
- L1 cache - 64 KB (32 KB data and 32 KB instructions) for each core
- L2 cache - 256 KB per core
- L3 Cache - 8 or 12 MB shared between all cores
- Built-in dual-channel (LGA1156) or triple-channel (LGA1366) DDR3-1066/1333 MHz RAM controller
- QPI bus running at 2.4 GHz (4.8 GB/s) or 3.2 GHz (6.4 GB/s) on LGA1366 models
- DMI bus (2 GB/s) on models for LGA1156
- Integrated PCI Express 2.0 controller (one x16 lane or two x8 lanes on non-integrated graphics models) on LGA1156 models
- Support for VT virtualization technology
- Support for 64-bit Intel EM64T instructions
- Support for Hyper-Threading Technology
- Support for Turbo Boost Technology
- SSE 4.2 instruction set
- AES-NIS instruction set for the i7-980X model
- Antivirus Technology Execute Disable Bit
- Enhanced SpeedStep Dynamic Frequency Technology
What to choose?
The Core i3-530 and 540 processors are quite powerful and inexpensive chips, and the price difference between them is negligible, so there is no point in getting a 530 unless you are on a tight budget.
The Core i3 series chips are direct competitors to the previous generation Core 2 Duo Exxx processors: they cost about the same and provide a comparable level of performance, although somewhat faster. However, while LGA1156 motherboards are more expensive than LGA775 motherboards, buying an i3 chip is a smarter long-term investment than a Core 2 Duo because these processors are not only fast enough today, but can be swapped out for any LGA1156 chip in the future. - even on the super powerful Core i7. If the i3-530 is too expensive for you, you can pay attention to the Pentium G6950 ("boxed" version complete with a standard cooler will cost about 3200 rubles), which is slower than both "three" cards, but practically does not concede to most Core 2 Duo.
As for the quad-core Core 2 Quad, which are slightly more expensive than the dual-core Core i3 (for example, the "boxed" Core 2 Quad Q8300 costs about 5000 rubles), then buying them today only makes sense to upgrade an existing system to an LGA775 socket - in this case it is very reasonable choice.
All 600-series Core i5 processors are high performance, but if you do not need a chip with integrated graphics, there is not much point in buying a model in this family. These models are rather focused on the corporate market - an office computer does not need powerful graphics, and the simpler it is, the more convenient it is to maintain.
For the same money that they ask for chips of the 600th family, it is better to purchase a quad-core i5-750 - this is an ideal choice for building a powerful home PC for a reasonable price. If you make a choice within the 600 series, you should know that the 661 differs from the 660 only in slightly faster integrated graphics, but at the same time with increased power consumption and the lack of hardware support for VT-d I / O virtualization, which is relevant only for corporate users. In other words, if you are buying a CPU for a home computer, it makes sense to prefer the Core i5-661.
For building a powerful gaming PC, the best choice in terms of price / performance is the Core i7-860, all other options will cost significantly more, since you will need a more expensive motherboard based on the X58 Express chipset for the LGA1366 socket.
The six-core "extreme" Core i7-980X is an unsurpassed leader in performance not only for the entire modern line of Intel desktop processors, but also for competing AMD models. Therefore, do not be surprised that a system based on it will cost a rather impressive amount. Fans of the very best can prepare their wallets - this chip is about to appear on the shelves of Russian stores, replacing the previous flagship Core i7-975
Posted on October 30, 2017
We chose the Core i7 and Core i5 HQ and U series processors. These four models are used in most notebooks on the market. As you can see above, the two U-series processors are clocked faster than the Core i5-7300HQ and are generally offered at a lower price.Is this enough to win?
The short answer is NO. Full-fledged HQ series processors are still cooler.
Cinebench R15
Let's start with one of the iconic Cinebench processor benchmarks. We chose the multi-core scenario not only because most applications (including games) use multiple cores at once, but also to see how the result will be affected by the presence of additional processor cores (or the ability to execute more instruction streams).We see the same picture: HQ-series processors are tearing their U-series rivals to shreds. Moreover, the Core i5-7300HQ model not only outperforms the i5-7200U by as much as 40%, but also leaves the Core i7-7500U behind – by 22%!
X264 Benchmark
If the term "computing performance" sounds too vague for you, the X264 benchmark, which simulates video transcoding by the CPU, will help clarify the picture. The higher the result, the faster the processor can convert videos from one format to another.
 |
conclusions
If you expect decent performance from your computer, go for the HQ series processor.Don't let the "i7" name fool you. Even the i5-HQ processor will be faster than the i7-U! In addition to the number of cores and execution threads, HQ processors have other advantages, such as larger cache sizes, and therefore are better suited for high-performance laptops, including gaming models.
This does not mean that U-series processors are worse. They are just meant for other purposes. Their destiny is ultrabooks, for which mobility and low power consumption are priorities. When speed is most important, you should always choose HQ series processors.
It's been almost a month since Intel introduced the Coffee Lake family of processors, and the past weeks have clearly shown that they were released somewhat hastily. There are a lot of indicators of poor preparation of the announcement. The availability of new products in retail is extremely limited, and prices are noticeably inflated by sellers due to shortages. The situation with motherboards is not ideal either: there is a fairly wide selection of LGA1151 motherboards on the shelves based on the Z370 chipset compatible with Coffee Lake, but many of them cause serious complaints from users due to constantly revealed flaws in the firmware.
Nevertheless, despite all the problems, Coffee Lake-based platforms are rated very positively by the community. By adding additional processing cores to the new processors, Intel has done exactly what users have long wanted from it. The performance of mainstream Intel processors has made a noticeable leap, and as a result, the representatives of the new family have become very good candidates for getting into modern desktops, despite all the “childhood illnesses” and the existence of competing AMD Ryzen processors.
We have already expressed our own opinion about Coffee Lake in the review: testing then showed that Intel was able to quickly catch up with the lagging behind the competitor in certain aspects. Nevertheless, for all its merits, the Core i7-8700K is not very suitable for the mass user. Not only that, with the transition to the Coffee Lake design, Intel has increased its appetite and priced its new flagship mainstream processor more than before, raising the recommended price of the Core i7-8700K from the usual $339 to $359. In addition, real retail prices go far beyond this line. For example, in the largest North American online stores, they will ask for at least $410 for this chip (subject to availability in stock), and such limits do not constrain domestic retail.
Of course, not everyone is ready to buy a mass processor for an amount exceeding $400. Therefore, we decided to pay attention to new items in the lower class, which belong to the Core i5 family, and not Core i7. As before, such CPUs differ from their older counterparts by the lack of support for Hyper-Threading technology, that is, they retain the six-core structure. And this means that in terms of price and performance, Coffee Lake in the guise of Core i5 can be even more attractive than Core i7. They are also able to offer an increased number of computing cores compared to their predecessors, but even according to the official price list, their cost is at least $100 lower than that of the Core i7.
In the past, we often recommended unlocked Core i5 series processors for mid-range desktops, primarily for gaming. Now, it seems, having acquired a couple of additional cores, this series offers an even better combination of consumer characteristics. That is why we decided to conduct a detailed test of the older Coffee Lake Core i5 series and try to evaluate whether this option is much worse compared to the Hyper-Threaded Core i7 processor and how it stands up to the competing offerings of the Ryzen 7 and Ryzen 5 series, which, despite lineup upgrades carried out by Intel continue to dominate the number of threads and sometimes even cores.
Core i5-8600K in detail
The Core i5-8600K processor, like the Core i7-8700K, can be described as a typical representative of the Coffee Lake family - it has six processing cores at its disposal. The main difference from the older brother is the disabled Hyper-Threading technology: this is exactly what the desktop Core i5 has always been different from the Core i7 from the very moment these trademarks appeared in 2011. Intel's commitment to this principle makes today's Core i5-8600K especially attractive - compared to the predecessor of the Kaby Lake generation, the computing power of the new product has grown significantly: it has not only one and a half times more cores, but also increased operating frequencies. All this is perfectly visible when comparing the specifications.
| Core i5-8600K | core i5 -7 6 00K | |
| codename | coffee lake | Kaby Lake |
Production technology, nm |
14++ | 14+ |
| Cores/Threads | 6/6 | 4/4 |
| Base frequency, GHz | 3,6 | 3,8 |
| Frequency Turbo Boost 2.0, GHz | 4,3 | 4,2 |
| L3 cache, MB | 9 |
6 |
| Memory support | DDR4-2666 | DDR4-2400 |
| Integrated graphics | GT2: 24 EU | GT2: 24 EU |
| Max. graphics core frequency, GHz | 1,15 | 1,15 |
| PCI Express lanes | 16 | 16 |
| TDP, W | 95 | 91 |
| socket | LGA1151v2 | LGA1151v1 |
| Official price | $257 | $242 |
There are no improvements at the microarchitectural level in Coffee Lake, that is, with a single-threaded load and at the same clock frequency, the new processors are identical in performance to Kaby Lake. However, for the production of new products, an improved 14 ++ nm manufacturing process is used. While Intel is still unable to start producing large processor chips using a more advanced 10nm technology, the start of which was pushed back to at least the second half of 2018 for the manufacture of desktop processors, engineers are optimizing the old 14nm process technology. And by no means without success. Today's 14++nm technology compared to the original process technology was able to provide a solid reduction in leakage currents, resulting in a 52% reduction in heat generation at the same level of performance. It is thanks to this achievement that the Core i5-8600K has one and a half times more cores, and the maximum frequency in turbo mode has increased from 4.2 GHz to 4.3 GHz.
True, some concerns are caused by a decrease in the characteristics of the base frequency: for the Core i5-8600K, it is set to 3.6 GHz, which is 200 MHz less than that of the corresponding Kaby Lake. However, this lag should be compensated by the aggressive Turbo Boost 2.0 technology, which in Coffee Lake can increase the processor frequency much more than before. Even with a load on all six cores, if the power consumption and heat dissipation of the Core i5-8600K remains within the established limits, the operating frequency of the processor can increase to 4.1 GHz. As a result, given the active turbo mode, the Core i5-8600K should always be ahead of its quad-core predecessor.
| Rated frequency | Max Frequency Turbo Boost 2.0 | ||||||
| 1 core | 2 cores | 3 cores | 4 cores | 5 cores | 6 cores | ||
| Core i5-8600K | 3.6 GHz | 4.3 GHz | 4.2 GHz | 4.2 GHz | 4.2 GHz | 4.1 GHz | 4.1 GHz |
| Core i5-7600K | 3.8GHz | 4.2 GHz | 4.1 GHz | 4.1 GHz | 4.0 GHz | - | - |
In addition to increased frequencies and additional cores, the Core i5-8600K can offer a 3 MB increase in the third level cache, as well as official support for dual-channel DDR4-2666 with a bandwidth of up to 42.7 GB / s against DDR4-2400 with a bandwidth of 38.4 GB / s. With.
True, to get all the benefits provided by the new product, you will need a new motherboard based on the Intel Z370 logic set. The new version of LGA1151, which is used by Coffee Lake processors, has added additional power lines, and old LGA1151 boards based on the Z270 or Z170 (and other legacy chipsets) do not work with 8000-series processors. But without exception, all new boards compatible with the Core i5-8600K can provide its overclocking. It, like the Core i7-8700K, has an unlocked multiplier, so with a couple of manipulations in the BIOS of the motherboard, its operating frequency can be easily increased, as well as the frequency at which the L3 cache and system memory operate. At the same time, overclocker LGA1151 processors of the Coffee Lake family are declared to comply with a 95-watt thermal package, which means that theoretically their moderate overclocking is quite possible without the use of bulky air or liquid cooling systems.
There is no doubt that the Core i5-8600K is better than its Kaby Lake predecessor, the Core i5-7600K, in every way. However, now you need to compare this processor not only with internal competitors, but also with those processors that AMD offers in the same price segment. The real retail price of the Core i5-8600K today is about $300, and for this amount you can buy an eight-core Ryzen 7 1700. If you focus on official prices, then the direct competitor for the older Core i5 is the six-core Ryzen 5 1600X. Let's compare the Core i5-8600K specs with both AMD alternatives.
| Intel | AMD | ||
| Core i5-8600K | Ryzen 7 1700 | Ryzen 5 1600X | |
| socket | LGA1151v2 | SocketAM4 | SocketAM4 |
| Cores/Threads | 6/6 | 8/16 | 6/12 |
| base frequency | 3.6 GHz | 3.0 GHz | 3.6 GHz |
| Turbo/XFR | 4.3 GHz | 3.7/3.75 GHz | 4.0/4.1 GHz |
| Overclocking | There is | There is | There is |
| L2-cache | 256 KB per core | 512 KB per core | 512 KB per core |
| L3-cache | 9 MB | 2 × 8 MB | 2 × 8 MB |
| Memory | DDR4-2666 | DDR4-2666 | DDR4-2666 |
| PCI lanes | 16 | 16 | 16 |
| Graphics core | There is | Not | Not |
| TDP | 95 W | 65 W | 95 W |
| Official price | $257 | $329 | $249 |
From the point of view of formal characteristics, AMD's proposals continue to look attractive, even though Intel has significantly increased the number of cores in Coffee Lake processors. Ryzen 5 and Ryzen 7 continue to outperform their rivals, at least in terms of the number of executable threads and cache sizes. However, on the side of Coffee Lake, the leadership in terms of clock frequencies, plus we should not forget that modern Intel processor cores have a clear advantage in terms of IPC - the number of instructions executed per clock.
As our previous tests have shown, in resource-intensive applications, the six-core Core i7-8700K performs at least as well as the eight-core Ryzen 7 1700X. But the performance gap between the Core i5-8600K and the Ryzen 7 1700 is more significant: while Intel blocks Hyper-Threading in the new mid-range processors, Ryzen's SMT technology is present not only in the eight-core Ryzen 7, but also in the six-core Ryzen 5. And this means that the situation in the middle price segment may remain ambiguous even after updating the lineup of Intel processors.
Naturally, detailed tests will dot all the “yo”, but it’s too early to move on to them.
We were deceived: features of the turbo mode in Coffee Lake
When we first got acquainted with the processors of the Coffee Lake generation and tested, we noted that its real frequency always corresponds to the maximum allowed turbo frequency for the corresponding load. This had a positive effect on performance: still, the Core i7-8700K with a nominal frequency of 3.7 GHz, even with the maximum AVX load on all six cores, “spun” at 4.3 GHz, leaving no doubt about the superiority of the new processor technology design and 14++ nm. True, some bewilderment was caused by thermal and electrical indicators. The fact is that while the thermal package of the Core i7-8700K is set to 95 W, and the maximum allowable temperature is 100 degrees, its real consumption under maximum load reached 140-145 W, and the temperature with the highly efficient Noctua NH-U14S cooler - up to 88 degrees. It is very doubtful that this mode of operation of the CPU can be considered normal.
Even bigger questions regarding the correct operation of Coffee Lake processors in turbo mode began to arise when we began to get acquainted with the Core i5-8600K sample. This time, we had a serial copy of the CPU in our hands, and it was no longer possible to attribute the strangeness observed with consumption and temperatures to the features of the engineering sample. And the reasons for surprise only increased. The thing is that in the nominal mode at full AVX load, which we traditionally created with the LinX 0.8.0 utility, the temperature went beyond all reasonable limits.
As you can see from the above screenshot, the processor frequency under full load in LinX 0.8.0 is 4.1 GHz - this is the maximum possible frequency of the Core i5-8600K with all six cores involved. At the same time, CPU consumption reaches the already familiar 145 W, and the temperature reaches the maximum allowed by the specification - 99 degrees. And this is with the Noctua NH-U14S cooler, which is not the slightest reason to blame for the inability to withstand the high thermal power of the chip! It is clear that such a high temperature is largely due to the low efficiency of the internal thermal interface used in Intel processors, but at the same time it is quite obvious that there should not be any critical heating of the Core i5-8600K in the nominal mode anyway.
Therefore, we turned to Intel engineers for clarification, who gave a very discouraging comment: on many LGA1151 motherboards based on the Z370 chipset, Turbo Boost 2.0 technology is implemented incorrectly. In an attempt to squeeze the maximum efficiency out of new processors, board manufacturers deliberately ignore the established limits on processor power consumption, and this can indeed lead to overheating. Unfortunately, the ASUS Strix Z370-F Gaming motherboard we used turned out to be a prime example of a board with incorrectly configured turbo mode. Therefore, it is not surprising that when tested on this platform, the Core i7-8700K and Core i5-8600K showed off-scale temperatures and power consumption.
In fact, the processors of the Coffee Lake family, when the turbo mode is activated, should not at all work at the maximum frequencies determined for the load on a particular number of cores. This is just an upper bound, and some more conditions are attached to it. The main one: CPU consumption over long periods of time should not go beyond the established TDP limits (that is, beyond 95 W for the Core i7-8700K and Core i5-8600K) and can only briefly reach 120 W. However, these additional conditions have been blocked by many board manufacturers at the BIOS level, and Intel is now working with partners to ensure that Turbo Boost 2.0 technology can be restored correctly.
It is clear that this will entail some decrease in the performance of new processors with a high computational load, but the temperature regime of Coffee Lake will finally be able to not inspire any concerns. And representatives of Intel have already been able to achieve some success in advising board manufacturers. For example, in the latest BIOS versions for our ASUS Strix Z370-F Gaming board (0419 and 0420), the implementation of turbo mode is already quite up to par. After updating the firmware, the frequency of the Core i5-8600K, when tested in LinX 0.8.0, no longer holds at 4.1 GHz and drops to 3.5 GHz, so that the temperature and consumption remain within acceptable limits: 95 W and 72 degrees respectively.
In terms of performance, switching the motherboard to the correct multiplier operation resulted in a 10% drop in the Linpack performance score (from 330 to 300 Gflops), as expected. However, in this case, the maximum frequency underestimation takes place, since Linpack uses extremely power-hungry AVX2 instructions. For example, when testing in Prime95 with AVX instructions disabled, the operating frequency of the Core i5-8600K is already 3.9 GHz, which is noticeably closer to the maximum set for full load, but still not up to it.
Nevertheless, one cannot but pay attention to the fact that due to incorrect support for turbo mode by motherboards, the results of Coffee Lake performance measurements made at the time or before the announcement of the processors of this family turned out to be somewhat overestimated (this applies not only to ours, but also to the vast majority reviews available online). In fact, the performance of Coffee Lake in nominal mode with a heavy multi-threaded load will be somewhere between 3-7 percent lower than that obtained in initial tests, but in reality they will now be able to operate at a more adequate temperature and demonstrate much more moderate power consumption.
Such operation of processors with multipliers, when the frequency drops noticeably under a heavy computational load, and sometimes even below the base passport value, used to be typical only for the HEDT platform, where processors have a significant number of computational cores. However, with the introduction of the Coffee Lake design, ordinary mainstream models have also become multi-core, so there is nothing strange in the fact that the multiplier now dynamically adjusts to consumption in the LGA1151 platform.
That is why Intel has decided to stop describing in detail the values of the turbo frequency under various loads, limiting itself to indicating only the general maximum - the details now do not make much sense. The fact is that the frequencies that are incorporated in the turbo mode may in reality be unattainable. It all depends on the current level of power consumption, and it is not only determined by the nature of the load, but may also vary for different processor instances, depending on the quality of the semiconductor crystal and the nominal VID voltage.
Modern central processors are not easy to understand even for a specialist: many different models are produced, and their names seem to be specially designed to confuse the buyer.
And if a lot has been written about the Core and Core 2 series for almost five years since their appearance, then there is practically no systematized information about the chips of the three latest families of Core i3, i5 and i7, addressed to the consumer, not to the expert.
What are the features of the architecture of new processors, differences from their predecessors?
Finally, why are they better than the still quite relevant Core 2 Duo and Quad?
All processors of the "i" family are based on the latest Nehalem microarchitecture, which replaced the Core at the end of 2008.
The architecture, named after one of the Indian tribes, is an evolutionary development of the Core and differs from it in several fundamental innovations: the placement of all cores on a single chip, an integrated two- or three-channel DDR3 RAM controller, QPI or DMI system buses that replaced the FSB, a cache -memory of the third level, common for all cores, as well as the possibility of embedding a graphics core in a chip.
Nehalem is the first to implement the SSE 4.2 instruction set, their power consumption is 30% less than Core counterparts with comparable performance.
In addition, Hyper-Threading technology has returned to the new chips, which makes it possible to present one physical core as two virtual ones.
The first Nehalem was produced using 45nm technology, and in 2010 a gradual transition to the 32nm process began.
To install processors, a motherboard with LGA1156 or LGA1366 connectors is required.
There are currently four types of desktop processors based on the Nehalem architecture, codenamed Bloomfield, Clarkdale, Gulftown, and Lynnfield.
Of these, Clarkdale are dual-core and manufactured using 32nm technology, Bloomfield and Lynnfield are quad-core and manufactured using 45nm process technology, and Gulftown are 32nm six-core chips.
The bulk of the dual-core i3 and i5 are Clarkdale, the quad-core i5 are Lynnfield, the quad-core i7 are Bloomfield and Lynnfield, and the six-core i7 (it's one so far, it's the 980X) is Gulftown.
Block diagram of the Lynnfield processor
What is the difference between quad-core Bloomfield and Lynnfield?
First of all, Bloomfield has a three-channel memory controller, and Lynnfield has a dual-channel memory controller, which significantly affects the price.
Bloomfield implements a high-speed QPI system bus (25.6 Gb / s), which is used to communicate with the north bridge, which provides the PCI Express 2.0 interface, to which graphics accelerators are connected.
Lynnfield uses the DMI bus (2 Gb/s), and the PCI Express 2.0 graphics bus controller is built into the processor itself, which eliminates the fundamental need for a north bridge and allows the use of a single-chip system logic set - this was done in the Intel P55 Express chipset.
Finally, Lynnfield chips are designed for installation in the "mass" LGA1156 socket, and Bloomfield - in the LGA1366 socket, reserved for top-end systems.
By the way, about the Intel P55 Express chipset: this chipset was designed specifically for Lynnfield, and the LGA1156 processor socket appeared at the same time.
Motherboards based on P55 work without problems with dual-core Core i3 / i5 (Clarkdale), but there is one caveat: this chipset does not support the graphics core built into the processor (more on that below), that is, in any case, you will have to use a discrete video accelerator.
The H57, H55 and Q57 chipsets, presented simultaneously with the Clarksdale processors, work with the integrated graphics core.
The main characteristics of all four sets of logic can be found in the table.
Nehalem processors have a rather confusing labeling system, and even the name of the family does not say much about a particular chip, since they can have different architectures and capabilities.
Therefore, let's take a closer look at their capabilities and functionality.
Core i3 and i5 dual-core processors, quad-core and six-core Core i5 and i7 processors differ from their predecessors primarily in that, like AMD chips, they have built-in DDR3 RAM controllers and an external bus running at 133 MHz.
For comparison, Core 2 Duo (LGA775 socket) is compatible with both DDR3 and DDR2 memory, since the memory controller is implemented there at the system logic level.
In addition, the dual-core Core i3 and i5 have GMA HD graphics accelerators built into the chip.
Their capabilities can be briefly described as follows: if you just want to watch HD-video, and you are not interested in the latest three-dimensional computer games, then the performance of the graphics core built into the processor will be enough.
According to experts, GMA HD is somewhat faster than previous generations of Intel GMA graphics cores embedded in chipsets.
The GMA HD core enables simultaneous decoding of two HD video streams (for example, for picture-in-picture or picture-and-picture modes) and their simultaneous transmission to different digital outputs.
Supports 36-bit color depth and extended xvYCC color space, it is possible to transmit Dolby True HD and DTS-HD Master Audio audio streams.
Declared support for programming interfaces DirectX 10 (Shader Model 3.0) and Open GL 2.1.
Up to 1.7 GB (!) of system memory can be allocated for the frame buffer.
The graphics are fully compatible with HDMI 1.3 Universal Digital Interface.
