Produced on Nehalem, Bloomfield and Gulftown microarchitectures. In this case, the internal clock frequency fluctuates around 3000 MHz. Integrated graphics are not supported by all models. The frequency of the data bus, as a rule, does not exceed 5 GHz per second.
Some configurations are equipped with unlocked multipliers. To learn more about processors, you should look at Intel Processors Core i7 on specific microarchitectures.
CPU based on Nehalem microarchitecture
The Core processor has a clock speed of 2.8 GHz. In this case, there are four cores. The bus frequency of the CPU reaches 2400 MHz. The maximum voltage the system can withstand is 1.4 V. The Intel Core model is released for four cores. It has a clock frequency parameter of 2.53 GHz. The CPU multiplier is of the unlocked type. The main bus frequency fluctuates around 2400 MHz. The Core i7 2700K model has a clock speed of 2.93 GHz. The specified modification for four cores has an LGA connector. The bus frequency itself does not exceed 2400 MHz.
Bloomfield lineup
4720 has four cores. In this case, the chip area is 263 mm 2 . The clock frequency itself is 2.6 GHz. The configuration of the Core i7 4730 is provided for four cores. In total, 731 million transistors are involved in it. The CPU clock frequency is 2.8 GHz. The Intel modification is rated at 3.07 GHz. In this case, the chip area is 263 mm 2 . The bus itself is available at 213 MHz.
CPU based on Gulftown microarchitecture
The Core i7 970 model was released by the manufacturer for six cores. Its clock frequency does not exceed 3.2 GHz. The bus is available for the model at 2660 MHz. The Core i7 980 has a clock speed of exactly 3.3 GHz. The area of the chip in this situation is 239 mm 2 . The bus itself is provided for 2660 MHz. Core i7 processor 990 transistors has 1170 million units. The clock frequency of the model does not exceed 3.4 GHz. The LGA connector is supported in this case.
Main functions
The area of high-speed memory in processors based on the Gulftown microarchitecture is very extensive, so the Intel Core i7 deserves good reviews from the owners. The cache memory is directly related to the architecture. The kernels of the model are used dynamically. Thus, the system provides high performance. If we consider the Intel Core i7 4790, then the IM bus in this case is provided for 5 MHz. It plays an important role in the exchange of information.
The system bus in the processor on the Gulftown microarchitecture is used by CB. For transferring data to the controller unit, it fits perfectly. The interface is provided by the manufacturer with MI support. Direct connection is made through the system board. All major operating commands are supported by it.
Performance
The Intel Core i7 laptop is capable of supporting a maximum of four threads. In this case, the base frequency parameter is quite high. An IP program is provided for ordering instructions. Direct data processing does not take much time. It is also important to note that the clock frequency parameter directly depends on the speed of computing cycles.
Estimated power in Intel processors is specified through a dot. The maximum frequency setting is 38 GHz. Directly, the CPU power on the Gulftown microarchitecture is at the level of 83 watts. When running at the base frequency, all cores are used in the processor.
Memory Specifications
The Intel Core i7 CPU based on the Gulftown microarchitecture boasts a large amount of memory. In this case, it is supported by various formats. The number of channels directly affects system performance. There are two of them in this version. Additionally, it is important to mention that the Intel CPU supports flex memory.
Throughput is at a very high level. In this case, reading data does not take much time. This was largely achieved by supporting dual-channel memory. High data storage speed is another advantage of this system. ECC memory is supported by processors. The standard chipset for this is set.
Graphics Subsystem Specifications
On the Gulftown microarchitecture, the graphics frequency setting is at 350 MHz. In this case, it is also important to consider the render rate. It affects the base frequency quite strongly. The graphics subsystem itself allows you to significantly increase rendering.
Support for the NS format is provided for Intel models. If we consider the Intel Core i7 2600K, then the maximum system size is at the level of 1.7 GB. For interface support, this indicator is very important. It also affects memory availability. To increase the interaction of a personal computer with a processor, the PPC system is used. Its resolution is 4096 x 2304 pixels.
Direct support
When it is important to mention the support of "Direct". In this case, specific application collections are taken into account. "Direct" series 11.1 is excellent for processing system files. If we talk about the graphic component, it is important to mention the "Open Graph" system. It affects the calculation of tasks quite strongly. In this case, much depends on the support of multimedia files.
The Libera system is designed to display two-dimensional graphics. If we talk about the "Quick Video" technology, then in this case you need to take into account the speed of conversion. If you believe the opinion of experts, then the system interacts normally with portable media players. Another technology "Quick Video" affects the speed of video editing. In addition, it provides the placement of important information on the safety of work on the Web. Creating a video using this technology is very simple.
Expansion options
The Intel Core i7 computer uses the Express edition for data transfer. To date, there are many versions of it, which, in fact, do not differ much. However, in general, the Express edition is very important when it comes to connecting to personal computer various devices.
If we talk about version 1.16, then it can significantly increase the data transfer rate. This system can work only with devices of the PC type. It allows you to play up to 16 channels directly. At the same time, the basic modulator of the central processor is not involved in data processing.
Data protection technology
This technology allows you to work with the AE system, which is a set of commands. Due to it, you can quickly perform data encryption. This makes the process safe. The AE system is also used to decrypt data. Many tools of the program allow you to solve a wide range of tasks. In particular, the AE system is capable of handling cryptographic data. She solves problems with applications pretty quickly.
The Data Project technology itself was created to decipher random numbers. They provide authentication. Additionally, it should be noted that the Data Project technology includes the Kay system. It is designed to generate random numbers. It helps a lot in creating unique combinations. Also, the "Key" system is involved in the decoding of algorithms. It is well suited for enhancing data encryption.
Platform Protection Technology
The "Platform Protection" technology of the "Intel" CPU is provided for the 10.1 series. Speaking about it, first of all, it is important to mention the "Guard" system. It was created for safe work with various applications. In this case, you can perform various operations with them.
To connect microcircuits, the "Guard" system is also used. The Trusted program is used directly to protect the platforms. It allows you to work with a digital office. The measured launch feature is supported by Platform Protection technology.
Also available is the option of secure command execution. In particular, the system is able to isolate some flows. At the same time, running applications do not affect them. To cancel hardware programs, the Anti-Tef system is used. In this case, the vulnerability of the CPU is significantly reduced. The Anti-Tef system is also designed to fight malware.
2017 was a real test for Intel, something that has not been seen for many years after the debut of the Intel Core line on the market. First of all, this is due to the release of a very successful line, which required Intel to hastily prepare the third generation of 14nm processors in order to strengthen its positions.
Under other circumstances, Intel could have ditched the 14nm Intel Coffee Lake and Intel Kaby Lake R lines (8th generation mobile Intel Cores) entirely, focusing its resources on accelerating the launch of the 10nm Intel Ice Lake and Intel Cannon series. Lake respectively. Moreover, the computing power of Intel Kaby Lake processors is quite enough for a wide range of home, educational or office computers. But the competitor left no choice.
The first 8th generation Intel Core models were introduced at the end of August. They are aimed at the mobile market, and many laptop manufacturers have already announced new or updated products based on them. At the end of September, a presentation of the desktop line took place along with the Intel Z370 chipset, which we will talk about in a separate article.
The first on sale will be six models of processors, each of which is a landmark for its series. So, Intel Core i3-8100 and Intel Core i3-8350K are the first full-fledged 4-core CPUs in this series, which previously had only 2-core, 4-thread solutions. The line of Intel Core i5 for the first time replenished with 6-core, 6-thread representatives - Intel Core i5-8400 and Intel Core i5-8600K. And the Intel Core i7 series is now dominated by the 6-core, 12-thread Intel Core i7-8700 and Intel Core i7-8700K models, which have replaced the 4-core, 8-thread models. In the first half of 2018, the list of available processors in each series will be expanded. Other Intel 300-series chipsets will also appear and motherboards based on them.
8th generation Intel Core solutions are positioned primarily for gamers, content creators and overclockers. They will be especially useful in cases where the software is optimized for multithreading. In addition, Intel processors are traditionally characterized by excellent performance in single-threaded mode, so even in outdated applications and games they look decent.
Gamers are promised a performance increase of up to 25% (recorded in Gears of War 4 when comparing systems based on Intel Core i7-8700K and Intel Core i7-7700K) and a comfortable frame rate in multitasking mode, when you need to not only play, but simultaneously record a game session and broadcast it on the internet.
There's also mouth-watering facts for content creators: Up to 32% faster 4K video editing (Intel Core i7-8700K vs. Intel Core i7-7700K). And if we compare the performance of Intel Core i7-8700K and Intel Core i7-4790K (Intel Devil`s Canyon), then we can count on 4.5 times acceleration when creating HEVC video in PowerDirector, 65% when editing files in Adobe Photoshop Lightroom and 7.8x when transcoded to Handbrake Transcode.
In turn, overclockers are bribed with new features: overclocking a single core, increasing the memory multiplier to 8400 MT / s, monitoring memory delays in real time, and others. If you are afraid of a possible processor failure as a result of overclocking experiments, then you can optionally buy Performance Tuning Protection Plan. It allows you to replace the CPU once in case of damage during freelance operation. The cost of such a plan depends on specific model. For example, for Intel Core i7-7700K it is set at $30, and owners of Intel Core i9-7980XE will need to pay $150.
There is no mention of any microarchitectural changes in the presentation, although you can admire the wonders of engineering thought embodied in the crystals themselves.
The main emphasis in press materials is on the increase in the number of physical cores and cache memory, advanced overclocking capabilities and the use of an improved 14-nm process technology. More specifically, Intel Skylake is manufactured using 14 nm, Intel Kaby Lake is 14+ nm, and Intel Coffee Lake is 14 ++ nm.
In turn, the use of the new chipset is explained by increased requirements for the power subsystem due to the increased number of cores, support for new overclocking capabilities and faster DDR4-2666 memory.
At the hardware level, the incompatibility of new and old processors manifests itself in a different number of VCC pads of the Socket LGA1151 connector: Intel Coffee Lake has 146 of them, while Intel Kaby Lake and Intel Skylake have 128. An additional 18 were obtained by activating reserve pads, without making any or physical changes. That is, you can install a new processor on old motherboards or old processors on new motherboards, but such bundles will not work. Therefore, for Intel Coffee Lake, it is mandatory to buy a motherboard based on Intel 300 series chipsets.
Intel did not forget to remind about a companion product - Intel Optane Memory, which can significantly increase system responsiveness and speed up application launch. Although with the current volume (16 / 32 GB) and price level, it is difficult for him to compete in the market with the same M.2 or conventional 2.5-inch SSDs.
We got acquainted with the presentation, now it's time to move on to a more detailed study of the capabilities of the hero of this review - IntelCorei7-8700 K, which is also the flagship of the 8th generation of the Intel Core line.
Specification
|
Processor socket |
|
|
Base / dynamic clock frequency, GHz |
|
|
base multiplier |
|
|
Base system bus frequency, MHz |
|
|
Number of cores / threads |
|
|
L1 cache size, KB |
6 x 32 (data memory) |
|
L2 cache size, KB |
|
|
L3 cache size, MB |
|
|
microarchitecture |
Intel Coffee Lake |
|
codename |
Intel Coffee Lake-S |
|
Maximum design power (TDP), W |
|
|
Process technology, nm |
|
|
Critical temperature (T junction), °C |
|
|
Support for instructions and technologies |
Intel Turbo Boost 2.0, Intel Optane Memory, Intel Hyper-Threading, Intel vPro, Intel VT-x, Intel VT-d, Intel VT-x EPT, Intel TSX-NI, Intel 64, Execute Disable Bit, Intel AEX-NI, MMX, SSE, SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, EM64T, AES, AVX, AVX 2.0, FMA3, Enhanced Intel SpeedStep, Thermal Monitoring, Intel Identity Protection, Intel Stable Image Platform Program (SIPP) |
|
Built-in memory controller |
|
|
Memory type |
|
|
Supported frequency, MHz |
|
|
Number of channels |
|
|
Maximum memory, GB |
|
|
Integrated Intel UHD Graphics 630 |
|
|
Number of execution units (EU) |
|
|
Base / dynamic frequency, MHz |
|
|
Maximum video memory (allocated from RAM), GB |
|
|
Maximum screen resolution at 60 Hz |
|
|
Maximum supported displays |
|
|
Supported technologies and APIs |
DirectX 12, OpenGL 4.5, Intel Quick Sync Video, Intel InTru 3D, Intel Clear Video HD, Intel Clear Video |
|
Products webpage |
|
|
Processor Page |
|
Packaging, scope of delivery and appearance
Intel kindly provided us with an engineering sample of the Intel Core i7-8700K for testing without the appropriate packaging and delivery kit. Therefore, we will use the official press materials to evaluate the appearance of the box. Its front side unmistakably indicates that the processor belongs to the 8th generation of the Intel Core line and the corresponding series, and key advantages are listed on one of the sidewalls. It also indicated the need to use new products exclusively with motherboards based on Intel 300 series chipsets. The packages themselves also differ in thickness, that is, there will be options for sale with and without a complete cooler.
AndIntel Core i7-7700K
Externally, the Intel Core i7-8700K does not differ from its predecessor, of course, if you do not take into account the markings and other designations on the heat distribution cover. The very designation of the retail sample of the novelty will be different. First, instead of the inscription "Intel Confidential" the name of the model (Intel Core i7-8700K) will be indicated. Secondly, there will be a different Spec code instead of "QNMK". And, of course, the FPO code will change. In this case, it tells us that the processor was manufactured in Malaysia on the 19th week of 2017 (from 08.05 to 14.05).
AndIntel Core i7-7700K
On the reverse side, there are contact pads for the Socket LGA1151 connector. As we already know, their physical location has not changed, but the functional purpose of some legs has changed, which requires the use of new motherboards with Intel Coffee Lake processors.
Analysis specifications
To test the Intel Core i7-8700K, we used the ROG STRIX Z370-F Gaming motherboard and our stock Scythe Mugen 3 cooling system. First, we deactivated Intel Turbo Boost 2.0 technology and got the processor frequency at 3.7 GHz at 1.12 V .
The maximum frequency under load (AIDA64) with Intel Turbo Boost 2.0 technology enabled reached the declared 4.7 GHz in the specification. The temperature rose to 96°C, but there was no throttling.
When the system was idle, the processor frequency remained at 4.7 GHz, although the temperature dropped below 50°C.
If you put the system into power saving mode, then the speed of the Intel Core i7-8700K drops to 800 MHz.
Cache structure of Intel Core i7-8700 processorsKand Intel Core i7-77 00K
The cache structure of the novelty is as follows:
- 32 KB of L1 cache per core with 8 associativity channels is reserved for instructions and the same amount for data;
- 256 KB L2 cache with 4 associativity channels per core;
- 12MB shared L3 cache with 16 associativity channels.
Compared to its predecessor, the cache memory of each level has increased in proportion to the increased number of cores: L1 - by 64 KB for data and instructions, L2 - by 512 KB, and L3 - by 4 MB.
The built-in RAM controller is guaranteed to support DDR4-2666 MHz modules in 2-channel mode. Of course, you can try to overclock the RAM to higher frequencies at your own peril and risk, but there are no guarantees anymore and it all depends on the quality of the bars themselves, the capabilities of the motherboard and the user's skills. The maximum available RAM is 64 GB.
The maximum temperature on the official website is stated at 100 ° C. A similar indicator is also reported by AIDA64.
The Intel Core i7-8700K processor has an integrated Intel UHD Graphics 630 graphics core, which at the time of writing was poorly detected by the GPU-Z and AIDA64 utilities. According to official information, it includes 24 execution units and can use all available 64 GB of RAM for its needs. The base frequency of its operation is 350 MHz, and the dynamic frequency can be increased up to 1200 MHz.
While simultaneously loading the CPU and iGPU cores using the AIDA64 and MSI Kombustor benchmarks, the frequency of the processor cores remained at 4.7 GHz. But at the same time, the temperature rose to 99 ° C and throttling was observed.
Testing
When testing, we used the Stand for testing Processors No. 2
| Motherboards (AMD) | ASUS F1A75-V PRO (AMD A75, Socket FM1, DDR3, ATX), GIGABYTE GA-F2A75-D3H (AMD A75, Socket FM2, DDR3, ATX), ASUS SABERTOOTH 990FX (AMD 990FX, Socket AM3+, DDR3, ATX) |
| Motherboards (AMD) | ASUS SABERTOOTH 990FX R2.0 (AMD 990FX, Socket AM3+, DDR3, ATX), ASRock Fatal1ty FM2A88X+ Killer (AMD A88X, Socket FM2+, DDR3, ATX) |
| Motherboards (Intel) | ASUS P8Z77-V PRO/THUNDERBOLT (Intel Z77, Socket LGA1155, DDR3, ATX), ASUS P9X79 PRO (Intel X79, Socket LGA2011, DDR3, ATX), ASRock Z87M OC Formula (Intel Z87, Socket LGA1150, DDR3, mATX) |
| Motherboards (Intel) | ASUS MAXIMUS VIII RANGER (Intel Z170, Socket LGA1151, DDR4, ATX) / ASRock Fatal1ty Z97X Killer (Intel Z97, Socket LGA1150, DDR3, mATX), ASUS RAMPAGE V EXTREME (Intel X99, Socket LGA2011-v3, DDR4, E-ATX ) |
| Coolers | Scythe Mugen 3 (Socket LGA1150/1155/1366, AMD Socket AM3+/FM1/ FM2/FM2+), ZALMAN CNPS12X (Socket LGA2011), Noctua NH-U14S (LGA2011-3) |
| RAM | 2 x 4 GB DDR3-2400 TwinMOS TwiSTER 9DHCGN4B-HAWP, 4 x 4 GB DDR4-3000 Kingston HyperX Predator HX430C15PBK4/16 (Socket LGA2011-v3) |
| video card | AMD Radeon HD 7970 3 GB GDDR5, ASUS GeForce GTX 980 STRIX OC 4 GB GDDR5 (GPU-1178 MHz / RAM-1279 MHz) |
| HDD | Western Digital Caviar Blue WD10EALX (1TB, SATA 6Gb/s, NCQ), Seagate Enterprise Capacity 3.5 HDD v4 (ST6000NM0024, 6TB, SATA 6Gb/s) |
| Power Supply | Seasonic X-660, 660 W, Active PFC, 80 PLUS Gold, 120 mm fan |
| Operating system | Microsoft Windows 8.1 64-bit |
Choose what you want to compare Intel Core i7-8700K Turbo Boost ON Enhanced Performance to
We were in a hurry to prepare the material for the release of new products on sale, so we did not have time to test the Intel Core i7-8700K with Intel Turbo Boost 2.0 technology disabled. Usually, dynamic overclocking allows you to increase the performance level by a few percent, so it's best not to disable it yourself.
To begin with, let's analyze the situation in the internal model range. In synthetic tests, the Intel Core i7-8700K outperformed the previous flagship by an average of 39%. In games, the performance bonus was only 2%, since since the testing of the 4-core model, many gaming benchmarks have been replaced. In turn, the integrated graphics core Intel UHD Graphics 630 turned out to be on average 11% better than its counterpart, however, its gaming capabilities are still limited to undemanding projects with low quality settings in Full HD.
The comparison with the recently tested 8-core (16-thread) processor of the Intel Core X line turned out to be more interesting and intense. In synthetic tests, it came out ahead by an average of 1%, and parity was recorded in gaming tests. The difference between them in the recommended price tags is $240 ($359 vs. $599). That is, the Intel Core i7-8700K strikes not only at the positions of AMD's opponents, but also at Intel's own HEDT lineup.
And now, actually, about competitors. These include the 8-core AMD Ryzen 7 1700 ($349) and the 6-core AMD Ryzen 5 1600X ($249). But they haven't been tested by us yet, so we compared the results of the novelty with (nominally $440, but now the average price has dropped to $389) and (nominally $219, but now $240). In synthetics, the Intel Core i7-8700K outperformed the Ryzen 7 1700X by 17% and the Ryzen 5 1600 by 43%. But in games, the situation turned out to be interesting. The superiority of the novelty over the 8-core opponent was almost 5%, but the Ryzen 5 1600 is already pulling ahead by the same 5%. And all thanks to the low minimum Intel Core i7-8700K in the Tom Clancy's Rainbow Six Siege test. If you ignore it, then new flagship in games, it is 3% ahead of Ryzen 5 1600 and Intel Core i7-7820X. The results of the comparison with the Ryzen 7 1700X do not change, since this processor was not tested in it.
The situation with energy consumption is also very curious. A test system with an Intel Core i7-8700K and a discrete graphics card required a maximum of 276 watts. This is even more than a bunch of 8-core Intel Core i7-7820X (242W) and AMD Ryzen 7 1700X (182W). Perhaps this applies only to our engineering sample and the versions on sale have a more balanced power consumption and heat dissipation.
Overclocking
Already when analyzing the technical characteristics of the Intel Core i7-8700K processor, we fixed the processor throttling under a significant load in the nominal mode. That is, our test cooling system could not cope with its cooling. Again, this may be solely related to the test engineering sample, and in normal retail versions temperature regime will be much better.
Nevertheless, we failed to manually overclock the test instance: raising even up to 4.8 GHz led to active throttling and frequency reset. And only thanks to automatic overclocking on the motherboard ROG board STRIX Z370-F Gaming in "TPU II" mode managed to increase the core frequency to 5.0 GHz with a multiplier of "x50" and reduce the frequency by 300 MHz when executing AVX instructions. At the same time, the RAM speed was increased to 3200 MHz, and the maximum temperature during testing did not exceed 94 ° C, which allowed the system to work stably.
You can evaluate the impact of overclocking on performance using the following table:
|
Nominal |
Overclocked |
|||
|
Fritz Chess Benchmark 4.3 |
||||
|
Heavy Multitasking |
||||
|
1920x1080, DX12, Very High |
||||
|
Tom Clancy's The Division |
1920x1080, DX11 |
|||
|
1920x1080, DX11 |
||||
|
Mean |
||||
The average increase was 4.49%. Synthetic tests responded best to the increase in frequency, providing a bonus from 4% to 7%. But in games, the maximum recorded increase was 3%.
Results
What did we end up with? First of all, to be commended Intel for adding more cores and threads to Intel's Coffee Lake desktop processors, regardless of the reasons that led her to such a step. Secondly, the additional cores come with their own cache memory of all three levels, which also contributes to an increase in the overall performance level. This is especially noticeable in synthetic tests, where the 6-core is on average 39% ahead of the 4-core flagship of the previous generation and practically does not lag behind the more expensive 8-core Intel Core X series. In turn, overclockers will certainly like additional overclocking options.
Now to the weaknesses of the tested engineering sample. The first is the high heat dissipation: even under nominal load with a fairly powerful Scythe Mugen 3 tower cooler, the temperature rose to 96°C. For this reason, we were unable to carry out manual overclocking, and automatic overclocking allowed us to increase the speed to 5 GHz with a decrease to 4.7 GHz under load in the benchmark. Second, power consumption test stand was higher than comparable 8-core Intel and AMD processors. Thirdly, in games there is no noticeable preponderance of new items over competitors.
, Kingston , Noctua , Sea Sonic , Seagate , Scythe AndTwinMOS Technologies for the equipment provided for the test bench.Article read 37079 times
| Subscribe to our channels | |||||
The first processors under the Intel Core i7 brand appeared nine years ago, but the LGA1366 platform did not claim mass distribution outside the server segment. Actually, all the "consumer" processors for it fell into the price range from ≈$300 to full-fledged "piecebucks", 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 customers (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 cache memory of the third level."
Later, it was also inherited by models for the mass market LGA1156. Later, without changes, migrated to LGA1155. Even later, it was "noted" 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 partly share) - a bit late: could have been earlier. However, the “good, but not enough” claim applies not only to processor performance, but in general to any evolutionary changes in any market. The reason for this lies not in the technical, but in the psychological plane, which is far beyond the scope of our site's interests. Here's a test computer systems different generations to determine their performance and power consumption (even if only for a limited sample of tasks) we can. What are we going to do today.
Test stand 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-alle opens with three of the 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 the participants in today's testing: its recommended price was $562. In the future, no desktop quad-core Core i7 cost so much. And the quad-core processors of the Sandy Bridge family (as in the previous case, we have a representative of the second wave here, and not the "starter" i7-2600K) are the only ones of all 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 back too: it was later that the thermal interface in narrow but noisy circles began to be endowed with truly magical properties. Somewhere starting from 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 |
| Cache L3 (L4), MiB | 8 | 6 (128) |
| RAM | 2×DDR3-1600 | 2×DDR3-1600 |
| TDP, W | 88 | 65 |
The one we're 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, the 4790K is Haswell Refresh. Some were already waiting for Broadwell, but the company released processors of this family exclusively to the tablet and laptop market: where they were most in demand. And with desktop plans, they 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 they were only compatible with a couple of 2014 chipsets, and most importantly, they turned out to be the only “socket” models with a four-level cache. Formally - for the needs of the graphics core, although in practice L4 can be used by all programs. There were similar processors earlier and later - but only in the BGA version (that is, they were soldered directly to the motherboard). These are unique in their own way. Enthusiasts, of course, were not inspired due to low clock speeds and limited "overclocking", but we will check how this "side escape" correlates with 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 of its incompatible versions. However, we wrote about the difficult path of mass-produced six-core processors to the market quite recently: when they were tested for the first time. So we won't repeat ourselves. We only note that we tested the i7-8700K again: using not a preliminary, but a “release” copy, and even installing it on an already “normal” board with debugged firmware. The results have changed slightly, but in several programs have become 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 it is imperative to take a couple of the fastest modern dual- and quad-core processors of the Core i3 and Core i5 lines, since they have already been tested, and it’s interesting to see which of the old people they will catch up with and where (and whether they will catch up). We also managed to get our hands on a brand new six-core Core i5-8400, so we took the opportunity to test that as well.
| 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 "historic" FX-8350, which is the same age as the Core i7-3770K. Fans of this line have always argued that it is not only cheaper, but generally better - just few people know how to cook it. But if you use the "right programs", then you will immediately overtake everyone. Since this year we have at the request of workers reworked the testing methodology in the direction of "harsh multi-threading", so there is a reason to test this hypothesis - all the same, testing is historical. And modern models will require at least two. We would love a Ryzen 5 1500X, very similar to the old Core i7s, but we haven't tested it. Ryzen 5 1400 formally also fits ... but in fact, this model (and modern Ryzen 3) along with the halving of the cache memory "suffered" and the links between the CCX. Therefore, I also had to take the Ryzen 5 1600, where this problem is not present - as a result, it often overtakes the 1400 by more than one and a half times. Yes, and a couple of six-core Intel processors are also present in today's testing. Others are clearly too slow to compare with this inexpensive processor, but oh well - let it dominate.
Test Methodology
Methodology. Here we briefly recall 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 2017 games
Detailed results for all tests are available as a complete results spreadsheet (Microsoft Excel 97-2003 format) . Directly in the articles, we use already processed data. This is especially true for application tests, where everything is normalized relative to the reference system (AMD FX-8350 with 16 GB of memory, GeForce GTX 1070 graphics card and Corsair Force LE 960 GB SSD) and grouped by computer application areas.
iXBT Application Benchmark 2017
In principle, the assertions of AMD fans that FX were not so bad in "harsh multithreading", if we consider only performance, are justified: as we can see, the 8350, in principle, could compete on equal terms with the Core i7 of the same year of manufacture. However, here it also looks good against the background of the younger Ryzen, but between these two families, almost nothing was produced by the company for this market segment. Intel, on the other hand, has such a uniform lineup, which made it possible to double the performance within the framework of the “quad-core” concept. Although the cores are of great importance here - the best dual-core processor of 2017 still did not catch up with the quad-core Core of the "previous" generation (recall that it is still officially called so in the company's materials, clearly separated from the numbered ones starting from the second). And six-core models are good - and that's all. So Intel's reproaches that the company delayed their entry to the market too much can be considered fair to some extent.
All the 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 is quite comparable: the difference between the 880 and 7700K is still twofold, the i5-8400 is still second only to the latter, the i3-7350K still has not caught up with anyone. And it happened in the same seven years. We can assume that 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.
One has only to “weaken” the utilization of multithreading a little, and the position of newer processors immediately improves - albeit quantitatively weaker ones. However, the traditional "two ends" with other (relatively) equal comparison of the "previous" and "seventh" generations of Core gives us. Although it is easy to see that the “second” and ... “eighth” are drawn to the maximum extent for the “revolutionary”. But this is more than understandable: the latter increased the number of cores, and in the "second" the microarchitecture and process technology changed radically, and at the same time.
As we already know, Adobe Photoshop is a little “weird” (bad news - the problem has not been fixed in the latest version of the package at the moment; very bad news - now it will also be relevant for the new Core i3), so we do not consider processors without HT. But our main heroes have support for this technology, so no one bothers them all to work normally. As a result, in general, the state of affairs is similar to other groups, but there is a nuance: the most fast processor for LGA1150 it turned out not to have a high frequency i7-4790K, but i7-5775C. Well - in some places, intensive methods of increasing productivity are very effective. It’s a pity that not always: it’s easier to “work” with frequency. And it's cheaper: you don't need an additional eDRAM crystal, 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 - more than even the frequency. Although the Core i7-8700K looked worse in our first test, 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 have previously been positive about increasing the number of supported computation threads while maintaining a similar level of such frequency.
Moreover, sometimes only she “decides” and the number of computation threads. Basically, of course, there are certain nuances here, but “ there is no reception against scrap". The whole revolutionary architecture of Ryzen, for example, only allowed the 1400 to deliver performance on par with the FX-8350 or Core i7-3770K that hit the market in 2012. Given that its frequency is lower than both, and indeed this is a special budget model that actually uses only half of the semiconductor crystal, it's not so bad. But reverence does not cause. Especially against the background of another (and also inexpensive) representative of the Ryzen 5 line, which easily and noticeably overtook any quad-core Core i7 of any year of production :)
Although we abandoned the single-threaded unpacking test, this program still cannot be considered too “greedy” for cores and their frequency. It is clear why - the performance of the memory system is very important here, so the Core i7-5775C managed to overtake only the i7-8700K, and even then by less than 10%. It is a pity that there are no products so far where L4 is combined with six cores and memory with a high memory bandwidth: such a processor "without bottlenecks" in such tasks could show a miracle. Theoretically, at least, it is obvious that in desktop computers we will not see anything like this in the near future for sure.
It is characteristic that this offshoot from the "main line" of desktop processors demonstrates (so far!) high results in this group of programs as well. However, what unites them is mainly the intended purpose, and not the optimization methods chosen by the 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 somewhat decreased: 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, not bad: it 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, it is not enough. Especially if we compare the gains at each annual step, adding another Core i7-4770K (which is why we regretted above that this processor was not found).
At the same time, the company has had the opportunity to dramatically increase productivity at least in multi-threaded software (and this has long been a lot among resource-intensive programs) for a long time. Yes, and it was also implemented - but within the framework of completely different platforms with their own characteristics. Not without reason, many have been waiting for six-core models under LGA115x since 2014 ... But many did not expect any breakthroughs from AMD in those years - the first Ryzen tests turned out to be all the more impressive. No wonder - as we can see, even the inexpensive Ryzen 5 1600 can compete in performance with the Core i7-7700K, which was the fastest LGA1151 processor just a couple of months ago. Now a similar performance level is quite available for the Core i5, but it would be better if it happened earlier :) In any case, there would be less reason for complaints.
Energy consumption and energy efficiency
However, this diagram once again demonstrates why the performance of mass 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 took place against the background of a “non-increase” in power consumption. If possible, even reduce. It was possible to reduce it by architectural or any other methods - users of mobile and compact systems (which have long been sold much more than “typical desktop ones”) will be satisfied. Yes, and on the desktop market, a small step forward, since you can tweak the frequencies a little more, which was done in the Core i7-4790K at one time, and then entrenched in the “regular” Core i7, and even in the Core i5.
This is especially clearly seen in the evaluation of the power consumption of the processors themselves (unfortunately, for the LGA1155 it is impossible to measure it separately from the platform using simple tools). At the same time, it becomes clear why the company does not need to somehow change the requirements for processor cooling within the LGA115x line. Also, why more and more products in the (formally) desktop assortment begin to fit into the thermal packs traditional for laptop processors: this happens by itself 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. overclocking processors, the company considers it necessary to tighten the requirements for the cooling system, since there is a small (but not zero) chance that the buyer of a computer with such will overclock it and use all sorts of "stability tests". And mass 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 the LGA1150. In normal mode, of course - during overclocking, you can inadvertently return to 2010 :)
But, at the same time, modern economical processors are not necessarily slow at all - three to five years ago, the performance of "energy efficient" models against the background of 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 much faster than pure performance: here, when comparing the Core i7-7700K and i7-880, not twice, but all two and a half. However... the first "big leap" and immediately one and a half times fell on the introduction of LGA1155, so it's 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 Core i7-880 and i7-2700K, are of the greatest interest. Unfortunately, nothing good came of 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. Yes, and it’s clear why: many LGA1156 missed it altogether, or have already managed to migrate from it to other solutions for so many years. But with the Core i7-2700K there is another problem: its performance (recall - in normal mode) is still often enough to work at the level of the new Core i7. In general, such an indestructible legend: which (together with the older Core i5 for LGA1155) was first made a good gaming processor by high single-threaded performance (in those years, Intel strongly "clamped" Core i3 and Pentium in frequency), and then they started more or less efficiently all eight supported computation threads are utilized. Although the same level of performance in games is often achieved by more “simple” solutions for new platforms, there is sometimes a feeling that this is due not only and not so much to “pure” performance. Therefore, for those who are interested in the results in games to some extent, we recommend that you familiarize yourself with them using the full table, and here we will give only a couple of the most interesting and revealing diagrams.
Take Far Cry Primal for example. We immediately discard the results of the Core i7-880: the incorrect operation of the video card on the GTX 1070 with this platform is obvious. Perhaps, by the way, this is also common for LGA1155, although in general the frame rate cannot be called low here: in practice it is enough. But clearly lower than it could be. And LGA1151 also somehow does not shine, and LGA1150 looks like the best platform. Now we remember that a modified version of the Dunia Engine 2 (it is used here) was developed between 2013 and 2014, so they could just re-optimize. An indirect confirmation of which 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 don’t see such artifacts anymore. With the exception of the Core i7-880, which once again amused by the "brakes", but this correlates well with other games. And not just multi-core processors look best, but also those released since 2015, i.e. LGA1151 and AM4 platforms. The complete 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 in which there are more computing cores. But within one- different (especially, significantly different architecturally) with their help, you need to compare very carefully. If you want to compare, of course: in general, in both (and not only in them) on a system with a five-year-old processor and a “good” video card, you can play with much more comfort than with any processor, but on a budget video card for $ 200 In general, do games have requirements for processors or not, and gaming computer you need to collect "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 at all, and not even tripled;)
Total
Actually, all we wanted to do was compare several processors of different years at once when working with modern software. Moreover, some characteristics of the older Core i7 models have not changed much 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 oft-discussed "5% per year." And taking into account the fact that every year a normal user does not buy computers, but usually focuses on 3-5 years, during such a period there was an increase in performance, economy, and platform functionality. But could have been better. At the same time, some “weak points” are clearly visible: for example, an increase in clock frequency in 2014 did not allow achieving significantly higher performance either in 2015 or even at the beginning of 2017. We managed to “break away” noticeably from LGA1155 (as the software was optimized for processors, starting with Haswell, the results were more modest at the start), and that’s it. And then (suddenly) +30% performance, which was not there for a long time. In general, from a historical point of view, a smoother implementation of this process would look better. But what has been has already been.
Almost always, under any publication that somehow touches on the performance of modern Intel processors, sooner or later there will be several angry reader comments that progress in the development of Intel chips has stalled for a long time and there is no point in switching from the “good old Core i7-2600K ' to something new. In such remarks, most likely, it will be irritably mentioned about the increase in productivity at an intangible level of "no more than five percent per year"; about the low-quality internal thermal interface, which irreparably ruined modern Intel processors; or about the fact that in modern conditions buying processors with the same number of computing cores as several years ago is generally the lot of short-sighted amateurs, since they do not have the necessary groundwork for the future.
There is no doubt that all such remarks are not without foundation. However, it is very likely that they exaggerate the problems many times over. The 3DNews laboratory has been testing Intel processors in detail since 2000, and we cannot agree with the thesis that any kind of their development has come to an end, and what is happening with the microprocessor giant in recent years can no longer be called anything other than stagnation. Yes, some cardinal changes rarely occur with Intel processors, but nevertheless they continue to be systematically improved. Therefore, those chips of the Core i7 series that you can buy today are obviously better than the models offered several years ago.
| Generation Core | codename | Process technology | Development stage | Exit time |
| 2 | Sandy Bridge | 32 nm | So (Architecture) | I sq. 2011 |
| 3 | IvyBridge | 22 nm | Tick (Process) | II quarter. 2012 |
| 4 | Haswell | 22 nm | So (Architecture) | II quarter. 2013 |
| 5 | Broadwell | 14 nm | Tick (Process) | II quarter. 2015 |
| 6 | skylake | 14 nm | So (Architecture) |
III quarter. 2015 |
| 7 | KabyLake | 14+ nm | Optimization | I sq. 2017 |
| 8 | CoffeeLake | 14++ nm | Optimization | IV quarter. 2017 |
Actually, this material is precisely the counterargument for reasoning about the worthlessness of the strategy chosen by Intel for the gradual development of consumer CPUs. We decided to collect in one test the older Intel processors for mass platforms over the past seven years and see in practice how much the representatives of the Kaby Lake and Coffee Lake series have gone ahead relative to the "reference" Sandy Bridge, which over the years of hypothetical comparisons and mental oppositions in the minds of ordinary people have become a true icon of the processor industry.
⇡ What has changed in Intel processors from 2011 to the present
The starting point in the recent history of the development of Intel processors is considered to be the microarchitecture SandyBridge. And this is no accident. Despite the fact that the first generation of processors under the Core brand was released in 2008 based on the Nehalem microarchitecture, almost all the main features that are inherent in the modern mass CPUs of the microprocessor giant came into use not then, but a couple of years later, when the next generation became widespread. processor design, Sandy Bridge.
Now Intel has accustomed us to frankly unhurried progress in the development of microarchitecture, when there are very few innovations and they almost do not lead to an increase in the specific performance of processor cores. But just seven years ago, the situation was radically different. In particular, the transition from Nehalem to Sandy Bridge was marked by a 15-20% increase in IPC (the number of instructions executed per cycle), which was due to a deep reworking of the logical design of the cores with an eye to increasing their efficiency.
Sandy Bridge laid down many principles that have not changed since then and have become standard for most processors today. For example, it was there that a separate zero-level cache appeared for decoded micro-operations, and a physical register file began to be used, which reduces energy consumption during the operation of algorithms for out-of-order execution of instructions.
But perhaps the most important innovation was that Sandy Bridge was designed as a unified system-on-a-chip, designed simultaneously for all classes of applications: server, desktop and mobile. Most likely, public opinion put him, and not some Nehalem, and even less Penryn, as the great-grandfather of modern Coffee Lake, precisely because of this feature. However, the total amount of all the alterations in the depths of the Sandy Bridge microarchitecture also turned out to be quite significant. Ultimately, this design lost all of the old P6 (Pentium Pro) kinship that had been here and there in all previous Intel processors.
Speaking about the general structure, one cannot help but recall the fact that for the first time in the history of Intel CPUs, a full-fledged graphics core was built into the Sandy Bridge processor chip. This block went inside the processor after the DDR3 memory controller, shared L3 cache and PCI Express bus controller. To connect the computing cores and all other "extra-core" parts, Intel engineers implemented in Sandy Bridge a new, at that time, scalable ring bus, which is used to organize interaction between structural units in subsequent mass CPUs to this day.
If we go down to the level of the Sandy Bridge microarchitecture, then one of its key features is support for the AVX family of SIMD instructions designed to work with 256-bit vectors. By now, such instructions have become commonplace and do not seem to be something unusual, but their implementation in Sandy Bridge required the expansion of some of the computing actuators. Intel engineers wanted to make working with 256-bit data as fast as working with smaller vectors. Therefore, along with the implementation of full-fledged 256-bit executive devices, it was also necessary to increase the speed of the processor with memory. The logical execution units designed to load and save data in Sandy Bridge received twice the performance, in addition, the read throughput of the L1 cache was increased symmetrically.
It is impossible not to mention the cardinal changes made in Sandy Bridge in the operation of the branch prediction block. Thanks to optimizations in the algorithms used and increased buffer sizes, the Sandy Bridge architecture made it possible to reduce the percentage of branch mispredictions by almost half, which not only significantly affected performance, but also made it possible to further reduce the power consumption of this design.
Ultimately, from today's standpoint, Sandy Bridge processors could be called an exemplary embodiment of the "tock" phase in Intel's "tick-tock" principle. Like their predecessors, these processors continued to be based on the 32nm process technology, but the performance increase they offered turned out to be more than convincing. And it was fueled not only by an updated microarchitecture, but also by 10-15 percent increased clock speeds, as well as the introduction of a more aggressive version of Turbo Boost 2.0 technology. Given all this, it is clear why many enthusiasts still remember Sandy Bridge with the warmest words.
The Core i7-2600K became the senior offering in the Core i7 family at the time of the release of the Sandy Bridge microarchitecture. This processor received a clock frequency of 3.3 GHz with the possibility of auto-overclocking at partial load up to 3.8 GHz. However, the 32-nm representatives of Sandy Bridge were distinguished not only by relatively high clock frequencies for that time, but also by good overclocking potential. Among the Core i7-2600K, it was often possible to find specimens capable of operating at frequencies of 4.8-5.0 GHz, which was largely due to the use of a high-quality internal thermal interface in them - flux-free solder.
Nine months after the release of the Core i7-2600K, in October 2011, Intel updated the older offering in model range and offered a slightly accelerated Core i7-2700K model, the nominal frequency of which was increased to 3.5 GHz, and the maximum frequency in turbo mode to 3.9 GHz.
However, the life cycle of the Core i7-2700K turned out to be short - already in April 2012, an updated design replaced Sandy Bridge IvyBridge. Nothing special: Ivy Bridge belonged to the “tic” phase, that is, it was a translation of the old microarchitecture to new semiconductor rails. And in this regard, the progress was really serious - Ivy Bridge crystals were produced using a 22-nm process technology based on three-dimensional FinFET transistors, which were just coming into use at that time.
At the same time, the old Sandy Bridge microarchitecture at a low level remained practically untouched. Only a few minor cosmetic changes have been made that speed up the division operations in Ivy Bridge and slightly increase the efficiency of Hyper-Threading technology. True, along the way, the “non-nuclear” components were somewhat improved. The PCI Express controller received compatibility with the third version of the protocol, and the memory controller increased its capabilities and began to support high-speed overclocking DDR3 memory. But in the end, the increase in specific productivity in the transition from Sandy Bridge to Ivy Bridge amounted to no more than 3-5 percent.
The new technological process did not give serious reasons for joy. Unfortunately, the introduction of 22-nm standards did not allow to somehow fundamentally increase the clock speeds of Ivy Bridge. The older version of the Core i7-3770K received a nominal frequency of 3.5 GHz with the ability to overclock in turbo mode up to 3.9 GHz, that is, in terms of the frequency formula, it turned out to be no faster than the Core i7-2700K. Only energy efficiency has improved, but desktop users have traditionally been less concerned about this aspect.
All this, of course, can be attributed to the fact that at the tick stage no breakthroughs should occur, but in some ways Ivy Bridge turned out to be even worse than its predecessors. It's about acceleration. When introducing carriers of this design to the market, Intel decided to abandon the use of flux-free gallium soldering of a heat-spreading cover to a semiconductor chip in the final assembly of processors. Starting with Ivy Bridge, banal thermal paste began to be used to organize the internal thermal interface, and this immediately hit the maximum achievable frequencies. In terms of overclocking potential, Ivy Bridge has definitely become worse, and as a result, the transition from Sandy Bridge to Ivy Bridge has become one of the most controversial moments in the recent history of Intel's consumer processors.
Therefore, the next stage of evolution, Haswell, there were high hopes. This generation, in the "so" phase, should have seen major microarchitectural improvements, which were expected to be able to at least move the stalled progress forward. And to some extent it happened. The fourth-generation Core processors, which appeared in the summer of 2013, have indeed acquired noticeable improvements in the internal structure.
The main thing: the theoretical power of Haswell execution units, expressed in the number of micro-operations executed per clock cycle, has grown by a third compared to previous CPUs. In the new microarchitecture, not only was the rebalancing of the existing execution units, but also two additional execution ports appeared for integer operations, branch servicing and address generation. In addition, the microarchitecture received compatibility with an extended set of 256-bit AVX2 vector instructions, which, thanks to the three-operand FMA instructions, doubled the peak throughput of the architecture.
In addition to this, Intel engineers revised the capacity of internal buffers and, where necessary, increased them. The scheduler window has grown in size. In addition, the integer and real number physical register files were enlarged, which improved the processor's ability to reorder the execution order of instructions. In addition to all this, the cache memory subsystem has also changed significantly. L1 and L2 caches in Haswell received twice the bus width.
It would seem that these improvements should be enough to noticeably increase the specific performance of the new microarchitecture. But no matter how. Haswell's design problem was that it left the front end of the execution pipeline unchanged and the x86 instruction decoder retained the same performance as before. That is, the maximum decoding rate of the x86 code in a microinstruction remained at the level of 4-5 instructions per clock. And as a result, when comparing Haswell and Ivy Bridge at the same frequency and with a load that does not use the new AVX2 instructions, the performance gain turned out to be only at the level of 5-10 percent.
The image of the Haswell microarchitecture was also spoiled by the first wave of processors released on its basis. Based on the same 22nm process technology as Ivy Bridge, the new products could not offer high frequencies. For example, the older Core i7-4770K again received a base frequency of 3.5 GHz and a maximum turbo frequency of 3.9 GHz, that is, there has been no progress compared to previous generations of Core.
At the same time, with the introduction of the next technological process with 14-nm standards, Intel began to experience all sorts of difficulties, so a year later, in the summer of 2014, not the next generation of Core processors was brought to the market, but the second line of Haswell, which received the code names Haswell Refresh, or, if we talk about flagship modifications, then Devil's Canyon. As part of this update, Intel was able to noticeably increase the clock speeds of the 22nm CPU, which really breathed new life into them. An example is the new older processor Core i7-4790K, which took the mark of 4.0 GHz at the nominal frequency and received the maximum frequency, taking into account the turbo mode, at 4.4 GHz. Surprisingly, such a half-GHz acceleration was achieved without any technical process reforms, but only due to simple cosmetic changes in the processor power circuit and due to improved thermal conductive properties of the thermal paste used under the CPU cover.
However, even representatives of the Devil's Canyon family could not become especially complained about by enthusiasts. Against the background of the Sandy Bridge results, their overclocking could not be called outstanding, besides, achieving high frequencies required complex “scalping” - dismantling the processor cover and then replacing the standard thermal interface with some material with better thermal conductivity.
Due to the difficulties that plagued Intel when transferring mass production to 14-nm standards, the performance of the next, fifth generation of Core processors, Broadwell, it turned out very crumpled. The company could not decide for a long time whether it was worth launching desktop processors with this design on the market at all, since when trying to manufacture large semiconductor chips, the rejection rate exceeded acceptable values. Ultimately, the Broadwell quad-core desktop computers did appear, but, firstly, it happened only in the summer of 2015 - nine months late relative to the originally planned date, and secondly, two months after their announcement, Intel introduced the design next generation, Skylake.
Nevertheless, from the point of view of the development of the microarchitecture, Broadwell can hardly be called a secondary development. And even more than that, desktop processors of this generation used solutions that Intel had never resorted to before or since. The uniqueness of the desktop Broadwell was determined by the fact that they were penetrated by the productive integrated graphics core Iris Pro of the GT3e level. And this means not only that the processors of this family had the most powerful integrated video core at that time, but also that they were equipped with an additional 22nm Crystall Well crystal, which is an eDRAM-based L4 cache.
The point of adding a separate fast integrated memory chip to the processor is quite obvious and is due to the need for a high-performance integrated graphics core in a frame buffer with low latency and high bandwidth. However, the eDRAM memory installed in Broadwell was architecturally designed exactly as a victim cache, and CPU cores could also use it. As a result, the desktop Broadwell became the only mainstream processor of its kind with 128 MB of L4 cache. True, the volume of the L3 cache located in the processor chip suffered somewhat, which was reduced from 8 to 6 MB.
Some improvements have also been incorporated into the underlying microarchitecture. Despite the fact that Broadwell belonged to the “tick” phase, the alterations touched the input part of the execution pipeline. The out-of-order execution scheduler window was enlarged, the volume of the second-level associative address translation table was increased one and a half times, and, in addition, the entire translation scheme acquired a second miss handler, which made it possible to process two address translation operations in parallel. In sum, all innovations have increased the efficiency of out-of-order execution of commands and the prediction of complex code branches. Along the way, the mechanisms for performing multiplication operations were improved, which in Broadwell began to be processed at a much faster pace. As a result of all this, Intel was even able to claim that microarchitecture improvements increased the specific performance of Broadwell compared to Haswell by about five percent.
But despite all this, it was impossible to talk about any significant advantage of the first desktop 14-nm processors. Both the L4 cache and microarchitectural changes were only trying to compensate for Broadwell's main flaw - low clock speeds. Due to problems with the technological process, the base frequency of the older member of the family, the Core i7-5775C, was set only at 3.3 GHz, and the frequency in turbo mode did not exceed 3.7 GHz, which turned out to be worse than Devil's Canyon by as much as 700 MHz.
A similar story happened with overclocking. The limiting frequencies to which it was possible to heat desktop Broadwell without using advanced cooling methods were in the region of 4.1-4.2 GHz. Therefore, it is not surprising that consumers were skeptical about the release of Broadwell, and the processors of this family remained a strange niche solution for those who were interested in a productive integrated graphics core. The first full-fledged 14-nm chip for desktop computers, which was able to attract the attention of a wide range of users, was only the next project of the microprocessor giant - skylake.
The production of Skylake, like the previous generation processors, was carried out according to the 14-nm process technology. However, here Intel has already managed to achieve normal clock speeds and overclocking: the older desktop version of Skylake, the Core i7-6700K, received a nominal frequency of 4.0 GHz and auto-acceleration within the turbo mode up to 4.2 GHz. These are slightly lower values when compared to Devil's Canyon, but newer processors were definitely faster than their predecessors. The fact is that Skylake is "so" in the Intel nomenclature, which means significant changes in the microarchitecture.
And they really are. At first glance, there were not so many improvements in the design of Skylake, but they were all targeted and made it possible to eliminate the existing weaknesses in the microarchitecture. In short, Skylake received increased internal buffers for deeper out-of-order execution of instructions and higher cache memory bandwidth. Improvements have affected the branch prediction block and the input part of the execution pipeline. Also, the rate of execution of division instructions was increased, and the mechanisms for executing addition, multiplication, and FMA instructions were rebalanced. To top it off, developers have worked hard to improve the efficiency of Hyper-Threading technology. In sum, this allowed us to achieve about a 10 percent improvement in performance per clock compared to previous generations of processors.
In general, Skylake can be described as a fairly deep optimization of the original Core architecture, so that there are no bottlenecks in the processor design. On the one hand, by increasing the power of the decoder (from 4 to 5 micro-ops per clock) and the speed of the cache of micro-ops (from 4 to 6 micro-ops per clock), the rate of instruction decoding has increased significantly. And on the other hand, the efficiency of processing the resulting micro-operations has increased, which was facilitated by the deepening of out-of-order execution algorithms and the redistribution of the capabilities of execution ports, along with a serious revision of the execution rate of a number of ordinary, SSE and AVX commands.
For example, Haswell and Broadwell had two ports each for performing multiplications and FMA operations on real numbers, but only one port was intended for additions, which did not correspond well to real program code. In Skylake, this imbalance was eliminated and additions were already performed on two ports. In addition, the number of ports capable of working with integer vector instructions has grown from two to three. Ultimately, all this led to the fact that for almost any type of operation in Skylake there are always several alternative ports. And this means that in the microarchitecture, almost all possible causes of pipeline downtime were finally successfully eliminated.
Notable changes have also affected the caching subsystem: the throughput of the cache memory of the second and third levels has been increased. In addition, the associativity of the second-level cache was reduced, which ultimately made it possible to improve its efficiency and reduce the penalty for handling misses.
Significant changes have also taken place at a higher level. So, in Skylake, the bandwidth of the ring bus that connects all processor units has doubled. In addition, a new memory controller has settled in the CPU of this generation, which has received compatibility with DDR4 SDRAM. And in addition to this, a new DMI 3.0 bus with a doubled bandwidth has been used to connect the processor to the chipset, which made it possible to implement high-speed PCI Express 3.0 lines through the chipset as well.
However, like all previous versions of the Core architecture, Skylake was another variation on the original design. And this means that in the sixth generation of the Core microarchitecture, Intel developers continued to adhere to the tactics of phased implementation of improvements at each development cycle. In general, this is not a very impressive approach, which does not allow you to see any significant changes in performance immediately - when comparing CPUs from neighboring generations. But on the other hand, when upgrading older systems, a noticeable increase in performance is not difficult to notice at all. For example, Intel itself willingly compared Skylake with Ivy Bridge, demonstrating that in three years the speed of processors has increased by more than 30 percent.
And in fact, it was quite a serious progress, because then everything got much worse. After Skylake, any improvement in the specific performance of processor cores stopped altogether. Those processors that are currently on the market still continue to use the Skylake microarchitectural design, despite the fact that almost three years have passed since its introduction in desktop processors. The unexpected downtime came as Intel was unable to keep up with the introduction of the next version of the 10nm semiconductor process. As a result, the whole tick-tock principle crumbled, forcing the microprocessor giant to somehow get out and engage in multiple re-releases of old products under new names.
Generation processors KabyLake, which appeared on the market at the very beginning of 2017, became the first and very striking example of Intel's attempts to sell the same Skylake to customers for the second time. Close family ties between the two generations of processors were not particularly hidden. Intel honestly said that Kaby Lake is no longer a "tick" and not "so", but a simple optimization of the previous design. At the same time, the word "optimization" meant some improvements in the structure of 14-nm transistors, which opened up the possibility of increasing clock frequencies without changing the scope of the thermal package. For the modified process technology, a special term “14+ nm” was even coined. Thanks to this manufacturing technology, Kaby Lake's older mainstream desktop processor, dubbed the Core i7-7700K, was able to offer users a nominal frequency of 4.2 GHz and a turbo frequency of 4.5 GHz.
Thus, the increase in the frequencies of Kaby Lake compared to the original Skylake was about 5 percent, and that was all, which, frankly, called into question the legitimacy of referring Kaby Lake to the next generation of Core. Until that moment, each subsequent generation of processors, no matter if it belonged to the “tick” or “tock” phase, provided at least some increase in IPC. Meanwhile, in Kaby Lake there were no microarchitectural improvements at all, so it would be more logical to consider these processors as just the second Skylake stepping.
However, the new version of the 14nm process technology was still able to prove itself in some ways: the overclocking potential of Kaby Lake increased by about 200-300 MHz compared to Skylake, due to which the processors of this series were quite warmly received by enthusiasts. True, Intel continued to use thermal paste instead of solder under the processor cover, so scalping was necessary to fully overclock Kaby Lake.
Intel did not cope with the commissioning of 10-nm technology by the beginning of this year either. Therefore, at the end of last year, another type of processors built on the same Skylake microarchitecture was introduced to the market - CoffeeLake. But talking about Coffee Lake as the third guise of Skylake is not entirely correct. Last year was a period of a radical paradigm shift in the processor market. AMD returned to the “big game”, which was able to break the established traditions and create demand for mass processors with more than four cores. All of a sudden, Intel was catching up, and the release of Coffee Lake was not so much an attempt to fill the gap before the long-awaited introduction of 10nm Core processors, but a reaction to the release of AMD's six- and eight-core Ryzen processors.
As a result, Coffee Lake processors received an important structural difference from their predecessors: the number of cores in them was increased to six pieces, which happened for the first time with an Intel mass platform. However, at the same time, no changes were introduced at the microarchitecture level: Coffee Lake is essentially a six-core Skylake, assembled on the basis of exactly the same internal structure of computing cores, which are equipped with an L3 cache increased to 12 MB (according to the standard principle, 2 MB per core ) and are united by the usual ring bus.
However, despite the fact that we so easily allow ourselves to say “nothing new” about Coffee Lake, it is not entirely fair to say that there are no changes at all. Although nothing has changed in the microarchitecture again, Intel specialists had to spend a lot of effort to ensure that six-core processors could fit into a standard desktop platform. And the result came out quite convincing: six-core processors remained true to their usual thermal package and, moreover, did not slow down at all in terms of clock speeds.
In particular, the older representative of the Coffee Lake generation, the Core i7-8700K, received a base frequency of 3.7 GHz, and in turbo mode it can accelerate to 4.7 GHz. At the same time, the overclocking potential of Coffee Lake, despite its more massive semiconductor crystal, turned out to be even better than that of all its predecessors. Core i7-8700K is often brought by their ordinary owners to the five-GHz line, and such overclocking is real even without scalping and replacing the internal thermal interface. And this means that Coffee Lake, although extensive, is a significant step forward.
All this became possible only thanks to the next improvement of the 14-nm process technology. In the fourth year of its use for mass production of desktop chips, Intel has achieved really impressive results. The introduction of the third version of the 14-nm standard (“14++ nm” in the manufacturer's designation) and the re-arrangement of the semiconductor crystal made it possible to significantly improve performance in terms of each watt spent and increase the total computing power. With the introduction of six cores, Intel was perhaps able to take an even more significant step forward than any of the microarchitecture improvements that preceded it. And today, Coffee Lake looks like a very tempting option for upgrading older systems based on previous Core microarchitecture media.
| codename | Process technology | Number of cores | GPU | L3 cache, MB | Number of transistors, billion | Crystal area, mm 2 |
| Sandy Bridge | 32 nm | 4 | GT2 | 8 | 1,16 | 216 |
| Ivy Bridge | 22 nm | 4 | GT2 | 8 | 1,2 | 160 |
| Haswell | 22 nm | 4 | GT2 | 8 | 1,4 | 177 |
| Broadwell | 14 nm | 4 | GT3e | 6 | N/A | ~145 + 77 (eDRAM) |
| skylake | 14 nm | 4 | GT2 | 8 | N/A | 122 |
| Kaby Lake | 14+ nm | 4 | GT2 | 8 | N/A | 126 |
| coffee lake | 14++ nm | 6 | GT2 | 12 | N/A | 150 |
⇡ Processors and platforms: specifications
To compare the last seven generations of Core i7, we took the older representatives in the respective series - one from each design. The main characteristics of these processors are shown in the following table.
| Core i7-2700K | Core i7-3770K | Core i7-4790K | Core i7-5775C | Core i7-6700K | Core i7-7700K | Core i7-8700K | |
| codename | Sandy Bridge | Ivy Bridge | Haswell (Devil's Canyon) | Broadwell | skylake | Kaby Lake | coffee lake |
| Production technology, nm | 32 | 22 | 22 | 14 | 14 | 14+ | 14++ |
| release date | 23.10.2011 | 29.04.2012 | 2.06.2014 | 2.06.2015 | 5.08.2015 | 3.01.2017 | 5.10.2017 |
| Cores/Threads | 4/8 | 4/8 | 4/8 | 4/8 | 4/8 | 4/8 | 6/12 |
| Base frequency, GHz | 3,5 | 3,5 | 4,0 | 3,3 | 4,0 | 4,2 | 3,7 |
| Turbo Boost frequency, GHz | 3,9 | 3,9 | 4,4 | 3,7 | 4,2 | 4,5 | 4,7 |
| L3 cache, MB | 8 | 8 | 8 | 6 (+128 MB eDRAM) | 8 | 8 | 12 |
| Memory support | DDR3-1333 | DDR3-1600 | DDR3-1600 | DDR3L-1600 | DDR4-2133 | DDR4-2400 | DDR4-2666 |
| Instruction set extensions | AVX | AVX | AVX2 | AVX2 | AVX2 | AVX2 | AVX2 |
| Integrated graphics | HD 3000 (12 EU) | HD 4000 (16 EU) | HD 4600 (20 EU) | Iris Pro 6200 (48 EU) | HD 530 (24 EU) | HD 630 (24 EU) | UHD 630 (24 EU) |
| Max. graphics core frequency, GHz | 1,35 | 1,15 | 1,25 | 1,15 | 1,15 | 1,15 | 1,2 |
| PCI Express Version | 2.0 | 3.0 | 3.0 | 3.0 | 3.0 | 3.0 | 3.0 |
| PCI Express lanes | 16 | 16 | 16 | 16 | 16 | 16 | 16 |
| TDP, W | 95 | 77 | 88 | 65 | 91 | 91 | 95 |
| socket | LGA1155 | LGA1155 | LGA1150 | LGA1150 | LGA1151 | LGA1151 | LGA1151v2 |
| Official price | $332 | $332 | $339 | $366 | $339 | $339 | $359 |
It is curious that in the seven years since the release of Sandy Bridge, Intel has not been able to noticeably increase clock speeds. Despite the fact that the technological production process has changed twice and the microarchitecture has been seriously optimized twice, today's Core i7s have hardly moved forward in their operating frequency. The latest Core i7-8700K is clocked at 3.7GHz, just 6 percent faster than the 2011 Core i7-2700K.
However, such a comparison is not entirely correct, because Coffee Lake has one and a half times more computing cores. If we focus on the quad-core Core i7-7700K, then the increase in frequency still looks more convincing: this processor has accelerated relative to the 32-nm Core i7-2700K by a fairly significant 20 percent in megahertz terms. Although it is still hardly an impressive increase: in absolute terms, this is converted into an increase of 100 MHz per year.
There are no breakthroughs in other formal characteristics either. Intel continues to provide all of its processors with a 256 KB L2 individual cache per core, as well as a shared L3 cache for all cores, the size of which is determined at the rate of 2 MB per core. In other words, the main factor on which the greatest progress has been made is the number of computing cores. The development of Core started with quad-core CPUs and moved to six-core ones. Moreover, it is obvious that this is not the end yet, and in the near future we will see eight-core versions of Coffee Lake (or Whiskey Lake).
However, as you can see, Intel's pricing policy has hardly changed in seven years. Even the six-core Coffee Lake has risen in price by only six percent compared to the previous quad-core flagships. All other older processors of the Core i7 class for the mass platform have always cost consumers about $330-340.
It is curious that the biggest changes did not even occur with the processors themselves, but with their support for RAM. The throughput of dual-channel SDRAM has doubled from 21.3 to 41.6 GB/s since the release of Sandy Bridge until today. And this is another important circumstance that determines the advantage of modern systems compatible with high-speed DDR4 memory.
And in general, all these years, along with the processors, the rest of the platform has evolved. If we talk about the main milestones in the development of the platform, then, in addition to the growth in the speed of compatible memory, I would also like to note the appearance of support for the PCI Express 3.0 graphical interface. It seems that high-speed memory and a fast graphics bus, along with advances in processor frequencies and architectures, are strong reasons that modern systems become better and faster than the previous ones. Support for DDR4 SDRAM appeared in Skylake, and the transfer of the PCI Express processor bus to the third version of the protocol took place back in Ivy Bridge.
In addition, the chipsets accompanying the processors have received a noticeable development. Indeed, today's Intel 300-series chipsets can offer much more interesting features compared to the Intel Z68 and Z77 that were used in the LGA1155 motherboards for the Sandy Bridge generation processors. This is easy to see from the following table, in which we have brought together the characteristics of Intel's flagship chipsets for the mainstream platform.
| P67/Z68 | Z77 | Z87 | Z97 | Z170 | Z270 | Z370 | |
| CPU Compatibility | Sandy Bridge Ivy Bridge |
Haswell | Haswell Broadwell |
skylake Kaby Lake |
coffee lake | ||
| Interface | DMI 2.0 (2 GB/s) | DMI 3.0 (3.93 GB/s) | |||||
| PCI Express Standard | 2.0 | 3.0 | |||||
| PCI Express lanes | 8 | 20 | 24 | ||||
| PCIe M.2 support | Not | There is |
Yes, up to 3 devices | ||||
| PCI support | There is | Not | |||||
| SATA 6Gb/s | 2 | 6 | |||||
| SATA 3Gb/s | 4 | 0 | |||||
| USB 3.1 Gen2 | 0 | ||||||
| USB 3.0 | 0 | 4 | 6 | 10 | |||
| USB 2.0 | 14 | 10 | 8 | 4 | |||
In modern logic sets, the possibilities for connecting high-speed storage media have significantly developed. Most importantly, thanks to the transition of chipsets to the PCI Express 3.0 bus, high-performance NVMe drives can now be used in high-performance assemblies, which, even compared to SATA SSDs, can offer noticeably better responsiveness and more high speed reading and writing. And this alone can be a strong argument in favor of modernization.
In addition, modern system logic sets provide much richer opportunities for connecting additional devices. And it's not just about a significant increase in the number of PCI Express lanes, which ensures that the boards have several additional PCIe slots that replace conventional PCI. Along the way, today's chipsets also have native support for USB 3.0 ports, and many modern motherboards are equipped with USB ports 3.1 Gen2.
⇡ Description of test systems and testing methodology
In order to test seven fundamentally different Intel Core i7 processors released over the past seven years, we needed to assemble four platforms with processor sockets LGA1155, LGA1150, LGA1151 and LGA1151v2. The set of components that turned out to be necessary for this is described by the following list:
- Processors:
- Intel Core i7-8700K (Coffee Lake, 6 cores + HT, 3.7-4.7 GHz, 12 MB L3);
- Intel Core i7-7700K (Kaby Lake, 4 cores + HT, 4.2-4.5 GHz, 8 MB L3);
- Intel Core i7-6700K (Skylake, 4 cores, 4.0-4.2 GHz, 8 MB L3);
- Intel Core i7-5775C (Broadwell, 4 cores, 3.3-3.7 GHz, 6MB L3, 128MB L4);
- Intel Core i7-4790K (Haswell Refresh, 4 cores + HT, 4.0-4.4 GHz, 8 MB L3);
- Intel Core i7-3770K (Ivy Bridge, 4 cores + HT, 3.5-3.9 GHz, 8 MB L3);
- Intel Core i7-2700K (Sandy Bridge, 4 cores + HT, 3.5-3.9 GHz, 8 MB L3).
- CPU cooler: Noctua NH-U14S.
- Motherboards:
- ASUS ROG Maximus X Hero (LGA1151v2, Intel Z370);
- ASUS ROG Maximus IX Hero (LGA1151, Intel Z270);
- ASUS Z97-Pro (LGA1150, Intel Z97);
- ASUS P8Z77-V Deluxe (LGA1155, Intel Z77).
- Memory:
- 2 x 8 GB DDR3-2133 SDRAM, 9-11-11-31 (G.Skill TridentX F3-2133C9D-16GTX);
- 2 × 8 GB DDR4-3200 SDRAM, 16-16-16-36 (G.Skill Trident Z RGB F4-3200C16D-16GTZR).
- Video Card: NVIDIA Titan X (GP102, 12 GB/384-bit GDDR5X, 1417-1531/10000 MHz)
- Disk subsystem: Samsung 860 PRO 1TB (MZ-76P1T0BW).
- Power supply: Corsair RM850i (80 Plus Gold, 850 W).
Testing was performed on the Microsoft Windows 10 Enterprise (v1709) Build 16299 operating system using the following set of drivers:
- Intel Chipset Driver 10.1.1.45;
- Intel Management Engine Interface Driver 11.7.0.1017;
- NVIDIA GeForce 391.35 Driver.
Description of the tools used to measure computing performance:
Comprehensive benchmarks:
- Futuremark PCMark 10 Professional Edition 1.0.1275 - testing in Essentials scenarios (the usual work of an average user: launching applications, surfing the Internet, video conferencing), Productivity (office work with a text editor and spreadsheets), Digital Content Creation (creation of digital content: photo editing, non-linear video editing, rendering and visualization of 3D models). Hardware acceleration OpenCL was disabled in testing.
- Futuremark 3DMark Professional Edition 2.4.4264 - testing in the Time Spy Extreme 1.0 scene.
Applications:
- Adobe Photoshop CC 2018 - Graphics performance testing. Measured is the average execution time of a test script, which is a creatively redesigned Retouch Artists Photoshop Speed Test, which includes a typical processing of four 24-megapixel images taken by a digital camera.
- Adobe Photoshop Lightroom Classic CC 7.1 - performance testing for batch processing of a series of images in RAW format. The test scenario includes post-processing and export to JPEG at 1920 × 1080 resolution and maximum quality of two hundred 16-megapixel RAW images taken with a Fujifilm X-T1 digital camera.
- Adobe Premiere Pro CC 2018 - non-linear video editing performance testing. Measures rendering time to H.264 Blu-ray for a project containing HDV 1080p25 footage with various effects applied.
- Blender 2.79b - testing the speed of the final rendering in one of the popular free packages for creating three-dimensional graphics. The duration of building the final model from Blender Cycles Benchmark rev4 is measured.
- Corona 1.3 - testing the speed of rendering using the renderer of the same name. Measures the build speed of the standard BTR scene used to measure performance.
- Google Chrome 65.0.3325.181 (64-bit) - performance testing of Internet applications built using modern technologies. A specialized WebXPRT 3 test is used, which implements the algorithms actually used in Internet applications in HTML5 and JavaScript.
- Microsoft Visual Studio 2017 (15.1) - measuring the compilation time of a large MSVC project - a professional package for creating three-dimensional graphics Blender version 2.79b.
- Stockfish 9 - testing the speed of the popular chess engine. The speed of enumeration of options in the position "1q6/1r2k1p1/4pp1p/1P1b1P2/3Q4/7P/4B1P1/2R3K1 w" is measured;
- V-Ray 3.57.01 - testing the performance of a popular rendering system using the standard V-Ray Benchmark application;
- VeraCrypt 1.22.9 - cryptographic performance testing. The benchmark built into the program is used, which uses Kuznyechik-Serpent-Camellia triple encryption.
- WinRAR 5.50 - archiving speed testing. The time taken by the archiver to compress a directory with various files with a total volume of 1.7 GB is measured. The maximum compression ratio is used.
- x264 r2851 - testing the speed of video transcoding to H.264/AVC format. To evaluate the performance, the original [email protected] AVC video file with a bit rate of about 30 Mbps.
- x265 2.4+14 8bpp - testing the speed of video transcoding to the promising H.265/HEVC format. For performance evaluation, the same video file is used as in the x264 encoder transcoding speed test.
Games:
- Ashes of the Singularity. Resolution 1920 × 1080: DirectX 11, Quality Profile=High, MSAA=2x. Resolution 3840 × 2160: DirectX 11, Quality Profile=Extreme, MSAA=Off.
- Assassin's Creed Origins. Resolution 1920 × 1080: Graphics Quality = Very High. Resolution 3840 × 2160: Graphics Quality = Very High.
- Battlefield 1. Resolution 1920 × 1080: DirectX 11, Graphics Quality = Ultra. Resolution 3840 × 2160: DirectX 11, Graphics Quality = Ultra.
- Civilization VI. Resolution 1920×1080: DirectX 11, MSAA=4x, Performance Impact=Ultra, Memory Impact=Ultra. Resolution 3840×2160: DirectX 11, MSAA=4x, Performance Impact=Ultra, Memory Impact=Ultra.
- Far Cry 5. Resolution 1920 × 1080: Graphics Quality = Ultra, Anti-Aliasing = TAA, Motion Blur = On. Resolution 3840 × 2160: Graphics Quality = Ultra, Anti-Aliasing = TAA, Motion Blur = On.
- Grand Theft Auto V. Resolution 1920 × 1080: DirectX Version = DirectX 11, FXAA = Off, MSAA = x4, NVIDIA TXAA = Off, Population Density = Maximum, Population Variety = Maximum, Distance Scaling = Maximum, Texture Quality = Very High, Shader Quality = Very High, Shadow Quality = Very High, Reflection Quality = Ultra, Reflection MSAA = x4, Water Quality = Very High, Particles Quality = Very High, Grass Quality = Ultra, Soft Shadow = Softest, Post FX = Ultra, In-Game Depth Of Field Effects = On, Anisotropic Filtering = x16, Ambient Occlusion = High, Tessellation = Very High, Long Shadows = On, High Resolution Shadows = On, High Detail Streaming While Flying = On, Extended Distance Scaling = Maximum, Extended Shadows Distance = maximum. Resolution 3840 × 2160: DirectX Version = DirectX 11, FXAA = Off, MSAA = Off, NVIDIA TXAA = Off, Population Density = Maximum, Population Variety = Maximum, Distance Scaling = Maximum, Texture Quality = Very High, Shader Quality = Very High , Shadow Quality = Very High, Reflection Quality = Ultra, Reflection MSAA = x4, Water Quality = Very High, Particles Quality = Very High, Grass Quality = Ultra, Soft Shadow = Softest, Post FX = Ultra, In-Game Depth Of Field Effects = On, Anisotropic Filtering = x16, Ambient Occlusion = High, Tessellation = Very High, Long Shadows = On, High Resolution Shadows = On, High Detail Streaming While Flying = On, Extended Distance Scaling = Maximum, Extended Shadows Distance = Maximum.
- The Witcher 3: Wild Hunt. Resolution 1920 × 1080, Graphics Preset = Ultra, Postprocessing Preset = High. Resolution 3840 × 2160, Graphics Preset = Ultra, Postprocessing Preset = High.
- Total War: Warhammer II. Resolution 1920 × 1080: DirectX 12, Quality = Ultra. Resolution 3840 × 2160: DirectX 12, Quality = Ultra.
- Watch Dogs 2. Resolution 1920 × 1080: Field of View = 70°, Pixel Density = 1.00, Graphics Quality = Ultra, Extra Details = 100%. Resolution 3840 × 2160: Field of View = 70°, Pixel Density = 1.00, Graphics Quality = Ultra, Extra Details = 100%.
In all gaming tests, the results are the average number of frames per second, as well as the 0.01-quantile (first percentile) for fps values. The use of the 0.01-quantile instead of the minimum fps is due to the desire to clean up the results from random bursts of performance that were provoked by reasons not directly related to the operation of the main components of the platform.
⇡ Performance in complex benchmarks
The comprehensive PCMark 8 test shows the weighted average performance of systems when working in typical common applications of various kinds. And it well illustrates the progress that Intel processors have undergone at each stage of the design change. If we talk about the basic Essentials scenario, then the average speed increase per generation does not exceed the notorious 5 percent. However, the Core i7-4790K stands out from the general background, which, thanks to improvements in the microarchitecture and an increase in clock frequencies, was able to provide a good breakthrough in performance that goes beyond the average level. This breakthrough is also visible in the Productivity scenario, according to which the performance of the Core i7-4790K is comparable to the performance of older processors in the Skylake, Kaby Lake and Coffee Lake families.
The third scenario, Digital Content Creation, which combines resource-intensive creative tasks, paints a completely different picture. Here, the fresh Core i7-8700K boasts an 80% advantage over the Core i7-2700K, which can be regarded as a more than worthy result of a seven-year evolution of the microarchitecture. Of course, a significant part of this advantage is due to the increase in the number of cores, but even if we compare the performance of the quad-core Core i7-2700K and Core i7-7700K, then in this case, the speed increase reaches a solid value of 53 percent.
The synthetic game test 3DMark sticks out the advantages of the new processors even more. We use the Time Spy Extreme scenario, which has enhanced optimizations for multi-core architectures, and in it the final rating of the Core i7-8700K is almost three times higher than that of the Core i7-2700K. But a twofold advantage over Sandy Bridge is also shown by the representative of the Kaby Lake generation, which, like all predecessors, has four processing cores.
Curiously, the most successful improvement to the original microarchitecture, judging by the results, should be considered the transition from Ivy Bridge to Haswell - at this stage, according to 3D Mark, performance increased by 34 percent. However, Coffee Lake, of course, also has something to brag about, however, Intel processors of the 2017-2018 model have exactly the same microarchitecture as Skylake, and stand out solely due to extensive amplification - an increase in the number of cores.
⇡ Performance in resource intensive applications
In general, performance in applications over the past seven years of evolution of Intel processors has grown significantly. And we are not talking about five percent a year, which is usually joked about in the ranks of intel-haters. Today's Core i7s outperform their predecessors from 2011 by more than twice. Of course, the transition to a six-core system played a big role here, but microarchitectural improvements and an increase in clock frequency also made a significant contribution. Haswell turned out to be the most effective design in this regard. It significantly increased the frequency, and also introduced support for AVX2 instructions, which gradually became stronger in applications for working with multimedia content and in rendering tasks.
It is worth noting that in some cases, upgrading processors in systems that solve professional tasks can provide a truly breakthrough improvement in performance. In particular, a threefold increase in performance when moving from Sandy Bridge to Coffee Lake can be obtained when transcoding video with modern encoders, as well as in the final rendering using V-Ray. A good increase is also observed in non-linear video editing in Adobe Premiere Pro. However, even if your field of activity is not directly related to solving such problems, in any of the applications we tested, the increase was at least 50 percent.
Rendering:
Photo processing:
Video processing:
Video transcoding:
Compilation:
Archiving:
Encryption:
Chess:
Internet surfing:
In order to better visualize how the power of Intel processors changed during the last seven generations of microarchitecture, we have compiled a special table. It shows the percentages of average performance gains in resource-intensive applications obtained when switching from one flagship processor of the Core i7 series to another.
It's easy to see that Coffee Lake has been the most significant design update for mainstream Intel processors. The 1.5-fold increase in the number of cores gives performance a significant boost, thanks to which, when switching to the Core i7-8700K, even from processors of recent generations, you can get a very noticeable acceleration. Comparable performance growth since 2011 at Intel happened only once - with the introduction of the Haswell processor design (in an improved form of Devil's Canyon). Then it was due to serious changes in the microarchitecture, which were carried out simultaneously with a noticeable increase in clock frequency.
⇡ Game performance
The fact that the performance of Intel processors is steadily increasing is well seen by users of resource-intensive applications. However, among the players there is a different opinion. Still, games, even the most modern ones, do not use vector instruction sets, are poorly optimized for multithreading, and generally scale their performance at a much more restrained pace due to the fact that, in addition to computing resources, they also need graphics. So does it make sense to upgrade processors for those who use computers primarily for games?
Let's try to answer this question. To begin with, let's look at the results of tests in FullHD resolution, where processor dependence is more pronounced, since the graphics card is not a serious limitation on fps and allows processors to demonstrate what they are capable of more clearly.
The situation in different games similar, so let's look at the average relative gaming performance in FullHD. These are summarized in the following table, which shows the gains that can be gained from switching from one flagship Core i7 processor to another.
Indeed, gaming performance with the release of new generations of processors scales much weaker than in applications. If it was possible to say that over the past seven years, Intel processors have accelerated by about half, then in terms of gaming applications, the Core i7-8700K is only 36 percent faster than Sandy Bridge. And if we compare the latest Core i7 with some Haswell, then the advantage of the Core i7-8700K will be only at the level of 11 percent, despite the one and a half times increase in the number of computing cores. It seems that players who do not want to update their LGA1155 systems are right in some way. Such an increase as creative workers - content creators, they will not even get close.
The difference in the results is quite weak, in summary the situation is as follows.
It turns out that 4K players - owners of Core i7-4790K and later processors - have nothing to worry about now. Until a new generation of graphics accelerators comes to the market, such CPUs will not be a bottleneck when gaming at ultra-high resolutions, and performance is completely limited by the video card. A processor upgrade may only make sense for systems equipped with Sandy Bridge or Ivy Bridge retro processors, but even in this case, the frame rate increase will not exceed 6-9 percent.
⇡ Energy consumption
It would be interesting to supplement the performance tests with the results of energy consumption measurements. Over the past seven years, Intel has changed technology standards twice and its stated thermal envelope six times. In addition, Haswell and Broadwell processors, unlike the others, used a fundamentally different power supply scheme and were equipped with an integrated voltage converter. All this, of course, in one way or another influenced real consumption.
The Corsair RM850i digital power supply used by us in the test system allows you to control the consumed and issued electrical power which we use for measurements. The graph below shows the total consumption of systems (without a monitor), measured "after" the power supply, which is the sum of the power consumption of all components involved in the system. The efficiency of the power supply itself in this case is not taken into account.
In the idle state, the situation changed dramatically with the introduction of the Broadwell design, when Intel switched to using a 14nm process and introduced deeper power-saving modes into circulation.
When rendering, it turns out that the increase in the number of computing cores in Coffee Lake has noticeably affected its power consumption. This processor has become significantly more gluttonous than its predecessors. The most economical representatives of the Core i7 series are the carriers of the Broadwell and Ivy Bridge microarchitectures, which is quite consistent with the TDP characteristics that Intel declares for them.
Interestingly, at the highest loads, the consumption of the Core i7-8700K is similar to the consumption of the Devil's Canyon processor and no longer seems so prohibitive. But in general, the energy appetites of Core i7 processors of different generations differ very noticeably, and more modern CPU models do not always become more economical than their predecessors. A big step in improving the characteristics of consumption and heat dissipation was made in the Ivy Bridge generation, in addition, Kaby Lake is not bad in this regard. However, now it seems that improving the energy efficiency of flagship desktop processors has ceased to be an important task for Intel.
Addendum: performance at the same clock speed
Comparative testing of mass Core i7 processors of different generations can be interesting even if all participants are brought to a single clock frequency. Often, the performance of newer representatives is higher due to the fact that Intel increases the clock speeds in them. Tests at the same frequency make it possible to isolate an extensive frequency component from the overall result, which depends only indirectly on the microarchitecture, and focus on “intensification” issues.
Performance measured regardless of clock speeds may also be of interest to enthusiasts who operate the CPU outside the nominal modes, at frequencies that are very different from the standard values. Based on these considerations, we decided to add an additional discipline to the practical comparison - tests of all processors at the same frequency of 4.5 GHz. This frequency value was chosen based on the fact that it is not difficult to overclock to it almost any of the Intel processors of recent years. Only the representative of the Broadwell generation had to be excluded from such a comparison, since the overclocking potential of the Core i7-5775C is extremely limited and one can not even dream of taking the 4.5 GHz frequency. The remaining six processors went through another test cycle.
Even if we ignore the fact that the frequencies of Intel processors are growing at least slowly, Core i7 are getting better with each new generation only due to structural changes and optimizations in the microarchitecture. Judging by the speed in applications for creating and processing digital content, we can conclude that the average increase in specific productivity at each stage is about 15 percent.
However, in games where optimization program code under modern microarchitectures, it lags far behind, the situation with the increase in performance is somewhat different:
The games clearly show how the development of Intel microarchitectures stopped at the Skylake generation, and even an increase in the number of computing cores in Coffee Lake does little to increase gaming performance.
Of course, the lack of growth in specific gaming performance does not mean that the newer Core i7 is not interesting for gamers. In the end, don't forget that the above results are for frame rates for CPUs running at the same clock speed, and newer processors not only have higher nominal frequencies, but also overclock much better than older ones. This means that overclockers may be interested in switching to Coffee Lake, not because of its microarchitecture, which has remained unchanged since Skylake, and not because of six cores, which give a minimal increase in speed in games, but for another reason. - thanks to overclocking capabilities. In particular, reaching the 5 GHz limit for Coffee Lake is quite a feasible task, which cannot be said about its predecessors.
⇡ Conclusion
It so happened that Intel is usually scolded for the strategy chosen in recent years for the measured and unhurried introduction of improvements to the core Core architecture, which gives a not too noticeable increase in performance when switching to each next generation of CPU. but detailed testing shows that, in general, real productivity is growing at a not so sluggish pace. You just need to take into account two points. Firstly, many improvements added to new processors do not reveal themselves immediately, but only after some time, when the software acquires appropriate optimizations. Secondly, even if small, but systematic improvement in productivity that occurs every year, the sum gives a very significant effect, if we consider the situation in the context of longer time periods.
In confirmation, it is enough to cite one very revealing fact: the latest Core i7-8700K is more than twice as fast as its predecessor from 2011. And even if we compare the novelty with Core processor i7-4790K, which was released in 2014, it turns out that in four years the performance has managed to grow at least one and a half times.
However, you need to understand that the above growth rates relate to resource-intensive applications for creating and processing digital content. And this is where the dividing line lies: professional users who use their systems for work receive far greater dividends from the improvement of processors than those whose computer is purely for entertainment. And while for content creators, frequent upgrades of platforms and processors are more than a meaningful step to increase productivity, for gamers, the conversation turns out to be completely different.
Gaming applications are a very conservative industry that reacts extremely slowly to any changes in processor architecture. In addition, gaming performance is more dependent on the performance of graphics cards than processors. Therefore, it turns out that users of gaming systems see the development of Intel CPUs in recent years in a completely different way. Where "professionals" state a twofold increase in performance, players get at best only a 35% increase in the number of fps. And this means that in the pursuit of new generations of Intel CPUs, there is practically no point for them. Even the older processors of the Sandy Bridge and Ivy Bridge series have enough power to unlock the potential of a GeForce GTX 1080 Ti-level graphics card.
Thus, for now, players in new processors may be attracted not so much by performance growth as by new features. They may be some additional features that appear in fresh platforms, such as support for high-speed drives. Or the best overclocking potential, the limits of which, despite Intel's problems with the development of new technological processes, are still gradually moving to more distant boundaries. However, in order for players to receive a clear and understandable signal to upgrade, first of all, there must be a noticeable increase in the performance of gaming GPUs. Until then, even the owners of seven-year-old Intel CPUs will continue to feel completely not deprived of processor performance.
Nevertheless, this situation is quite capable of changing the processors of the Coffee Lake generation. The increase in the number of computing cores that has taken place in them (up to six, and up to eight in the future) carries a powerful emotional charge. Due to this, the Core i7-8700K seems to be a very successful upgrade for almost any PC user, because many people think that six-core due to the potential inherent in them will be able to remain a relevant option for a longer period. Whether this is true is hard to say now. But, summing up all the above, we can confirm that upgrading the system to Coffee Lake in any case makes much more sense than the upgrade options that the microprocessor giant has offered so far.
