two-in-one processor / Processors and memory

two-in-one processor / Processors and memory


Today’s Intel® Core ™ desktop processors have traditionally been divided into two groups: “regular” and “enthusiast”. And although real buyers are not always guided by this positioning, it describes well what can be expected from this or that chip.

The enthusiast processors, which are marked with the letter K in the name, are overclocking models with unlocked multipliers, higher clock speeds and higher thermal package limits. Obviously, these proposals are aimed at more experienced users, for whom it is not a problem to configure many different parameters in the system and choose the right set of components, including the right motherboard, a decent power supply and a highly efficient cooling system.

Processors, which Intel is targeting ordinary users, cost less, but they are slower and more economical. These processors are designed in such a way as to cause a minimum of problems for end users: they usually do not require any special system configuration and work with a plug and play principle. In addition to the fact that such CPUs do not give access to changing the multiplier, they also have more restrained frequencies and low consumption, which allows them to be equipped with relatively simple cooling systems and not to bother with choosing boards and other strapping.

Surprisingly, processors belonging to different groups often have almost the same model numbers, although it is obvious that their performance and other consumer qualities can vary greatly. This was especially evident in the Comet Lake generation. In it, overclocking CPU models noticeably increased their energy appetites and received a typical heat dissipation characteristic of 125 watts. Processors for conventional systems continued to exist within the typical for this class thermal package of 65 W, that is, their consumption and heat dissipation turned out to be almost half that of their overclocking counterparts. And still, along with the Core i9-10900K, there is a Core i9-10900, a Core i7-10700 is delivered next to the Core i7-10700K, and instead of a Core i5-10600K, you can purchase a Core i5-10600.

The coexistence of, on the one hand, similar, but on the other – completely different models prompted us to think about the need to conduct special testing, which could show how great the difference between them is and how one should generally relate to such a filling of the lineup of Core 10 processors generations. Much of this interest is fueled by the fact that “simple” models do not seem useless to advanced users either. On the one hand, they are 10-20% cheaper, but at the same time offer exactly the same number of cores and threads as the flagship overclocking models. On the other hand, their base clock frequencies are lower by a whole gigahertz, which will surely make a frightening impression on an inexperienced user.

But even if significant differences in nominal frequencies have real weight, this is not a reason to deliberately put an end to “non-K” models. Firstly, the efficiency indicated in the specifications can carry a completely positive message for a noticeable part of users. Secondly, if you look at the maximum turbo frequencies of “ordinary” models, the situation with the base frequencies ceases to seem so terrible, because according to this characteristic, the advantage of overclocking modifications with a TDP of 125 W is only 100-300 MHz.

In this article, we will try to give a clear answer to the question of what those who choose a processor for their system without the “K” index in the name will have to face: either with a multi-core CPU with not the most impressive performance, where energy efficiency is paramount, then whether with a cheaper and only slightly slowed down version of the flagship, or even with something in between. As part of a partnership project, Intel provided us with a Core i7-10700 processor for a detailed study – a promising eight-core processor that attracts with an official price of $ 323.

⇡ # Core i7-10700 in detail

Speaking about the Core i7-10700, we assume that you are well acquainted with Intel’s flagship eight-core processor for the LGA1200 platform – the Core i7-10700K processor. If this is not the case, we strongly recommend that you refer to the corresponding review on our website.

The fact is that if we discard all overclocking possibilities, then the Core i7-10700 without the K index turns out to be a very close relative of its overclocking brother. At least both of them are based on the same silicon, which means that the differences between them really exist only at the level of clock frequencies, as well as thermal and energy characteristics. In general, the Core i7-10700 is an eight-core processor based on the Skylake microarchitecture with support for Hyper-Threading technology, and has an L3 cache of 16 MB. It is built on a semiconductor Q0 stepping crystal, which is manufactured using a 14-nm process technology (with some number of plus signs) and initially has 10 cores, a pair of which is hardware locked and inoperative. At the same time, the key property of the Core i7-10700 is economy: its thermal package is twice as strict as that of the Core i701700K, and the target heat dissipation is set at 65 W.

Speaking of the Core i7-10700K, we drew close parallels between it and the last generation eight-core processor, the Core i9-9900K. Following this logic further, I would like to say that the Core i7-10700 is the ideological successor to the 65-watt Core i9-9900, and this is almost true. However, there is one oddity: while the overclocking Core i7-10700K slightly outperforms the Core i9-9900K in operating frequencies, the regular Core i7-10700, on the contrary, is slightly inferior to its predecessor for the LGA1151v2 platform.

How significant these differences are can be estimated from the following table:

Core i7-10700K Core i7-10700 Core i9-9900K Core i9-9900
Platform LGA1200 LGA1200 LGA1151v2 LGA1151v2
Technological process, mm fourteen fourteen fourteen fourteen
Kernels / threads 8/16 8/16 8/16 8/16
Rated frequency, GHz 3.8 2.9 3.6 3.1
Max. turbo, 1 core, GHz 5.1 4.8 5.0 5.0
Max. turbo, all cores, GHz 4.7 4.6 4.7 4.6
L3 cache, MB sixteen sixteen sixteen sixteen
TDP, W 125 65 95 65
Memory DDR4-2933 DDR4-2933 DDR4-2666 DDR4-2666
PCIe Lines 16 × Gen3 16 × Gen3 16 × Gen3 16 × Gen3
Integrated graphics there is there is there is there is
Price $ 374 $ 323 $ 488 $ 423

The logic is that if in the last generation the Core i9-9900K and Core i9-9900 processors were as close as possible to each other according to the formal frequency formula, and the difference between them was in the value of TDP, then in the Comet Lake generation, Intel moved away the Core i7-10700K and Core i7-10700 from each other, in particular, and in terms of frequencies. It is for this reason that the Core i7-10700 looks a little less interesting compared to the Core i9-9900.

However, one should not forget about the nuances. The maximum frequencies that the processors are allowed to develop in turbo mode for the Core i7-10700 and Core i9-9900 are configured with a gap of 100-200 MHz.

Maximum frequency in turbo mode, GHz Base, GHz
1 core 2 cores 3 cores 4 cores 5 cores 6 cores 7 cores 8 cores
Core i7-10700K 5.1 5.1 5.0 4.8 4.8 4.7 4.7 4.7 3.8
Core i7-10700 4.8 4.8 4.7 4.7 4.6 4.6 4.6 4.6 2.9
Core i9-9900K 5.0 5.0 4.9 4.8 4.8 4.7 4.7 4.7 3.6
Core i9-9900 5.0 5.0 4.9 4.8 4.8 4.7 4.7 4.6 3.1

At the same time, although the same thermal package is declared for the Core i7-10700 and Core i9-9900, the newer generation processor has a much more liberal PL2 limit (limiting consumption under short-term loads), which allows it to go beyond the 65-watt limit significantly further than a representative of the Coffee Lake generation.

PL1, W PL2, W Tau, seconds
Core i7-10700K 125 229 56
Core i7-10700 65 224 28
Core i9-9900K 95 119 28
Core i9-9900 65 81 28

In fact, for short-term loads of less than 28 seconds, the Core i7-10700 is able to run smoothly at its maximum turbo frequency. The PL2 limit of 224 W, if it is able to somehow affect the frequency formula, then this influence will be extremely insignificant. Some noticeable slowdown of the Core i7-10700 can occur only when a serious load on the processor will be long and continuous.

How it looks in practice can be seen in the following graph. It showed the real frequencies and power consumption of the Core i7-10700 during the rendering test in Cinebench R20.

The first 20 seconds of the test, the Core i7-10700 runs smoothly at a maximum frequency of 4.6-4.7 GHz. In this case, its power consumption reaches about 150 W, but this is within the framework of the rules, because it does not go beyond the PL2 boundary. The processor reaches 65-watt consumption only after some time, and in order to meet the specification and not go beyond the PL1 limit, its frequency is reset to 3.6-3.7 GHz, that is, by 20-25%.

It is very important that all the consumption limits that are described by the specification for the Core i7-10700 are in fact optional. Although the processor in question is not an overclocker, and its multiplier cannot be increased, it is not forbidden to change the consumption limits manually for it – the corresponding settings are available in the BIOS of almost any motherboard. As a result, users who do not need to limit power consumption can easily configure the Core i7-10700 so that it is guided solely by the frequency formula for turbo mode, while ignoring all passport restrictions regarding heat and power consumption.

It turns out that the Core i7-10700 is the very case when we can talk about a two-in-one processor: either energy efficient, if you follow all the specifications exactly and observe the PL1 and PL2 limits, or rather high-spirited, if these limits are turned off. How these two entities are combined, we have displayed in the following graph, which shows the real frequencies of the Core i7-10700 in the multi-threaded rendering test Cinebench R20: in one case – provided that the processor has a specified 65-watt limit, and in the other – in condition with the limits PL1 and PL2 removed.

Naturally, all this affects the performance. Working without limits, the Core i7-10700 scores about 4890 points in Cinebench R20, but in the energy-efficient 65-watt state this figure drops to 3880 points.

All this leads us to the conclusion that the Core i7-10700 is not at all like the Core i7-10700K, which, when the PL1 and PL2 power consumption limits are activated, loses very little in performance. It’s impossible to say the same about the Core i7-10700: the inclusion of power consumption limits turns it into a completely different one in terms of the practical characteristics of the CPU. Performance, judged by the Cinebench R20 score, is reduced by about 20%.

The comparison of the actual consumption of the Core i7-10700 in the two configurations is shown in the following graph, which was compiled based on measurements in Cinebench R20 with various restrictions on the number of active threads. This graph clearly shows that the 65 W Core i7-10700 allocated for it only fits with four-thread rendering.

And here is how the temperatures of this processor look in the limited 65-watt mode and when the power consumption restrictions are lifted. To remove heat in this test, we used a Noctua NH-U14S cooler.

The conclusion from these graphs is quite obvious: if the consumption limits remain in effect, then a very economical and rather cold processor is obtained from the Core i7-10700, despite the fact that we are talking about a 14nm eight-core. Therefore, you shouldn’t be surprised at the complete cooler included in the box with this processor. It has a relatively small size and is devoid of any heat pipes. But it will certainly fit even into compact Mini-ITX systems.

It is curious that in comparison with the cooling systems with which Intel has equipped its processors of previous generations, the cooler for the Core i7-10700 has noticeable external differences: it is anodized black, which makes it look much more modern and noble.

⇡ # Description of the test system and testing methodology

The Core i7-10700K and Core i7-10700 processors are priced at a whopping $ 50. This amount can be a strong argument in favor of a cheaper option. Especially if we take into account its not so significant differences from its older brother in frequency, provided that the consumption limits are removed.

That is why we had two questions, the answer to which this testing was devoted to. Question one: how far does the Core i7-10700 lag behind the Core i7-10700K, if all the PL1 and PL2 limits are canceled for the former. And the second: what proportion of performance will be lost by those users who want from the Core i7-10700 not chasing its older brother, namely economy. After all, in the end, a 65-watt eight-core is, in particular, a very attractive option for use in systems with small dimensions.

To answer these questions, we conducted a benchmarking test using the following components:

  • processors:
    • Intel Core i7-10700K (Comet Lake, 8 cores + HT, 3.8-5.1 GHz, 16 MB L3);
    • Intel Core i7-10700 (Comet Lake, 8 cores + HT, 2.9-4.8 GHz, 16 MB L3);
  • processor cooler: Noctua NH-U14S;
  • motherboard: ASUS ROG Maximus XII Hero (Wi-Fi) (LGA1200, Intel Z490);
  • memory: 2 × 16 GB DDR4-3600 SDRAM, 16-19-19-39 (G.Skill TridentZ Neo F4-3600C16D-16GTZNC);
  • video card: NVIDIA GeForce RTX 2080 Ti (TU102, 1350/14000 MHz, 11 GB GDDR6 352-bit);
  • disk subsystem: Samsung 970 EVO Plus 2TB (MZ-V7S2T0BW);
  • PSU: Thermaltake Toughpower DPS G RGB 1000W Titanium (80 Plus Titanium, 1000W).

We tested the Core i7-10700K processor with the settings that motherboards set to it by default – that is, immediately with the canceled consumption limits. At the same time, the Core i7-10700 was tested in two states – an economical 65-watt state and when the PL1 and PL2 limits were removed.

Testing was performed on Microsoft Windows 10 Pro (v2004) Build 19041.208 using the following set of drivers:

  • Intel Chipset Driver 10.1.18295.8201;
  • NVIDIA GeForce 451.67 Driver.

Description of the tools used to measure computational performance:

Applications:

  • 7-zip 19.00 – testing the speed of archiving. The time taken by the archiver to compress a directory with various files with a total volume of 3.1 GB is measured. The LZMA2 algorithm and the maximum compression ratio are used.
  • Adobe Photoshop 2020 21.2.1 – performance testing for graphics processing. This measures the average execution time of the Puget Systems Adobe Photoshop CC Benchmark 18.10 test script, which simulates typical processing of a digital camera image.
  • Adobe Photoshop Lightroom Classic 9.3 – performance testing in 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 16MP RAW images taken with a Fujifilm X-T1 digital camera.
  • Adobe Premiere Pro 2020 14.3.1 – performance testing for non-linear video editing. This measures the render time to YouTube 4K of a project containing HDV 2160p30 footage with various effects applied.
  • Blender 2.83.3 – testing the speed of the final rendering in one of the popular free packages for creating three-dimensional graphics. The time taken to build the final pavillon_barcelona_v1.2 model from Blender Benchmark is measured.
  • Topaz Video Enhance AI v1.3.8 – performance testing in an AI-based program to improve video detail. The test uses the original video at 640 × 360, which is doubled using the Theia-Detail model: UE, P.
  • V-Ray 4.10.03 – testing the performance of the popular rendering system using the standard V-Ray Benchmark Next application;
  • x265 3.2 + 9 10bpp – testing the speed of video transcoding to H.265 / HEVC format. For performance evaluation, the original 2160p @ 24FPS AVC video file is used, which has a bitrate of about 42 Mbps.

Games:

  • Assassin’s Creed Odyssey. 1920 × 1080 resolution: Graphics Quality = Ultra High. 2560 × 1440 resolution: Graphics Quality = Ultra High.
  • Civilization VI: Gathering Storm. 1920 × 1080 resolution: DirectX 12, MSAA = 4x, Performance Impact = Ultra, Memory Impact = Ultra. 2560 × 1440 resolution: DirectX 12, MSAA = 4x, Performance Impact = Ultra, Memory Impact = Ultra.
  • Far Cry 5. Resolution 1920 × 1080: Graphics Quality = Ultra, HD Textures = On, Anti-Aliasing = TAA, Motion Blur = On. 2560 × 1440 resolution: Graphics Quality = Ultra, Anti-Aliasing = Off, Motion Blur = On.
  • Hitman 2. Resolution 1920 × 1080: DirectX 12, Super Sampling = 1.0, Level of Detail = Ultra, Anti-Aliasing = FXAA, Texture Quality = High, Texture Filter = Anisotropic 16x, SSAO = On, Shadow Maps = Ultra, Shadow Resolution = High. 2560 × 1440 resolution: DirectX 12, Super Sampling = 1.0, Level of Detail = Ultra, Anti-Aliasing = FXAA, Texture Quality = High, Texture Filter = Anisotropic 16x, SSAO = On, Shadow Maps = Ultra, Shadow Resolution = High.
  • Shadow of the Tomb Raider. 1920 × 1080 resolution: DirectX12, Preset = Highest, Anti-Aliasing = TAA. 2560 × 1440 resolution: DirectX12, Preset = Highest, Anti-Aliasing = Off.
  • Total War: Three Kingdoms. 1920 × 1080 resolution: DirectX 12, Quality = Ultra, Unit Size = Extreme. 2560 × 1440 resolution: DirectX 12, Quality = Ultra, Unit Size = Extreme.
  • World War Z. Resolution 1920 × 1080: DirectX11, Visual Quality Preset = Ultra. 2560 × 1440 resolution: DirectX11, Visual Quality Preset = Ultra.

In all gaming tests, the results are the average number of frames per second, as well as the 0.01-quantile (first percentile) for the FPS values. The use of 0.01-quantile instead of the minimum FPS indicators is due to the desire to clear the results from random performance spikes that were provoked by reasons not directly related to the operation of the main platform components.

⇡ # Performance tests

⇡ # Performance in complex tests

The PCMark 10 test allows you to assess the performance situation when using the system in some common scenarios that ordinary users have to perform in their daily work. And as the results show, in this case, the difference between the Core i7-10700K and the Core i7-10700 is almost not felt. Moreover, if the Core i7-10700 is locked within the 65-watt limit, it will also show almost the same performance. И это значит, что, пока речь не идёт о какой-то ресурсоёмкой вычислительной нагрузке, все восьмиядерники Intel приблизительно эквивалентны.

В 3DMark Time Spy ситуация иная. В этом бенчмарке создаётся серьёзная процессорная многопоточная нагрузка, и ограничения по энергопотреблению начинают играть заметную роль. В процессорном подтесте 65-ваттный вариант Core i7-10700 отстаёт от себя же безлимитного очень заметно. Но при этом отрадно, что если не фокусироваться на экономичности, то разрыв в производительности Core i7-10700 и Core i7-10700K сокращается до минимума.

⇡#Производительность в приложениях

Нет ничего удивительного в том, что 65-ваттное ограничение, действующее для Core i7-10700, в ресурсоёмких задачах выливается в серьёзное снижение производительности. Как было показано в первой части статьи, частота процессора из-за такого ограничения может снижаться вплоть до 3,6-3,7 ГГц. Поэтому в ряде случаев скорость работы экономичной версии Core i7-10700 может быть ниже, чем у Core i7-10700K, на существенные 25-30 %. Однако справедливости ради стоит сказать, что такое случается далеко не всегда. Например, при архивации или обработке изображений разрыв в результатах энергоэффективного и оверклокерского процессоров не такой уж и заметный.

Кроме того, не нужно забывать, что у Core i7-10700, помимо энергоэффективного режима, есть ещё и «режим максимального турбо». В нём никакие ограничения по потреблению не действуют, и производительность подтягивается вплотную к уровню Core i7-10700K. Разница в результатах сокращается до 2-3 %, что служит отличной иллюстрацией гибкости Core i7-10700, хотя из спецификации впечатление о нём складывается как о восьмиядернике, в котором экономичность поставлена выше быстродействия.

Рендеринг:

Обработка фото:

Работа с видео:

Перекодирование видео:

Архивация:

⇡#Производительность в играх

Игровые бенчмарки оказались наиболее любопытной частью в тестировании. Как выяснилось, при нагрузке такого рода практически всё равно, какой вариант Core i7-10700 используется в системе – ограниченный строгими рамками теплового пакета или же работающий в безлимитном режиме. Но и это закономерно: в современных играх основная часть нагрузки лежит на видеокарте, процессор же хотя и занят расчётами, они не столь интенсивны. Игры не могут загрузить его настолько, чтобы он долговременно потреблял бы более 65 Вт. Собственно, об этом говорится уже давно: процессоры с восемью ядрами для игр избыточны, поэтому загрузка такого CPU в играх практически никогда не доходит до 100 %. Кроме того, в современных играх в первую очередь используются целочисленные скалярные операции, а они не порождают существенного энергопотребления.

В результате, даже если говорить о производительности процессоров в разрешении Full HD, максимальная разница в частоте кадров, обеспечиваемой 65-ваттным Core i7-10700 и безлимитным Core i7-10700K, составляет лишь малозначительные 3 %. А это, в свою очередь означает, что приобретать для игровых систем именно процессор с буквой K в модельном номере нет особого смысла.

⇡#Энергопотребление

В тестах энергопотребления в первую очередь мы хотели посмотреть, насколько экономичной окажется платформа на базе Core i7-10700 в сборе. И тесты вполне удовлетворили наше любопытство: система с таким процессором в энергоэффективном, 65-ваттном режиме при максимальной процессорной нагрузке потребляет не более 120 Вт, что более чем вдвое меньше потребления той же платформы, но с процессором, который не сдерживает себя какими-то рамками энергетических аппетитов.

Попутно выяснилась и ещё одна любопытная деталь. В номинальном режиме, с отключёнными лимитами PL1 и PL2, оверклокерский процессор Core i7-10700K оказался экономичнее своего более слабого собрата. При максимальной AVX-нагрузке в Prime95 30.3 разница составила заметные 40 Вт. (В скобках напомним, что приведённые на диаграммах числа – это суммарное потребление тестовых систем, измеренное на выходе из блока питания).

Таким образом, не стоит думать, что для 65-ваттных процессоров Intel выбирает какие-то особые полупроводниковые кристаллы, способные работать при меньших напряжениях и потому имеющие лучшие показатели экономичности. Наш образец Core i7-10700 полностью опровергает это предположение. Скорее даже наоборот, лучший кремний попадает в Core i7-10700K, а пониженное энергопотребление Core i7-10700 обеспечивается простой регулировкой тактовой частоты.

⇡#Выводы

Главный вопрос, на который нам хотелось получить ответ в проведённом исследовании Core i7-10700: насколько сильно он может упасть в глазах энтузиастов из-за того, что его паспортное тепловыделение ограничено величиной 65 Вт? Ответ на него порадует многих: Core i7-10700 можно сконфигурировать так, чтобы из него получился вполне достойный восьмиядерник с хорошим уровнем производительности и, что немаловажно, существенно меньшей ценой, если сравнивать со стоимостью Core i7-10700K. А если в контексте цены вспомнить о версии Core i7-10700 без графического ядра, которая проходит под названием Core i7-10700F, то особо нужно будет сказать, что это – самый дешёвый вариант среди всех восьмиядерных процессоров последнего поколения на российском рынке. Иными словами, сторониться 65-ваттных восьмиядерных процессоров Intel определённо не нужно.

Секрет извлечения из Core i7-10700 производительности, близкой к уровню Core i7-10700K, довольно прост. Intel не стала ограничивать возможности управления пределами потребления, поэтому обозначенную для Core i7-10700 в спецификации величину TDP в 65 Вт можно попросту отменить. Это позволяет вывести рабочие частоты такого процессора на уровень, близкий к частотам Core i7-10700K. Разница при этом составит лишь 100-300 МГц, что означает очень небольшой разрыв в быстродействии, в особенности если говорить про геймерский аспект – частоту кадров в играх. Единственное, не нужно забывать при этом о том, что, несмотря на неоверклокерскую сущность Core i7-10700, использовать его настоятельно рекомендуется в платах на базе набора логики Z490, поскольку только с ними его можно укомплектовать быстрой оперативной памятью, а не DDR4-2933.

Что же касается экономичности Core i7-10700, то и эта его сторона выглядит очень неплохо. Процессор действительно можно вписать в 65-ваттные рамки, для соответствия которым он будет динамически сбрасывать частоту при росте нагрузки, подстраиваясь под установленный предел. Такой сценарий интересен далеко не всегда, но в ряде случаев он всё-таки пригодится: благодаря ему Core i7-10700 становится восьмиядерным процессором, который без каких-либо проблем может вписаться в холодные, тихие и экономичные системы. И хотя потери в производительности относительно Core i7-10700K в пиковых случаях ресурсоёмкой вычислительной нагрузки (вроде рендеринга или перекодирования видео) при этом могут достигать 25-30 %, в большинстве распространённых задач и, самое главное, в играх отличие в быстродействии оказывается не слишком значительным. Например, по данным наших тестов, отставание 65-ваттного Core i7-10700 от Core i7-10700K по частоте кадров в играх в разрешении Full HD не превысило 3 %, что указывает на возможность использовать Core i7-10700 в компактных геймерских компьютерах, например формата Mini-ITX.

Таким образом, главный вывод, который можно сделать по итогам, стоит сформулировать так: среди восьмиядерных процессоров Intel поколения Comet Lake интерес для энтузиастов представляет не только оверклокерский Core i7-10700K, но и также его младший собрат без литеры К в названии – Core i7-10700. Он дешевле, но при этом удивительно гибок. С одной стороны, это экономичный CPU, хорошо подходящий для систем небольшого формата, а с другой — при отключении пределов потребления он удивительным образом превращается в некое подобие Core i7-10700K, разве только без оверклокерских возможностей. Иными словами, метафора «два в одном» – это как раз про Core i7-10700.

Если Вы заметили ошибку — выделите ее мышью и нажмите CTRL+ENTER.

Leave a Comment