Intel Core i7

Intel Core i7 is a family of three Intel desktop x86-64 processors, the first processors released using the Intel Nehalem microarchitecture and the successor to the Intel Core 2 family. All three models are quad-core processors.[1][2][3][4] The Core i7 identifier applies to the initial family of processors[5][6] codenamed Bloomfield.[7] Intel representatives state that the moniker Core i7 does not have any deeper meaning. The name continues the use of the successful Core brand.[8] Core i7 is manufactured in Costa Rica [9] and was officially launched on November 17, 2008.



Features
The Nehalema architecture has many new features, some of which are present in the Core i7. The ones that represent significant changes from the Core 2 include:

* The new LGA 1366 socket is incompatible with earlier processors.
* On-die memory controller: the memory is directly connected to the processor.
o Three channel memory: each channel can support one or two DDR3 DIMMs. Motherboards for Core i7 have four (3+1) or six DIMM slots instead of two or four, and DIMMs should be installed in sets of three, not two.
o Support for DDR3 only.
o No ECC support.
* The front side bus is replaced by QuickPath interface. Motherboards must use a chipset that supports QuickPath. As of 5 December 2008 (2008 -12-05)[update], Intel, EVGA, ASUS, MSI, Foxconn, Supermicro, and Gigabyte have all released X58 motherboards, all supporting the i7's LGA1366 Socket interface.
* Single-die device: all four cores, the memory controller, and all cache are on a single die.
* "Turbo Boost" technology allows all active cores to intelligently clock themselves up in steps of 133 MHz over the design clock rate as long as the CPU's predetermined thermal and electrical requirements are still met.[11] This mode isn't enabled when the CPU is manually over-clocked by the user.
* Re-implemented Hyper-threading. Each of the four cores can process two threads simultaneously, so the processor appears to the OS as eight CPUs. This feature was present in the older NetBurst architecture but was dropped in Core.
* On-die, shared, inclusive 8MB L3 cache.
* Only one QuickPath interface: not intended for multi-processor motherboards.
* 45nm process technology.
* 731M transistors.
* Sophisticated power management can place an unused core in a zero-power mode

Drawbacks
The Core i7 does not support error-correcting memory.
Some high-end motherboards that support the Core i7 advertise support for ECC memory, for example Supermicro's C7X58 and X8SAX (for example, see Supermicro's X8SAX page), however, in the same motherboard manuals, it is made clear that ECC is only supported if the CPU have the feature enabled.

Processors
* The clock rates listed here are as specified by Intel for normal mode. "Turbo boost" can increase the rate on active cores in steps of 133 MHz up to a predetermined limit for short periods when required.
* The 965 XE has separate unlocked multipliers for memory and cores.
o Core clock above those in the table are not guaranteed by Intel.[2] Rates above 5GHz have been reported.
o Memory rates above those in the table are not guaranteed by Intel.[2] Rates above DDR3-2000 have been reported.
* The processor has a Thermal Design Power of 130W and will slow itself down if this power is exceeded. This feature can be disabled from an option in most of the new motherboards' BIOS.[12]
* Prices are per unit in lots of 1,000 in USD

Performance
* In some benchmarks using overclocking techniques, core clocks of up to 4GHz have been reported, but are not supported by Intel[14]
o IT OC Taiwan overclocked the chip, which sports an unlocked multiplier, to a respectable 4.20 GHz, at a QPI speed of 200 MHz, and multiplier value of 21.0x. A vCore setting of 1.72 V was used, which is above the danger-mark, taking Intel's own warnings into account.[15]

A 2.93 GHz Core i7 940 system has been used to run a 3DMark Vantage benchmark and gave a CPU score of 17,966.[16] The 2.66 GHz Core i7 920 scores 16,294. An earlier generation Core processor, the very expensive 3.20 GHz Intel Core 2 Extreme QX9770, scores 13,182. Another one, the 2.66 GHz Core 2 Quad Q9450, scores 11,131.[17]

AnandTech tested the Intel QuickPath Interconnect (4.8 GT/s version) and found the copy bandwidth using triple-channel 1066 MHz DDR3 was 12.0 GB/s. A 3.0 GHz Core 2 Quad system using dual-channel 1066 MHz DDR3 achieved 6.9 GB/s.[18]

Over-clocking will be possible with the 900 series and a motherboard equipped with the X58 chipset. In early October 2008, reports surfaced that it will not be possible to use "performance" DDR3 DIMMs that require voltages higher than 1.65v, because the integrated memory controller within the Core i7 will be damaged.[19] Some tests, however, have demonstrated that the voltage limit does not apply, like on a MSI board, and manufacturers can choose to bond CPU voltage to memory or not. By the end of that month, performance memory vendors had announced 1.65v DDR3 memory kits with clock rates up to 2GHz.

Some early articles suggested that i7's design is not ideal for gaming performance. In a test done on leaked hardware, a Core i7 940 compared to a QX9770 shows the Core i7 is slower than Yorkfield clock for clock in 2 while being faster in the other two. The difference in all cases is small, and is due to the significantly smaller sized L2 cache on the processor cores, with each core able to access its own 256 kB of L2 cache. In contrast, the most recent Yorkfields have up to 12 MB of L2 cache. To help compensate, the Core i7 also has a new L3 cache of 8 MB, shared among all four cores, similar to AMD's "Barcelona" processors.[20] However, more recent testing done on all clock rates of official hardware with final drivers and BIOS revisions show that Core i7 at the very least beats Yorkfield clock-for-clock, and in most cases exceeds it by an average of about 17%.[21] But when it comes to high-end multi-GPU environments (Nvidia 3-way SLI and ATI Crossfire X), the i7 is revealed to be a lot faster than Yorkfield (QX9770) in clock-for-clock.[22] In the single-threaded Super PI 1M test, a Core i7 920 running at 2.66 GHz finished the test in 15.36 seconds, while a QX9770 (3.2 GHz) did the test in 14.42 seconds[23], so the Core i7 executed 20% more instructions per clock cycle on this test.

The Core i7 has three memory channels, and the channel bandwidth can be selected by setting the memory multiplier. However, in early benchmarks, when the clock rate is set higher than a threshold (1333 for the 965XE) the processor will only access two memory channels simultaneously. A 965XE has higher memory throughput with 3xDDR3-1333 than with 3xDDR3-1600, and 2xDDR3-1600 has almost identical throughput to 3xDDR3-1600.[14]

Since the Core i7 is a quad-core processor, Hyperthreading cannot yield any performance improvement for application workloads with fewer than five simultaneous threads when all the cores are fully powered on, and some applications will suffer a performance degradation when hyperthreading is enabled.[24] Hyperthreading will provide its best improvement when the workload has eight or more simultaneous threads.

Overclocking
The process of overclocking the Core i7 architecture is similar to that of the AMD architecture due to the on-die MCH. As of December 2008, the highest clockspeed attained with the Core i7 (965 XE) is 5510 MHz.[26]