Tuesday, August 14, 2007




CORE 2 PROCESSORS

The Core 2 brand refers to a range of Intel's mobile and desktop 64-bit dual- and quad-core x86 CPUs based on the Intel Core microarchitecture, a microarchitecture that evolved from the 32-bit dual-core Yonah mobile processor. The Yonah comprised two interconnected single cores (the Pentium M branded microprocessor derivatives) coupled as a single die silicon chip (IC). The Core 2 marked a relegation of the Intel's Pentium brand to a lower-end market, and a reunification of Intel's notebook and desktop brand names, previously divided into the Pentium M and Pentium 4 lines.

The Core 2 brand was launched on July 27, 2006, comprising Duo (dual-core), Quad (quad-core), and Extreme (dual- or quad-core CPUs with higher speeds and unlocked multiplier) branches. The brand covers processors based on various cores; "Conroe" and "Allendale" (dual-core for higher- and lower-end desktop use, respectively), "Merom" (dual-core for notebooks), "Kentsfield" (quad-core for desktops), and their variants named "Penryn" (dual-core for notebooks), "Wolfdale" (dual-core for desktops) and "Yorkfield" (quad-core for desktops). Although the "Woodcrest", "Clovertown" and the upcoming "Tigerton" core CPUs for servers and workstations are also based on Core architecture, they are marketed under the Xeon brand.[1] The first Core 2 Duo processors were fabricated on 300 mm wafers using a 65 nm manufacturing process.

Unlike the NetBurst architecture of Pentium 4 or Pentium D branded processors, the Core architecture does not stress extremely high clock speeds, but rather improvements in the processor's usage of both available clock cycles and power. This translated into more efficient decoding stages, execution units, caches, and buses, etc, reducing the Core 2 CPU's power consumption, while enhancing their processing capacity. With a TDP of up to only 65 W, the Core 2 dual-core Conroe consumed only half the power of some of the higher end dual-core Pentium D branded desktop chips[2] with a TDP of up to 130 W[3] (a high TDP requires many noisy cooling fans or a more expensive noiseless cooling system) and was more capable compared to them.

Intel Core 2 processors feature Intel 64 (also known as EM64T), Virtualization Technology (except T5500 or lower end E4x00), Execute Disable Bit, and SSE3. Core 2 also introduced SSSE3, Trusted Execution Technology, Enhanced SpeedStep, and Active Management Technology (iAMT2).

Typically for CPUs, the Core 2 Duo E4000/E6000, Core 2 Quad Q6600, Core 2 Extreme X6800, QX6700 and QX6800 CPUs were affected by bugs, but apparently minor.


Current processor cores
Core 2 Duo brand logo
Core 2 Duo brand logo
Core 2 Quad brand logo
Core 2 Quad brand logo
Core 2 Extreme brand logo
Core 2 Extreme brand logo

[edit] Conroe

The first Intel Core 2 Duo processor cores, code-named Conroe and given the Intel product code 80557, were launched on July 27, 2006 at Fragapalooza, a yearly gaming event in Edmonton, Alberta, Canada. These processors are built on a 65 nm process and are intended for desktops, replacing the Pentium 4 and Pentium D. Intel has claimed that Conroe provides 40% more performance at 40% less power compared to the Pentium D. All Conroe processors are manufactured with 4 MiB L2 cache; however, for marketing purposes, the E6300 and E6400 versions based on this core have half their cache disabled, leaving them with only 2 MiB of usable L2 cache.

The lower end E6300 (1.86 GHz) and E6400 (2.13 GHz), both with a 1066 MT/s FSB, were released on July 27, 2006. Traditionally, CPUs of the same family with less cache simply have the unavailable cache disabled, since this allows parts that fail quality control to be sold at a lower rating. As yields improve, they may be replaced with versions that only have the cache amount needed on the die, to bring down manufacturing cost. At launch time, Intel's prices for the Core 2 Duo E6300 and E6400 processors were US$183 and US$224 each in quantities of 1000. Conroe CPUs have improved capabilities over previous models with similar processor speeds. According to reviews, the larger 4 MiB L2 cache vs. the smaller 2 MiB L2 cache at the same frequency and FSB can provide a 0–9% performance gain with certain applications and 0–16% performance gain with certain games.[5][6] The higher end Conroe processors are labeled as the E6600 and E6700 Core 2 Duo models, with the E6600 clocked at 2.4 GHz and the E6700 clocked at 2.67 GHz. The family has a 1066 MT/s front side bus, 4 MiB shared L2 cache, and 65 watts TDP. These processors have been tested against AMD's current top performing processors (Athlon 64 FX Series), which were, until this latest Intel release, the fastest CPUs available. Conroe chips also experience much lower heat output compared to their predecessors — a benefit of the new 65 nm technology. At launch time, Intel's prices for the Core 2 Duo E6600 and E6700 processors were US$316 and US$530, respectively, each in quantities of 1000.

E6320 and E6420 Conroe CPUs at 1.86 and 2.13 GHz respectively were launched on April 22, 2007 featuring a full 4 MiB of cache and are considered Conroes.

Intel released four additional Core 2 Duo Processors on July 22nd, 2007. The release coincided with that of the Intel Bearlake (x3x) chipsets. The new processors are named Core 2 Duo E6540, E6550, E6750, and E6850. Processors with a number ending in "50" have a 1333 MT/s FSB. The processors all have 4 MiB of L2 cache. Their clock frequency is similar to that of the already released processors with the same first two digits (E6600, E6700, X6800).[7] An additional model, the E6540, was launched with specifications similar to the E6550 but lacking Intel Trusted Execution Technology and vPro support. These processors are slated to compete with AMD's Stars processor line and are therefore priced below corresponding processors with a 1066 MT/s FSB.[8]

[edit] Conroe XE

The Core 2 Extreme was officially released on July 29, 2006. However some retailers appeared to have released it on July 13, 2006, though at a higher premium. The less powerful E6x00 models of Core 2 Duo were scheduled for simultaneous release with the X6800, which are both available at this time. It is powered by the Conroe XE core and replaces the dual-core Pentium Extreme Edition processors. Core 2 Extreme has a clock speed of 2.93 GHz and a 1066 MT/s FSB, although it was initially expected to be released with a 3.33 GHz and 1333 MT/s. The TDP for this family is 75–80 watts. At full load the X6800 does not exceed 45 °C (113 °F), and with SpeedStep enabled the average temperature of the CPU when idle is essentially that of the ambient atmosphere.[9]

At launch time, Intel's price for the Core 2 Extreme X6800 was US$999 each in quantities of 1000. Like the desktop Core 2 Duo, it has 4 MiB of shared L2 cache available. This means that the only major difference between the regular Core 2 Duo and Core 2 Extreme is the clock speed and unlocked multiplier, usual advantages of the "Extreme Edition." The unlocked upward multiplier is of use to enthusiasts which allow the user to set the clockspeed higher than shipping frequency without modifying the FSB unlike mainstream Core 2 Duo models which are downward unlocked only.

[edit] Allendale

There was contention as to whether the previously-available low-end Core 2 Duo desktop processors (E6300, at 1.86 GHz and E6400, at 2.13 GHz, both with 2 MiB L2 cache) are specimens of the Allendale core. Prior to Q1 2007, all E6300 and E6400 processors released were Conroe (4 MiB L2 cache) cores with half their L2 cache disabled. The Allendale core, manufactured with 2 MiB L2 cache in total, offers a smaller die size and therefore greater yields.

Quoted from The Tech Report:

You'll find plenty of sources that will tell you the code name for these 2 MB Core 2 Duo processors is "Allendale," but Intel says otherwise. These CPUs are still code-named "Conroe," which makes sense since they're the same physical chips with half of their L2 cache disabled. Intel may well be cooking up a chip code-named Allendale with 2 MB of L2 cache natively, but this is not that chip.[10]

Another difference between the premium E6000 series (Conroe core) or (Allendale core) and the E4000 series (Allendale core) is the front side bus clock rating. The E4000 series are rated to run on a quad-pumped 200 MHz front side bus ("800 MT/s") while the E6000 series are rated to run on a quad-pumped 266 MHz front side bus ("1066 MT/s"). The E4000 series also lack support for Intel VT-x instructions.

The currently available Core 2 Duo E4300 only uses an Allendale core, released on January 21, 2007. The Allendale processors use a smaller mask with only 2 MiB of cache, thereby increasing the number of chips per wafer. Allendale processors are produced in the LGA775 form factor, on the 65 nm process node. It is unclear and a matter of contention whether the E6300 and E6400 models are still Conroe processors with half their L2 cache disabled or Allendale processors. It is possible that the E6300 and E6400 models can be both Allendale and Conroe, differentiated by the L2 and B2 steppings respectively.

Initial list price per processor in quantities of one thousand for the E4300 was US$163. A standard OEM price was US$175, or US$189 for a retail package. Price cuts were enacted on April 22, 2007, when the E4400 was released at $133 and the E4300 dropped to $113. Allendale processors with half their L2 cache disabled were released in mid-June 2007 under the Pentium Dual-Core brand name.

On July 22nd, 2007, an E4500 Allendale was launched, phasing out the E4300 model.[8] This was accompanied by a price cut for the E4400 model.

[edit] Merom

Merom, the first mobile version of the Core 2, was officially released on July 27, 2006 but quietly began shipping to PC manufacturers in mid-July alongside Conroe.[11][12] Merom is Intel's premier line of mobile processors, with largely the same features as Conroe, but with more emphasis on low power consumption to enhance notebook battery life. Merom-based Core 2 Duo provides 20% more performance yet maintains the same battery life as the Yonah-based Core Duo. Merom is the first Intel mobile processor to feature Intel 64 architecture.

The first version of Merom is "drop-in" compatible with Napa platform for Core Duo, requiring at most a motherboard BIOS update. It has a similar thermal envelope of 34 W and the same 667 MT/s FSB rate.[13] Merom features 4 MiB L2 cache (budget T5xxx models have only 2 MiB L2 cache).

A second wave of Merom processors featuring an 800 MT/s FSB and using the new Socket P was launched on May 9, 2007.[14] These chips are part of Santa Rosa platform.[15] Low voltage versions were also released on May 9, 2007.

Merom (מרום) is the Hebrew word for a higher plane of existence or a level of heaven, BaMerom (במרום) means "in the heavens". The name was chosen by the Intel team in Haifa, Israel, who designed this processor.

See the Merom section of "List of Intel Core 2 microprocessors" for a list of Merom processors.

[edit] Kentsfield

The Kentsfield was the first Intel desktop quad core CPU branded as Core 2 (and Xeon for lower-end servers and workstations). The top-of-the-line Kentsfields were Core 2 Extreme models numbered QX6xx0, while the mainstream ones branded Core 2 Quad were numbered Q6xx0. All of them featured two 4 MB L2 caches. The mainstream Core 2 Quad Q6600, clocked at 2.4 GHz, was launched on January 8, 2007 at US$851 (reduced to US$530 on April 7, 2007, and to US$266 on July 22, 2007). July 22, 2007 marked the release of the next Core 2: Quad Q6700 and Extreme QX6850 Kentsfields at US$530 and US$999 respectively, and also price reduction of Core 2: Quad Q6600 and Extreme QX6800 Kentsfields to US$266 and US$999 respectively.[16]

Kentsfield, like the Pentium D Presler, comprised two separate silicon dies.[17] Each of Kentsfield's two dies was dual-core with the Core architecture, as did each die constituting alone every Core 2's dual-core only CPU. Hence, the max. power consumption (TDP) of the Kensfield (QX6800 - 130 watts,[18] QX6700 - 130 W,[19] Q6600 - 95 W[20]) was approx. double of its similarly clocked Core 2 Duo counterpart. For example, the QX6700 consisted of two E6700 chips connected together by a 1066 MT/s FSB on one MCM, resulting in lower costs but less bandwidth to the northbridge. The Kentsfield was one socket processor sitting in a LGA775 socket, as well as Core 2 Duo (AMD Quad FX consisted of two dual-core processors in two separate sockets on one motherboard with a 2 x 125 W[21]= 250 W TDP).

A quad-core CPU, like the Kentsfield, processes very well with multi-threaded applications (typical for video editing, ray-tracing, or rendering), where its processing ability may approach double that of each of its halves comprising the equally clocked dual-core CPU. Similarly, Pentium D Presler had two dice in one package. A two-die CPU, however, can rarely double the processing ability of each of its constituent halves (e.g. the Kentsfield rarely doubles the ability of the Conroe), due to a loss of performance resulting from connecting them (i.e. sharing the narrow memory bandwidth, and operating system overhead of handling twice as many cores and threads).

Single or dual-threaded applications alone, including most games, do not benefit from the second pair of cores of a quad-core CPU over an equally clocked dual-core CPU. For example, with no increase in FSB speed and a mild reduction in processor frequency of the quad-core Extreme QX6700 over the older dual-core Extreme X6800, overall performance of the Kentsfield (QX6700) for those applications does not improve. Nevertheless, the simultaneous use of several processor-intensive single/dual-threaded applications on a quad-core CPU will generally lead to a dramatic overall performance increase over an equally clocked dual-core CPU. A quad-core CPU is useful also to run both the client and server processes of a game without noticeable lag in either thread, as each instance (up to four) could be running on a different core.

[edit] Kentsfield XE

The first Kentsfield, named Core 2 Extreme QX6700 (product code 80562) and clocked at 2.67 GHz, was released on November 2, 2006 at US$999.[22][23][24][25][26] It was the first x86 quad-core processor ever, featured the Kentsfield XE core, and complemented the Core 2 Extreme X6800 dual-core processor based on the Conroe XE core. The CPUs with the Kentsfield XE core had the highest and unlocked multipliers like their Extreme predecessors. On April 8, 2007, a new top Kentsfield XE - the Core 2 Extreme QX6800 - was released within the 135 W TDP thermal envelope, and clocked at 2.93 GHz. It went to high end OEM-only systems at US$1,199.[27] On July 22, 2007, the new champ Core 2 Extreme QX6850 arrived, clocked at 3.0 GHz, with a faster 1.33 GHz FSB, and at US$999 - a typical price for a top Kentsfield XE CPU.

The formerly available Core 2 Extreme QX6700 was relaunched on July 22nd, 2007 as the Core 2 Quad Q6700, clocked at 2.66 GHz, but it consumes less power than QX6700. The price of the Q6600 has been dropped to US$266 on July 22nd, 2007.

[edit] Future processors
Current event marker This section contains information about scheduled or expected future computer chips.
It may contain preliminary or speculative information, and may not reflect the final specification of the product.

[edit] Merom XE

Merom XE is a laptop CPU designed for ultra-high end laptops. It will be released in two models, the X7900 and the X7800. These will feature an 800 MHz FSB. The X7800 will be clocked at 2.6 GHz and will cost around $795.The processor will feature a 44WT TDP and will require the new Intel Centrino (Santa Rosa) platform. The X7900 will be clocked at 2.8 GHz, but its cost is unknown, but expect it to be more than 1000 dollars per CPU in packs of 1000.

The X7900 processor is currently used in the top-end iMacs, released in August 2007.

[edit] Penryn

The successor to the Merom core currently used for the Core 2 Duo T5000/T7000 series mobile processors, code-named Penryn, will debut the 45 nanometer process that will also be used for the Conroe sequel, Wolfdale (see below). Many details about Penryn appeared at the April 2007 Intel Developer Forum. Its successor is expected to be Nehalem.

Important advances[28] include the addition of new instructions (including SSE4, also known as Penryn New Instructions) and new fabrication materials (most significantly a hafnium-based dielectric).

Penryn is intended to be paired with a new chipset, Bearlake,[29] which will include an increase in bus speed (connection to the northbridge, etc.) for certain models to 1333 MT/s and will include support for DDR3 SDRAM; Intel believes[30] that DDR3 is particularly useful in the power- and heat-constrained environments within mobile equipment.

Penryn will also be released in a quad-core version for desktop replacement notebooks.

[edit] Wolfdale and Yorkfield

Wolfdale will be the desktop version of Penryn, with two cores sharing 6 MiB of L2 cache and 1333 MHz FSB. Two Wolfdale dies on a single module will make up a Yorkfield chip, the successor to the Kentsfield processor.[31] Each die will share 6 MiB L2 cache, for a total of 12 MiB, and this chip will also have a 1333 MHz FSB. These processors are expected to become available in late 2007 or early 2008, and the platform will support DDR3. At the Intel Developer Forum 2007 a Yorkfield processor was compared with a Kentsfield processor.[32][33]

[edit] Successors

Even further, the latest known codenames for future processors (perhaps based on a new post-Core microarchitecture) are Nehalem and Sandy Bridge (formerly known as Gesher), and little is known about either of them. Nehalem will be a 45 nm process debuting in 2008, followed by a 32 nm shrink codenamed Westmere (Nehalem-C), and Sandy Bridge will be 32 nm with a new microarchitecture debuting in 2010, provided that Intel stays on target with its roadmap.[34] In 2011, Intel will launch its first processor based on a 22 nm process. Based on Intel's cycle alternating new architectures and die shrinks every two years, it is currently assumed that this will be a shrink of Sandy Bridge.[35]

[edit] System requirements

[edit] Motherboard compatibility

Conroe, Conroe XE and Allendale all use Socket LGA775; however, not every motherboard is compatible with these processors.

Supporting chipsets are:

* Intel: 865PE/G/GV/G, 945P/PL/G/GZ/GC, P65P/G, 975X, P/G/Q965, Q963, 946GZ/PL; X3x, P3x, G3x, Q3x
* NVIDIA: nForce4 Ultra/SLI X16 for Intel, nForce 570/590 SLI for Intel and nForce 650i Ultra/650i SLI/680i LT SLI/680i SLI.
* VIA: P4M800, P4M800PRO, P4M890, P4M900, PT880 Pro/Ultra, PT890.
* SiS: 662, 671, 671fx, 672, 672fx
* ATI: Radeon Xpress 200 and CrossFire Xpress 3200 for Intel

See also: List of Intel chipsets

Although a motherboard may have the required chipset to support Conroe, some motherboards based on the above mentioned chipsets do not support Conroe. This is because all Conroe-based processors require a new power delivery feature set specified in Voltage Regulator-Down (VRD) 11.0 [1]. This requirement is a result of Conroe's significantly lower power consumption, compared to the Pentium 4/D CPUs it is replacing. A motherboard that has both a supporting chipset and VRD 11 supports Conroe processors, but even then some boards will need an updated BIOS to recognize Conroe's FID (Frequency ID) and VID (Voltage ID).

[edit] DDR2 memory modules

Unlike the previous Pentium 4 and Pentium D design, the Core 2 technology sees a greater benefit from memory running synchronously with the Front Side Bus (FSB). This means that for the Conroe CPUs with FSB of 1066 MT/s, the ideal memory speed is PC2-4200. In some configurations, using PC2-5300 can actually decrease performance. Only when going to PC2-6400 is there a significant performance increase. While expensive DDR2 memory models with tighter timings do improve performance, the difference in real world games and applications is negligible.[36]

Optimally, the memory bandwidth afforded should match the bandwidth of the FSB closely. The AGTL+ PSB used by all NetBurst processors as well as current and medium-term (pre-CSI) Core 2 processors provide a 64-bit data path. Current chipsets provide for two DDR or two DDR2 channels.
Matched processor and RAM ratings Processor Model DDR or DDR2 rating Memory channels Bandwidth
T5200 PC2-4200 (DDR2-533) Single channel 4.267 GB/s
T5n00 and T7n00 PC2-5300 (DDR2-667) Single channel 5.333 GB/s
Socket P T7n00 PC2-3200 (DDR2-400) Dual channel 6.400 GB/s
PC2-6400 (DDR2-800) Single channel 6.400 GB/s
E4n00/Pentium E21n0/Celeron 4n0 PC3200 (DDR400) or PC2-3200 (DDR2-400) Dual channel 6.400 GB/s
PC2-6400 (DDR2-800) Single channel and dual channel 6.400 GB/s
E6n00, E6n20, X6n00, Q6n00 and QX6n00 PC2-4200 (DDR2-533) Dual channel 8.533 GB/s
PC3-8500 (DDR3-1066) Single channel 8.533 GB/s
E6n50 PC2-5300 (DDR2-667) Dual channel 10.667 GB/s
PC3-10600 (DDR3-1333) Single channel 10.667 GB/s

On jobs requiring large amounts of memory access, the quad-core Core 2 processors can benefit significantly[37] from using a PC2-8500 memory, which runs exactly twice as fast as the FSB; this is not an officially supported configuration, but a number of motherboards offer it.

The Core 2 processor does not require the use of DDR2. While the Intel 975X and P965 chipsets require this memory, some motherboards and chipsets support both the Core 2 and DDR memory. When using DDR memory, performance may be reduced because of the lower available memory bandwidth.

[edit] Chip bugs

The Core 2 Memory management unit (MMU) does not operate as previously specified or implemented in previous generations of x86 hardware. This may cause problems, many of them serious security and stability issues, with existing operating system software. Intel's documentation states that their programming manuals will be updated "in the coming months" with information on recommended methods of managing the Translation Lookaside Buffer (TLB) for Core 2 to avoid issues, and admits that, "in rare instances, improper TLB invalidation may result in unpredictable system behavior, such as hangs or incorrect data."[38]

Among the issues noted:

* Write-protect or non-execute bit for a page table entry is ignored.
* Floating point instruction non-coherencies.
* Allowed memory corruptions outside of the range of permitted writing for a process by running common instruction sequences.

Intel errata AI65, AI79, AI43, AI39, AI90, AI99 are said to be particulary serious.

Among those who have noted the errata to be particularly serious are OpenBSD's Theo de Raadt[2] and DragonFly BSD's Matthew Dillon[3]. Taking a contrasting view was Linus Torvalds, calling the issue "totally insignificant", adding, "The biggest problem is that Intel should just have documented the TLB behavior better."[4]

[edit] Pricing

The pricing for various models of Core 2, in lots of 1000 to OEMs, at the time the processors were released, can be found in the list of Intel Core 2 microprocessors. It should be noted that these prices are what it costs system builders such as Dell and HP to stock Core 2 processors. There are no set MSRPs for Core 2 CPUs in the retail channel — prices at retailers are usually very close to the aforementioned prices, but are dependent on what the supplier is charging to stock these CPUs as well as supply and demand.

[edit] Nomenclature and abbreviations

With the release of the new Core 2 processor, the abbreviation C2 has come into common use, as well variants C2D (the present Core 2 Duo), and C2Q, C2E to refer to the Core 2 Quad and Core 2 Extreme processors respectively. C2QX stands for the Extreme-Editions of the Quad (QX6700 or QX6800).