Intel Core i series

<<   Intel Core i
Logos of the Intel Core i family based on Haswell
Production:	since 2008
Producer:	Intel
Processor clock:	1.06 GHz to 5.0 GHz
L3 cache size:	3 MiB to 24.75 MiB
Manufacturing :	45, 32, 22, 14 and 10 nm
Instruction set :	x86 (16 bit), x86-32, x86-64
Microarchitecture :	Nehalem , Sandy Bridge , Ivy Bridge , Haswell , Broadwell , Skylake , Kaby Lake , Coffee Lake
Base: LGA 1156 (H1) LGA 1155 (H2) , BGA 1023 LGA 1150 (H3) , BGA 1168 , BGA 1364 LGA 1151 LGA 1366 , BGA 1288 LGA 2011 LGA 2011-v3 LGA 2066 PGA 988A (G1) , PGA 988B (G2) , BGA 1224
Names of the processor cores: Lynnfield, Clarksfield, Bloomfield Clarkdale, Arrandale, Gulftown Sandy Bridge, Sandy Bridge-M, Sandy Bridge-E Ivy Bridge, Ivy Bridge-M, Ivy Bridge-E Haswell, Haswell-E, ... Broadwell, Broadwell-E, ... Skylake, Skylake-X, ... Kaby Lake, Kaby Lake-X, ... Coffee Lake, ... Whiskey Lake Amber Lake Comet Lake, ... Cannon Lake Ice Lake

The Core-i series of Intel is a family of medium to high-priced x86 - microprocessors for mobile and desktop computers in typical applications such as office work, Internet, multimedia, leisure and play. At the end of 2008 it replaced the Core 2 family; The current 7th generation is in the area of ​​CPUs with more than 8 cores (categories i7 and i9), the 8th generation in the area of ​​up to 8 cores (categories i3, i5 and i7) and the 9th generation in the area of ​​mobile processors. The 10th generation for i3, i5, i7 and i9 processors is newly released (as of Jan 2020).

The Core i series serves the largest market segment; Intel still offers the affordable Pentium and even cheaper Celeron processors in a price class below that, while the Xeon processors are intended for servers. Like the previous series, it supports the Intel 64 extension and all SSE extensions up to SSE4.2.

designation

Categories (Intel Core i3-, i5-, i7-, i9-)

With the categories i3, i5, i7 and, since 2017, i9, Intel divides the processors into performance classes. The assignment does not depend on the socket used , but on various processor properties. The number of cores , Hyper-Threading support, the width of the memory connection and the support for short-term overclocking (turbo mode) play a role, but features such as Trusted Execution Technology (TXT) and parts of the cache can also be used with processors with lower Model numbers must be disabled. However, this classification is not a reliable indicator of performance. So it is quite possible that a Core i5 desktop processor calculates significantly faster than a Core i7 for notebooks.

The Core i7 processor series is initially intended for the high and maximum performance segment, which is why there is also an Extreme Edition with a free multiplier in the tradition of previous Intel series. Core i5 is positioned directly below the Core i7 . At the beginning of 2010, new dual-core processors ( i3 and partly i5 ) in 32 nm production followed, with which the older Core 2 duos were replaced. In May 2016, Intel expanded the i7 series with 6, 8 and 10-core CPUs for the desktop based on Broadwell E in 14 nm production for the desktop, which are not aimed exclusively at the high-performance segment.

Before the appearance of a new microarchitecture, Xeon processors are sometimes sold as i-Core family members, which are then usually referred to as Core i7 variants or Extreme Editions . They typically have more than 4 processor cores, the server processor socket, 4 main memory channels and have no integrated processor graphics; for these reasons, they also require special motherboards. Code names for these processor series were: Sandy Bridge E, Ivy Bridge E, Haswell E, Broadwell E and, most recently, Skylake X.

In May 2017, Intel announced the Core i9 platform for the third quarter of 2017, the top model of which, the Core i9-7980XE, has 18 cores. These are modifications of the Xeon Scalable Processors with Skylake-X cores, which use the 2066 socket with DDR4 main memory channels. The first CPUs with the processor core called Coffee Lake have been available since October 2017, for which mainboards must be equipped with the chipset of the 300 series.

Technically, the Intel Core i series still includes the Intel Pentium G and Intel Celeron G processors , which have been slimmed down, but are based on the same architecture in contrast to the J and N series .

Model numbers (3, 4, or 5 digits)

The following important information is missing from this article or section:

Mobile processors like Clarksfield , Arrandale and Amber Lake are missing. A complete table broken down by desktop, mobile and enthusiast processors can be found under en: Tick – tock model .

Help Wikipedia by researching and pasting it .

Line of Core i processors

Node	architecture	Development step	Gene.	Processors	Release	Cores	R.A.M.	base	Structure of the model number	example
45 nm	Nehalem	Nehalem	01	Lynnfield	Aug. 09	04th	2	Socket 1156	007 xx or  8 xx	i7-860
				Bloomfield	Nov. 08	04th	3	Socket 1366	009 xx	i7-920
32 nm		Westmere		Clarkdale	Apr. 10	02	2	Socket 1156	005 xx or  6 xx	i5-680
				Gulftown	Jul 10	06th	3	Socket 1366	009 xx	i7-970
	Sandy Bridge	Sandy Bridge	02	Sandy Bridge	Jan. 11	2-4	2	Socket 1155	02 xxx	i7 2 600
				Sandy Bridge-E	Nov. 11	4-6	4th	Base 2011	03 xxx	i7-3930K
22 nm		Ivy Bridge	03	Ivy Bridge	Apr. 12	2-4	2	Socket 1155	03 xxx	i7 3 770
				Ivy Bridge-E	Sep 13	4-6	4th	Base 2011	04 xxx	i7-4930K
	Haswell	Haswell	04th	Haswell	Jun. 13	2-4	2	Socket 1150	04 xxx	i7 4 770
				Haswell-E	Aug. 14	6-8	4th	Base 2011-3	05 xxx	i7-5930K
14 nm		Broadwell	05	Broadwell	Jun. 15	2-4	2	Socket 1150	05 xxx	i7 5 775C
				Broadwell-E	May 16	6-10	4th	Base 2011-3	06 xxx	i7-6950X
	Skylake	Skylake	06th	Skylake	Aug. 15	2-4	2	Socket 1151	06 xxx	i7- 6 700K
				Skylake-X	May 17	6-18	4th	Base 2066	07 xxx	i7-7820X
		Kaby Lake	07th	Kaby Lake	Aug 16	2-4	2	Socket 1151	07 xxx	i7 7 700K
				Kaby Lake-X	Jun. 17	04th	2	Base 2066		i7 7 740X
		Coffee Lake	08th	Coffee Lake	Aug. 17	4-6	2	Socket 1151	08 xxx	i7- 8 700K
				Coffee Lake-S	Oct. 18	6-8	2		09 xxx	i7-9700K
				Coffee Lake-X	Q4 18th	8-18	4th	Base 2066	09 xxx	i9-9980XE
		Whiskey Lake		Whiskey Lake	Aug. 18	2-4	2	FCBGA 1528	08 xxx	i7- 8 565U
		Comet Lake	0?	Comet Lake-U	Aug. 19	2-6	2	FCBGA 1528	10 xxx	i7-10710U
		Cascade Lake X	0?	Cascade Lake X	Q4 19th	10-18	4th	Base 2066	10 xxx	i9-10980X
10 nm		Cannon Lake	0?	Cannon Lake	after 18	02	2	BGA 1356	08 xxx	i3-8121U
	Ice Lake	Ice Lake	0?	Ice Lake	Nov. 19	2-4	2	FCBGA 1528	10 xx G x	i7-1065G7
		Tiger Lake	0?	Tiger Lake	?

generation

The category (“i3” to “i9”) is followed by a model number, which in the first generation ( Nehalem ) consisted of three digits. From the second generation, Intel put a digit in front of the generation, so that the model number now consisted of four digits. From the 10th generation, the model number has four or five digits. The first two digits stand for the generation (micro-architecture) of the processor (gen number) .

From the 7th "generation" onwards, this scheme is broken by Intel: the 10 nm manufacturing process is delayed by more than 3 years, contrary to the original planning, and Intel's development model process architecture (also called tick-tock) is no longer feasible Microarchitecture called Cannon Lake is ultimately eliminated completely. Intel is now highlighting a further development step : process - micro-architecture - optimization . In this sense, the developments called micro-architecture from Skylake to Ice-Lake (the jump to the 10 nm process) are all optimizations; the micro-architecture still largely corresponds to the Skylake micro-architecture. Nevertheless, new generation digits are assigned to Kaby Lake, Coffee Lake, Whiskey Lake and Comet Lake. In the server area ( Intel Xeon ) these microarchitectures are largely omitted, here a switch is made from Skylake to Cascade Lake. Since an Extreme series (with the ending "-X") is always reused by the server CPUs , the assignment generation - microarchitecture gets completely confused: Skylake-X, like Kaby-Lake, is run under Generation 7, Cascade-Lake- X, like Comet-Lake and Ice-Lake, is run under Generation 10. We can therefore no longer speak of a schema.

High-end desktop processor (HEDT): requires server board. The digits in front of the model number are increased by 1 compared to the generation number.
Desktop processor: but requires server board.
Desktop processor: Refresh with increased clock frequency, normal board, digits in front of the model number are increased by 1 compared to the generation number.
Mobile processor: reduced power consumption, lower performance, solder base

item number

The last two or three digits stand for the article number (Stock Keeping Unit digits) . Intel awards them according to changing rules. Number correlates with performance and even more so with price.

Suffixes (max. 2 letters)

This section needs to be revised: The following table is still half-baked and incorrect, many duplications ( E, H, P, X, Q ), TDP statements imprecise (<17 ... 18 W), statements about L are incorrect , etc. . → Helpful link to it
Please help to him improve , and remove then this mark.

The model number can be followed by up to two capital letters:

Letters (combinations) attached to model numbers

	suffix	TDP	Further meaning
CPU	E.		An E or a combination of TE , UE , ME , LE , QE or EQ always stands for embedded .
	X		An X or in combination an XM , MX or XE stands for eXtreme. It is i. d. Typically Xeon processors, which were sold as particularly powerful Core-i before the release date of the successor series. Mostly they are processors of the previous generation, i. H. a 5th generation Xeon is "re-branded" into a 6th generation Core-i-Extreme.
	F.		An F stands for a missing graphics unit (from Core ™ i? -9xxx).
	K		An open-top clock multiplier
Desktop CPU	C.		An upwardly open clock multiplier on the Core ™ i5-5675C and Core ™ i7-5775C.
	S.		An S stands for energy savings through reduced performance, mostly through later / less frequent activation of the "turbo mode" (performance-optimized, "performance-optimized lifestyle").
	T		A T stands for energy savings through reduced equipment, mostly equipped with fewer cores than the regular model without suffix (Energieoptimiert, "Power-Optimized Lifestyle). This variant includes the energy optimizations of the S series (later switching on of the turbo mode) .
	E.	<55 W	An E stands for energy-efficient processors. This suffix was used in the Core Duo and is the forerunner of the S and T suffixes .
	P		Desktop CPUs with either a slimmed-down or without an integrated graphics unit.
	H		Both letters stand for high-performance graphics. R stands for desktop CPUs that are to be mounted on a notebook socket, while CPUs with H are to be installed on regular desktop sockets. H can be combined with other letters, e.g. B. HQ for a quad-core with high-performance graphics.
	R.
Mobile CPU	LM		Mobile CPUs with reduced ( LM , low power mobile ) or greatly reduced ( UM , ultra low power mobile ) TDP. As of the Core-i-2000M series, the suffixes LM and UM are no longer applicable , so that it is necessary to study the data sheet.
	AROUND
	U	<17 ... 18 W.	From the Core i 3000 series notebook processors with reduced voltage. U stands for ultra low power .
	Y	<13 W	From the Core i 3000 series notebook processors with reduced voltage. Y for extremely low power .
	L.	12 ... 19 W.	An L or a P stand for a particularly energy-saving processor. These letters were used in the Core Duo and are the technical precursors of the U and Y suffixes . Contradiction: i3-9100H L or i7-2655 L E have a TDP of 25 W.
	P	20 ... 29 W.
	M.		Mobile dual cores for mobile
	QM		Mobile Quad-Cores for Quad-Core Mobile
	MQ ,  HQ ,  HK	35 ... 45 W.	Mobile quad cores, the K indicates a free multiplier.
	G		Processors with integrated AMD Radeon graphics unit
Chipset- PCH	H		Entry class for non-K processors (not overclockable)
	B.		Entry-level models with better features than the H chipset
	Q		Professional models for companies
	Z		Chipset for mainstream overclockable processors
	X		High-end chipset for overclockable high-end desktop models

See also : Intel Core 2 # Prefix

Generations

See also : List of Intel Core i processors

Nehalem (1st generation)

→ Main article : Intel Nehalem microarchitecture

With the Nehalem generation, the Front Side Bus (FSB) and the Northbridge as a system connection for memory and peripherals were bid farewell. Instead, the main memory is connected directly to the CPU and communication between the chipset and processor takes place via a point-to-point connection called QuickPath Interconnect (QPI). In addition, the CPU itself has PCI Express lanes, to which the graphics card is usually connected. Previously, these transfers were made via an additional component, the northbridge, which was thus obsolete. Simultaneous Multithreading (SMT) also appears again, which was already used in Pentium 4 processors under the name Hyper-Threading, but disappeared with the core architecture.

Further innovations include a further expansion stage of the streaming SIMD extensions, SSE4.2, and that all quad-core processors are not composed of two dies, as is the case with Yorkfield and Kentsfield, but consist of one die.

Westmere Shrink

→ Main article : "Westmere" in the article Intel-Nehalem-Mikroarchitektur

Westmere represents the shrink on 32 nm. Furthermore, six-core processors appear for the first time in the non-server area.

Sandy Bridge (2nd generation)

→ Main article : Intel Sandy Bridge microarchitecture

Sandy Bridge

Sandy Bridge processors are available in both a quad-core version and a dual-core version with an integrated GPU . There are two versions of the dual-core version, one with the larger GPU with 12 shader units and another with a smaller GPU with only 6 shader units. This generation also has models for the desktop area that take place on the 1155 socket, and models for the mobile area that use the PGA 988B (G2), BGA 1023 and BGA 1224 sockets.

Sandy Bridge E.

The six-core models sold in the desktop segment are actually native eight-cores with two deactivated cores. These processors do not have an integrated graphics processor, but the integrated PCIe controllers offer more than twice as many lanes than is the case with the Sandy Bridge (without E). The Patsburg chipset X79 is still connected via DMI with 20 Gbit / s. Due to the additional PCIe lanes, the quad-channel memory controller, and the origin of the CPU from the server segment, where there are also QPI interfaces (which are switched off on the desktop pendant) for connecting further CPUs, the number of socket pins increases to 2011 pins on.

In February 2012, Intel also presented a quad-core offshoot for the 2011 socket. The processor design is identical to the eight-core version, with the exception of the reduced L3 cache and fewer cores and thus fewer transistors and a smaller chip area. However, the quad-core variant does not offer an open multiplier.

Ivy Bridge (3rd generation)

→ Main article : "Ivy Bridge" in the article Intel-Sandy-Bridge-Mikroarchitektur

Ivy Bridge processors are available in both quad-core and dual-core versions. As with the Sandy Bridge, there are variants with a full expansion stage of the GPU with 16 shader units, but also those with only 6 shader units.

Haswell (4th generation)

→ Main article : Intel Haswell micro-architecture

Haswell processors are available in both quad-core and dual-core versions. All models available so far contain an integrated GPU . In contrast to the previous generation, Haswell processors use the newly developed LGA1150 socket, which means they are not compatible with the mainboards of the previous series.

Broadwell (5th generation)

→ Main article : Intel Broadwell micro-architecture

Within the Broadwell generation there are again versions with and without integrated processor graphics. The fastest processor graphics HD Graphics 6000 or Iris Pro 6200 now have 48 shader units, the Iris Pro 6200 an additional 128 MB embedded DRAM on the processor package.

The sockets BGA 1168 , BGA 1364 (mobile computer) and LGA 1150 (desktop) continue to be used as with the Haswell generation, i3 / i5 / i7-5xxx have 2 DDR4 main memory channels and up to 4 cores.

The i7-68xx and i7-69xx versions are derived from the Xeon E5-v4 server processors and, like these, have the 2011-3 socket , 4 DDR4 main memory channels, 6, 8 or 10 cores (i7-6950X Extreme Edition) and no integrated processor graphics. You will Broadwell E called.

Skylake (6th generation)

Skylake generation graphics cores

Graphics core		Shader	4k @ 60Hz	eDRAM
HD Graphics 510	GT1	012		00-	no more analog interfaces (VGA)
HD Graphics 520	GT2	024	Yes	00-	eDP, DP , HDMI 1.4, DVI (three displays simultaneously)
Iris ™ Graphics 540	GT3e	048	Yes	064 MB	hardly any increase in performance compared to the Iris Pro Graphics 6200
Iris ™ Pro Graphics 580	GT4e	072	Yes	128 MB	Performance increase over Iris Pro Graphics 6200

→ Main article : Intel Skylake micro-architecture

In summer 2015 the first Skylake Core ™ i5 / i7 will appear.

In January 2016 the first Core ™ i7 processors of the Skylake generation with Iris Pro graphics (e.g. Core ™ i7-6970HQ) will appear

New this time is a fourth performance level of the integrated graphics, the Iris Pro Graphics 580.

The performance data of the other Core ™ i3 / i5 / i7-6xxx models are within the usual scope of the Core i processors: 2-4 cores, 2 main memory channels and a total of four different integrated processor graphics. These no longer have a VGA output, but can control up to three screens at the same time. Compared to the Broadwell generation, the processors come with new sockets: Sockets LGA 1151, BGA 1356 and new chipsets, so they need new motherboards.

Skylake-X

In the summer of 2017, Intel announced the Skylake-X processor series. Processors with more than eight cores are assigned to the Core ™ i9 series. Several trade magazines criticized insufficient product maturity. Apparently the announcement was hasty and a reaction to the competition AMD, whose Zen Threadripper processor was released in August 2017. Skylake-X is a reused server processor from the Xeon Scalable Processor series. This also differs in the processor core from the Core ™ i3 / i5 / i7-6xxx models, because for the first time two SIMD - AVX-512 instruction set expansion units are installed. Special motherboards are again required for this series of processors, this time a different processor socket than for the server models is used, socket 2066 . Four DDR4 main memory channels, up to 128 GB working memory, but no error correction (ECC) are supported. The Skylake-X models have 6 to 18 cores (Extreme Edition) and require the X299 chipset.

The 2018 releases of the Skylake X generation are still 6th generation processors. Intel deviates from the nomenclature it has been using since 2008. The following were changed:

• Relabelling (without architecture changes, the first digit is increased from "7" to "9", i9-9820X ​​is no longer listed as i7)
• Better thermal interface between die and heatspreader (again soldered with indium, verified at least on the i9-9980XE)
• Use of the high core count die also for the "small" CPUs (i9-9900X, i9-9820X, i7-9800X)
• Use of the L3 cache of unused cores (plus 5 ½ MB: i9-9900X, i7-9800X; plus 2 ¾ MB: i9-9920X, i9-9820X)
• Activation of all 44 PCI lanes (i9-9820X, i7-9800X)
• Use of two additional cores (i9-9820X, i7-9800X)
• Increase the core tension by about 12% at maximum cycle
• Increase the TDP to 165 W for all models

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i9-7980XE	Q3 17th	18 (36)	044	2.6 GHz	4.4 GHz	165 W	24 ¾  MB	up to here from the high core count - which arose
Core ™ i9-7960X		16 (32)		2.8 GHz			22 MB
Core ™ i9-7940X		14 (28)		3.1 GHz			19 ¼ MB
Core ™ i9-7920X		12 (24)		2.9 GHz		140 W	16 ½ MB
Core ™ i9-7900X	Q2 17th	10 (20)		3.3 GHz	4.5 GHz		13 ¾ MB	from here on from the low core count - they emerged
Core ™ i7-7820X		08 (16)	028	3.6 GHz			11 MB
Core ™ i7-7800X		06 (12)		3.5 GHz	4.0 GHz		08 ¼ MB
Core ™ i9-9990XE	Jan. 19	14 (28)	044	4.0 GHz	5.0 GHz	255 W	19 ¼ MB	OEM only
Core ™ i9-9980XE	Q4 18th	18 (36)	044	3.0 GHz	4.5 GHz	165 W	24 ¾ MB	all CPUs are derived from the high core count - die
Core ™ i9-9960X		16 (32)		3.1 GHz			22 MB
Core ™ i9-9940X		14 (28)		3.3 GHz			19 ¼ MB
Core ™ i9-9920X		12 (24)		3.5 GHz
Core ™ i9-9900X		10 (20)		3.5 GHz
Core ™ i9-9820X		10 (20)		3.3 GHz	4.2 GHz		16 ½ MB
Core ™ i7-9800X		08 (16)		3.8 GHz	4.5 GHz

Kaby Lake (7th generation)

→ Main article : Intel Kaby Lake microarchitecture

The Kaby Lake generation first went on sale in August 2016.

Kaby Lake-X

Two Kaby Lake processors require expensive mainboards with an LGA 2066 socket, but hardly use the advantages of these. Only two memory channels and 16 PCIe lanes are used, for which an LGA-1155 board would be sufficient. The only difference to the Kaby Lake-S processors for socket 1151 is the higher TDP. These processors were discontinued just 11 months after their market launch.

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i7-7740X	Q2 17th	04 (8)	016	4.3 GHz	4.5 GHz	112 W.	08 MB	only two main memory channels
Core ™ i5-7640X		04 (4)		4.0 GHz	4.2 GHz		06 MB

Coffee Lake (8th generation)

→ Main article : Intel Coffee Lake microarchitecture

In early October 2017, Intel released the first Coffee Lake processors for desktop PCs. For the first time there are six-core processors in the desktop area (Core ™ i7-8700K, Core ™ i5-8600K and Core ™ i5-8400).

On April 3, 2018, more Coffee Lake processors came on the market, which were intended for the lower price segment. The Core ™ i3 category included quad-core processors for the first time. The new models include the following:

• Celeron G4900 with 3.1 GHz and G4290 with 3.2 GHz and Pentium Gold G5400 to G5600 with 3.7 to 3.9 GHz
• Core ™ i3-8300, Core ™ i5-8500 and Core ™ i5-8600

The following models are among the currently strongest and most important Coffee Lake CPUs:

The most important Intel Core Coffee Lake CPUs at a glance

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i7-8086K	June 18	06 (12)	016	4.0 GHz	5.0 GHz	095 W	12 MB	Die size 16.6mm × 9.2mm
Core ™ i7-8700K	Oct. 17			3.7 GHz	4.7 GHz	095 W
Core ™ i7-8700	Apr. 18			3.2 GHz	4.6 GHz	065 W
Core ™ i7-8700T	Apr. 18			2.4 GHz	4.0 GHz	035 W
Core ™ i5-8600K	Oct. 17	06 (6)		3.6 GHz	4.3 GHz	095 W	09 MB
Core ™ i5-8400	Oct. 17			2.8 GHz	4.0 GHz	065 W
Core ™ i5-8500T	Apr. 18			2.1 GHz	3.5 GHz	035 W
Core ™ i3-8350K	Oct. 17	04 (4)		4.0 GHz		091 W	06 MB
Core ™ i3-8100	Oct. 17			3.6 GHz		065 W

Coffee Lake Refresh (9th generation)

All of the Core i5 and i7 processors listed here have a soldered heatspreader, see cooling problems . The Core i3 category included processors with Turbo Boost for the first time. The Core ™ i7 category has more cores than the previous generation, but loses Hyper-Threading.

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i9-9900KS	Oct. 19	08 (16)	016	4.0 GHz	5.0 GHz	127 W.	16 MB	only 1 year manufacturer warranty
Core ™ i9-9900K	Oct. 18			3.6 GHz	5.0 GHz	095 W		Die size 19.4mm × 9.2mm
Core ™ i9-9900KF	Q1 19th			3.6 GHz	5.0 GHz	095 W
Core ™ i9-9900	May 19			3.1 GHz	5.0 GHz	065 W
Core ™ i9-9900T	May 19			2.1 GHz	4.4 GHz	035 W
Core ™ i7-9700K	Oct. 18	08 (8)		3.6 GHz	4.9 GHz	095 W	12 MB
Core ™ i7-9700	May 19			3.0 GHz	4.7 GHz	065 W
Core ™ i7-9700T	May 19			2.0 GHz	4.1 GHz	035 W
Core ™ i5-9600K	Oct. 18	06 (6)		3.7 GHz	4.6 GHz	095 W	09 MB
Core ™ i5-9600	May 19			3.1 GHz	4.6 GHz	065 W
Core ™ i5-9600T	May 19			2.3 GHz	3.9 GHz	035 W
Core ™ i5-9500	May 19			3.0 GHz	4.4 GHz	065 W
Core ™ i5-9500T	May 19			2.2 GHz	3.7 GHz	035 W
Core ™ i5-9400	March 19			2.9 GHz	4.1 GHz	065 W
Core ™ i3-9350K	March 19	04 (4)		4.0 GHz	4.6 GHz	091 W	08 MB
Core ™ i3-9100	May 19			3.6 GHz	4.2 GHz	065 W	06 MB

Whiskey Lake (8th generation)

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i7-8565U	Aug. 18	04 (8)	016	1.8 GHz	4.6 GHz	15 W (10-25 W)	08 MB	Intel UHD Graphics 620
Core ™ i5-8265U				1.6 GHz	4.1 GHz		06 MB
Core ™ i3-8145U		02 (4)		2.1 GHz	3.9 GHz		04 MB

Comet Lake (10th generation)

Another chip generation in 14 nm that Intel has announced. More information has been on Github since January 29, 2019.

There will be three categories:

• Comet Lake-U with 2 to 6 cores
• Comet Lake-H with 6 to 10 cores
• Comet Lake-S with 6 to 10 cores

The HD Graphics 520 GT2 is used as the graphics. The CPUs support AVX2, but not AVX512.

The most important Intel Core Comet Lake CPUs at a glance

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i7-10710U	Aug. 19	06 (12)	016	1.1 GHz	4.7 GHz	015 W	12 MB
Core ™ i7-10510U	Aug. 19	04 (8)		1.8 GHz	4.9 GHz		08 MB
Core ™ i5-10210U	Aug. 19	04 (8)		1.6 GHz	4.2 GHz		06 MB
Core ™ i3-10110U	Aug. 19	02 (4)		2.1 GHz	4.1 GHz		04 MB

Cannon Lake Shrink

Launched on May 15, 2018. Economical mobile processor without graphics unit. Sample devices available since the end of December 2018, the chip never went into mass production due to manufacturing problems in the 10 nm process, and end devices with these chips were never sold. Cannon Lake is thus skipped as a 10 nm CPU architecture, the first 10 nm devices come with Ice Lake CPUs (see below).

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i3-8121U	Dec. 18	2 (4)	016	2.2 GHz	3.2 GHz	15 W	04 MB	no integrated GPU, die area approx. 71 mm²

Cascade Lake (10th generation)

Like Skylake-X, this series is a re-use of the Xeon server CPUs from the Cascade Lake series with its own socket ( socket 2066 ) for 4 DDR4 main memory channels and therefore requires its own motherboard again. As with Skylake-X, the X299 is used as the chipset, so that Cascade-Lake-X is socket-compatible with Skylake-X CPUs.

• see Intel Cascade Lake microarchitecture and Intel Xeon (Cascade Lake)

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i9-10980XE	Q4 19th	18 (36)	048	3.0 GHz	4.6 GHz	165 W	24 ¾ MB
Core ™ i9-10940X		14 (28)		3.3 GHz			19 ¼ MB
Core ™ i9-10920X		12 (24)		3.5 GHz
Core ™ i9-10900X		10 (20)		3.7 GHz	4.5 GHz

Ice Lake (10th generation)

• Launched August 1st, 2019
• Economical mobile processor
• System-on-a-chip, the actual CPU occupies just under 25 percent of the chip area with quad-core
• Native support for typical notebook functions
• Probably no more classic CPU-PCI-Express lanes, only PCI-Express lanes via the Platform Controller Hub
• Native support for Thunderbolt 3 and USB 3.1
• Integrated graphics and DP 1.4
• Support of LP-DDR up to 3733 MT / s
• Intel IPU
• New Gaussian Neural Accelerator

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	Remarks
				base	turbo
Core ™ i7-1065G7U	Aug. 19	4 (8)	016	1.3 GHz	3.9 GHz	15 W	08 MB	The area 122.5 mm² 4 ×  Thunderbolt 3
Core ™ i7-1060G7U		4 (8)		1.0 GHz	3.8 GHz	09 W.	08 MB
Core ™ i5-1035G7U		4 (8)		1.1 GHz	3.7 GHz	15 W	06 MB

Lakefield (10th generation)

• Launched June 10, 2020
• Economical mobile processor with graphics unit and RAM
• The entire chip is 12 × 12 × 1 mm
• Uses Intel's Foveros technology to stack chips
• extremely small die of the CPU with 82 mm²
• Up to 8gb LP-DDR4-4266 PoP-RAM integrated directly in the chip
• 7 watt TDP in turbo mode max 9 watt TDP for i5 variant
• Extremely low standby consumption with 2.5 mW standby SoC power

model	Release	Cores (threads)	PCIe lanes	Tact		TDP	L3 cache	GPU clock	GPU EU's	Remarks
				base	turbo
Core i5-L16G7	Jun. 20	5 (5)	06th	1.4 GHz	3.0 GHz	7 W	04 MB	500 MHz	64	The area 92 mm²
Core i3-L16G4		5 (5)		0.8 GHz	2.8 GHz	07 W	04 MB		48

Cooling problems

Numerous hardware manufacturers complained that processors from the Haswell generation onwards tend to overheat under full load and get hotter than their predecessor processors at the same rate. The cause is an Intel with the introduction of the generation Ivy Bridge altered production process, after which the space between the the or the unhoused processor and the heatspreader not with indium (thermal conductivity: 82 W / K · m) soldered, but a thermal grease ( max. 12.5 W / K m) is filled. The thermal resistance increases by about 0.1 K / W. With the same power loss and cooling, the processor dies are 10 to 13 K hotter. In the case of processors of the 9th generation, Intel re-soldered the heat spreaders, as at least the processors presented so far all have a turbo clock of at least 4.6 GHz.

Product logos

• 

  Intel Core i7
  (Westmere)

• 

  Intel Core i3
  (Sandy Bridge)

• 

  Intel Core i5
  (Sandy Bridge)

• 

  Intel Core i7
  (Sandy Bridge)

• 

  Intel Core i3
  (Haswell)

• 

  Intel Core i5
  (Haswell)

• 

  Intel Core i7
  (Haswell)

See also

• List of Intel Core i processors
• Intel Core M series
• Intel Pentium Dual-Core (Desktop)
• Intel Celeron dual core

Web links

• Details on the Core i7 at Intel (English).
• Details on the Core i7 XE at Intel (English).
• Nehalem - Everything You Need to Know about Intel's New Architecture. In: AnandTech , article on the Nehalem architecture, November 3, 2008 (English).
• Intel Core i7 Extreme 965, 940 and 920 in the test (Bloomfield). In: Au-Ja , detailed test with benchmarks, overclocking and details about the architecture, from November 3, 2008.
• Intel Core i5 750 and Core i7 870 processors with Lynnfield core in the test. In: Allround-PC.com , details on technology and benchmarks, September 8, 2009.

Individual evidence

1. ↑ Core i9-9980HK , intel.com
2. ↑ Core i9-9980XE , intel.com
3. ↑ Test: Core i7-6950X and i7-6900K: Broadwell-E for Socket 2011-v3 in the test. In: PC Games Hardware , May 31, 2016.
4. ↑ heise.de
5. ↑ Core i7-8700K, i5-8600K and i5-8400 in the CPU test: Intel at the limit - Coffee Lake is here In: PC Games Hardware , October 5, 2017.
6. ↑ Intel's Sandy Bridge. In: AnandTech , September 14, 2010
7. ↑ https://www.heise.de/ct/artikel/Intel-10-Nanometer-Processoren-erst-2019-4036836.html
8. ↑ Information on Intel processor numbers, www.intel.de
9. ↑ Intel Skylake: Two new processors in the desktop portfolio. In: PC Games Hardware , January 22, 2016
10. ↑ ^{a b c d} Intel Desktop Chipset Roadmap H2 2016 to 2018. In: AnandTech Forum. September 12, 2014, accessed December 22, 2017 . 
11. ↑ Michael Günsch: Intel Basin Falls: Skylake-X with up to 12 cores at Computex at the end of May. In: ComputerBase. April 21, 2017. Retrieved December 22, 2017 . 
12. ↑ Intel: Intel® Core ™ i5-5575R Processor. In: Intel Product Specification. June 1, 2015, accessed September 27, 2017 . 
13. ↑ Heise Newsticker - Iris Pro graphics for desktop PCs. Retrieved December 10, 2017 . 
14. ↑ Intel's Skylake processors. Retrieved December 10, 2017 . 
15. ↑ Intel Ark, Product Directory. Retrieved December 10, 2017 . 
16. ↑ Heise Newsticker, first Skylake processors with Iris Pro graphics. Retrieved December 10, 2017 . 
17. ↑ Intel Product Specifications - Intel Core X Series. Retrieved December 10, 2017 . 
18. ↑ Intel Product Specifications - X299 Chipset. Retrieved December 10, 2017 . 
19. ↑ https://www.youtube.com/watch?v=LuXL0GN3SAk
20. ↑ https://www.youtube.com/watch?v=s1Ww2vNAjN0
21. ↑ https://www.anandtech.com/show/13804/intel-core-i9-9990xe-up-to-5-ghz-auction-only
22. ↑ Volker Rißka: Intel Kaby Lake-X: After eleven months, the CPU flop is stopped. In: ComputerBase. May 1, 2018, accessed May 17, 2018 . 
23. ↑ heise online: More Coffee Lake processors for desktop PCs. Retrieved January 22, 2018 . 
24. ↑ Michael Söldner: Coffee Lake: New Intel processors for beginners . In: PC WORLD . ( pcwelt.de [accessed on January 22, 2018]).
25. ↑ i5 or i7: Which CPU is better for gaming? In: Gaming Gear comparisons - only4gamers.de . January 18, 2018 ( only4gamers.de [accessed January 22, 2018]).
26. ↑ https://www.computerbase.de/2018-06/intel-core-i7-8700t-i5-8500t-cpu-test-coffee-lake/#abschnitt_intel_coffee_lake_mit_35_watt_in_der_uebersicht
27. ↑ Intel usually gives a 3-year manufacturer's guarantee on CPUs, without reversing the burden of proof and with replacement CPUs, even if these are hardly available in stores. EU warranty without reversal of evidence by dealers is only available for half a year, after which the customer is obliged to prove an error when purchasing.
28. ↑ https://www.youtube.com/watch?v=Ccqid7FcjOU
29. ↑ https://www.notebookcheck.com/Intel-Comet-Lake-U-Noch-eine-Generation-auf-14-nm-Basis.308680.0.html
30. ↑ https://www.computerbase.de/2018-06/intel-10-nm-cpu-cannon-lake-die-size
31. ↑ https://www.cpu-monkey.com/en/cpu-intel_core_i7_1065g7-933
32. ↑ https://www.computerbase.de/2020-06/intel-lakefield-3d-hybrid-cpu-foveros/?amp=1
33. ↑ https://www.computerbase.de/2020-06/intel-lakefield-3d-hybrid-cpu-foveros/?amp=1
34. ↑ Jens Stark: Haswell is getting too warm. In: Computerworld.ch. June 11, 2013, accessed March 26, 2016 . 
35. ↑ Haswell: Almost uncontrollable heat hotspots? In: tweakpc.de. July 24, 2013, accessed March 26, 2016 . 
36. ↑ Core i7-4770K decapitated, sanded & polished: 12 degrees temperature gain. In: 3dcenter.org. July 22, 2013, accessed March 26, 2016 . 
37. ↑ Officially presented: Intel Core i5-9600K, Core i7-9700K and Core i9-9900K. In: hardwareluxx.de. October 8, 2018, accessed March 16, 2019 .