Jump to content

Nehalem (microarchitecture): Difference between revisions

From Wikipedia, the free encyclopedia
Content deleted Content added
Added because http://www.tcmagazine.com/comments.php?shownews=24467&catid=2 shows LGA 1155
LGA-1155
Line 359: Line 359:
| rowspan=4 | Desktop
| rowspan=4 | Desktop
| rowspan=2 | 4 (8)
| rowspan=2 | 4 (8)
| rowspan=4 | LGA-1156/LGA 1155
| rowspan=4 | LGA-1156/LGA-1155
| rowspan=4 |
| rowspan=4 |
|
|

Revision as of 22:17, 8 June 2009

Intel CPU core roadmaps from NetBurst to Sandy Bridge

Nehalem is the codename for an Intel processor microarchitecture,[1] successor to the Core microarchitecture. The first processor released with the Nehalem architecture is the desktop Core i7,[2] which was released on November 15, 2008 in Tokyo and November 17, 2008 in the USA.[3] The first computer to use Nehalem-based Xeon processors was the Mac Pro workstation announced on March 3, 2009.[4] Nehalem-based Xeon EX processors for larger servers are expected in Q4 2009.[5] Mobile Nehalem-based processors will follow in 2010.

Initial Nehalem processors use the same 45 nm manufacturing methods as Penryn. A working system with two Nehalem processors was shown at Intel Developer Forum Fall 2007,[6] and a large number of Nehalem systems were shown at Computex in June 2008.

The architecture is named after the Nehalem River in Northwest Oregon,[citation needed] which is in turn named after the Nehalem Native American nation in Oregon.[citation needed] The code name itself had been seen on the end of several roadmaps starting in 2000. At that stage it was supposed to be the latest evolution of the NetBurst architecture. Since the abandonment of NetBurst, the codename has been recycled and refers to a completely different project.

Technology

Microarchitecture of the quad-core implementation

Various sources have stated the specifications of processors in the Nehalem family:

  • Two, four, six, or eight cores
  • 45 nm manufacturing process
  • Integrated memory controller supporting two or three memory channels of DDR3 SDRAM or four FB-DIMM channels
  • Integrated graphics processor (IGP) located off-die, but in the same CPU package[7]
  • A new point-to-point processor interconnect, the Intel QuickPath Interconnect, replacing the legacy front side bus
  • Simultaneous multithreading (SMT) by multiple cores which enables two threads per core. Intel calls this hyperthreading. Simultaneous multithreading has not been present on a consumer desktop Intel processor since 2006 with the Pentium 4 and Pentium XE. Intel reintroduced SMT with their Atom Architecture.
  • Native (monolithic, i.e. all processor cores on a single die) quad- and octo-core processors[8]
  • The following caches:
    • 32 KB L1 instruction and 32 KB L1 data cache per core
    • 256 KB L2 cache per core
    • 2–3 MB L3 cache per core shared by all cores
  • 33 % more in-flight micro-ops than Conroe[9]
  • Second-level branch predictor and second-level Translation Lookaside Buffer[9]
  • Modular blocks of components such as cores that can be added and subtracted for varying market segments[10]

Performance and power improvements

It has been reported that Nehalem will have a focus on performance, which accounts for the increased core size.[11] Compared to Penryn, Nehalem will have:

  • 1.1x to 1.25x the single-threaded performance or 1.2x to 2x the multithreaded performance at the same power level
  • 30 % lower power usage for the same performance
  • According to a preview from AnandTech "expect a 20–30 % overall advantage over Penryn with a 10 % increase in power usage. It looks like Intel is on track to delivering just that in Q4."[12]
  • Core-wise, clock-for-clock, Nehalem will provide a 15–20 % increase in performance compared to Penryn.[13]

PC Watch found that a Nehalem "Gainestown" processor has 1.6x the SPECint_rate2006 integer performance and 2.4x the SPECfp_rate_2006 floating-point performance of a 3.0 GHz Xeon X5365 "Clovertown" quad-core processor.[11]

A 2.93 GHz Nehalem "Bloomfield" system has been used to run a 3DMark Vantage benchmark and gave a CPU score of 17,966.[14] The 2.66 GHz variant scores 16,294. A 2.4 GHz Core 2 Duo E6600 scores 4,300.[15]

AnandTech tested the Intel QuickPath Interconnect ("QPI", 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.[16]

Overclocking will be possible with Bloomfield processors and the X58 chipset. The Lynnfield processor will use a PCH removing the need for a northbridge chipset.[17]

The Nehalem processors are the first to incorporate the SSE 4.2 SIMD instructions, adding 7 new instructions to the SSE 4.1 set available in the Core 2 series.

Variants

Template:Future chip

These tables list all the processors of Nehalem microarchitecture to have been leaked so far. The table is ordered roughly by descending performance, which usually means descending price and power too. Released processors are set in bold.

Notes:

Codename Market Cores
(Threads)
Socket Brand Processor No. Clock rate Turbo TDP Interfaces L3 cache Release 1k Unit Price
Base Core Uncore Chipset Memory PCIe
Beckton MP server 8 (16) LGA-1567 130/105/90 W 4x QPI 4x [DDR3 with SMB motherboard] n/a 24 MB Q1 2010[18]
Gainestown DP server [19] 4 (8) LGA-1366 Xeon[20] W5580 133 MHz 3.2 GHz Yes 130 W 2x QPI 6.4 GT/s 3x DDR3 1333 MT/s [Note 1] n/a 8 MB 29 Mar 2009[21] $1600
X5570, X5560, X5550 2.93, 2.8, 2.66 GHz 95 W $1386, $1172, $958
E5540, E5530, E5520 2.53, 2.4, 2.26 GHz 80 W 2x 5.86 GT/s 3x DDR3 1066 MT/s [Note 1] $744, $530, $373
L5520, L5518 2.26 GHz, 2.13 GHz 60 W
4 (4) E5506 2.13 GHz No 80 W 2x 4.80 GT/s 3x DDR3 800 MT/s [Note 1] 4 MB $266
L5506 2.13 GHz 60 W
E5504 2 GHz 80 W $224
2 (2) E5502 1.86 GHz $188
Bloomfield UP server [22] 4 (8) LGA-1366 Xeon [23] W3570 133 MHz 3.2 GHz Yes 130 W 1x QPI 6.4 GT/s 3x DDR3 800–1333 MT/s n/a 8 MB 29 Mar 2009 [23] $999
W3540 2.93 GHz 1x QPI 4.8 GT/s 3x DDR3 800–1066 MT/s $562
W3520 2.66 GHz $284
Bloomfield High Performance Desktop [24] 4 (8) LGA-1366 Core i7 Extreme 975 [25] 133 MHz 3.33 GHz Yes 130 W 1x QPI 6.4 GT/s 3x DDR3 800–1066 MT/s (Unlocked since release, supporting XMP) n/a 8 MB 31 May 2009 $999
965 3.20 GHz 2.66 GHz 17 Nov 2008
Core i7 950 [25] 3.06 GHz 1x QPI 4.8 GT/s (Unlocked to 6.4 GT/s since release) 31 May 2009 $562
940 2.93 GHz 2.13 GHz 17 Nov 2008
920 2.66 GHz $284
Lynnfield Desktop 4 (8) LGA-1156/LGA-1155 133 MHz 2.93 GHz Yes 95 W 2x/4x DMI 2x DDR3 1066–1333 MT/s 1x16 / 2x8 8 MB Q3 2009 $562
2.8 GHz $284
4 (4) 2.66 GHz $196
4 (8) 2.13 GHz 65 W or 45 W [2] Q1 2010
Clarksfield Mobile 4 (8) mPGA-989 2.0 GHz 35 W [3] or 45/55 W [4] 2x/4x DMI 1x16 / 2x8 8 MB Q3 2009 $1054
1.73 GHz $546
1.6 GHz 6 MB $364

[Note 1] Though there is only one memory controller and it has only three channels, Intel states the Gainestown processors have six memory channels. Gainestown processors have dual QPI links and have a separate set of memory registers for each link[26] so thereby, in effect, a multiplexed six channel system.


The Havendale and Auburndale variants have been cancelled.[27]

Westmere (formerly Nehalem-C) is the name given to the 32 nm die shrink of Nehalem. Westmere should be ready for a late 2009 release provided that Intel stays on target with its roadmap. However, it appears that the bulk of Westmere's versions, excluding mobile versions, will be released sometime in 2010.[28][29] Westmere's features and improvements from Nehalem have been reported as follows:

  • Native six-core and possibly dual-die 16-core processors. [30]
    • The successor to Bloomfield and Gainestown is six-core.
  • A new set of instructions that gives over 3x the encryption and decryption rate of Advanced Encryption Standard (AES) processes compared to before.[31]
    • Delivers six new instructions that will be used by the AES algorithm. Also an instruction called PCLMULQDQ that will perform carry-less multiplication.[32] These instructions will allow the processor to perform hardware-accelerated encryption, not only resulting in faster execution but also protecting against software targeted attacks.
    • AES-NI may be included in the integrated graphics of Westmere.
  • integrated graphics, released at the same time as the processor.
  • Improved virtualization latency.[33]
  • New virtualization capability: "VMX Unrestricted mode support" -- which allows 16-bit guests to run. (real mode and big real mode)
Codename Market Cores
(Threads)
Socket Brand Processor No. Clock rate TDP Interfaces L3 cache Release 1k Unit Price
Base Core Uncore Chipset Memory PCIe
Gulftown[34] DP server, extreme desktop 6 (12) LGA-1366 130 W 2x QPI 3x DDR3 800–1066 MT/s n/a 12 MB Q2 2010[33] $999?
Clarkdale[35] Mainstream/value desktop 2 (4) LGA-1156 133 MHz 73 W 2x/4x DMI 2x DDR3 1066-1333 MT/s 1 x16/2 x8 4 MB Q4 2009[27] > $84 [5]
Arrandale[34] Mainstream/value mobile 2 (4) mPGA-989 18/25/35 W [6] 2x DDR3 800-1066 MT/s [7] 4 MB Q4 2009[27]

For the desktop, Gulftown is to be an "Extreme Edition" CPU and so will coexist with Bloomfield. [8] It will have Turbo Boost and similar clock speeds to Bloomfield. [9]

Lynnfield and Clarksfield may make the 32 nm transition in the middle of 2010[34], sometime after Q2 [10], while Beckton will move to 32 nm at the end of 2010.[36] The 32 nm CPUs will not have significantly different clock speeds compared to 45 nm CPUs.[37]

The integrated GPUs in Clarkdale and Arrandale are 45 nm, and support switchable graphics.[27][37] The lowest-power variant of Arrandale may have a 10 W CPU TDP, and a maximum clock speed of 1.6 GHz. [11]

A successor to Bloomfield and entry level server chips are also expected in Q2 2010.[27][27][34][36]

Successor

Template:Future chip

The successor to Nehalem and Westmere will be Sandy Bridge, scheduled for release in 2010, according to Intel roadmaps.[38] The successor to Sandy Bridge will be Haswell, scheduled for release in 2012. It will come with a new cache subsystem, a FMA (fused multiply-add) unit, and a vector coprocessor.[13]

See also

References

  1. ^ "http://www.intel.com/pressroom/archive/releases/20070328fact.htm" (Press release). Intel Corporation. 2007-03-28. {{cite press release}}: External link in |title= (help)
  2. ^ Gruener, Wolfgang (2008-08-10). "Nehalem = i7: Intel unveils new Core processor brand". TG Daily.
  3. ^ Hillis, Scott (2008-05-21). "Intel says Nehalem chips on track for '08 2nd half". Reuters. {{cite web}}: Unknown parameter |coauthor= ignored (|author= suggested) (help)
  4. ^ Crothers, Brooke (2009-03-03). "Apple beats Intel to Nehalem-EP chip launch".
  5. ^ Morgan, Timothy (2009-02-10). "Intel confirms Nehalem Xeons imminent".
  6. ^ Lal Shimpi, Anand (2007-09-18). "IDF 2007 – Day 1: Nehalem, Intel's GPUs, 32nm and More". AnandTech.
  7. ^ Laird, Jeremy (2007-09-19). "IDF: Not all Intel Nehalem CPUs single-die". Tech.co.uk.
  8. ^ Laird, Jeremy (2007-07-19). "IDF: Intel reveals eight-core PC processor". TechRadar UK.
  9. ^ a b Lal Shimpi (2008-03-17). "Opening the Kimono: Intel Details Nehalem and Tempts with Larrabee". AnandTech. {{cite web}}: More than one of |author= and |last= specified (help)
  10. ^ Gelsinger, Patrick P. (2008-03-17). "Intel Architecture Press Briefing" (PDF). Intel.
  11. ^ a b "Intel's dual teamed approached to micro-architecture development". PC Watch (in Japanese). 2008-01-29.
  12. ^ Lal Shimpi, Anand (2008-06-05). "The Nehalem Preview: Intel Does It Again". AnandTech.
  13. ^ a b Doc Teraboule (2008-04-08). "IDF 2008 Shanghaï – From Nehalem to Haswell" (in French). Canard PC. Retrieved 2009-03-01.
  14. ^ Nguyen, Tuan (2008-07-09). "'Nehalem' 2.93 GHz Benches Revealed". Tom's Hardware.
  15. ^ "Intel Bloomfield 2.66 GHz: First Comprehensive Evaluation". techPowerUp!. 2008-07-09.
  16. ^ "Intel does it again". AnandTech. 2008-06-05.
  17. ^ Botezatu, Bogdan (2008-04-22). "Intel: No Overclocking for Mainstream Nehalems". Softpedia.
  18. ^ Nebojsa Novakovic (2009-02-12). "Intel's next bunch of fun CPUs moves to 2010". CNN International. Retrieved 2009-03-01.
  19. ^ "Intel® Xeon® Processor 5500 Series Product brief, (Document Number: 321579-001US)" (PDF). Intel. 2009-04-01. Retrieved 2009-04-10.
  20. ^ "Core i7 to make leap to Xeon in early 2009". Legacy.macnn.com. Retrieved 2008-11-24.
  21. ^ Fuad Abazovic (2009-01-28). "Nehalem based Xeon comes Mar 29th". Fudzilla.com. Retrieved 2009-01-28.
  22. ^ "Intel® Xeon® Processor 3500 Series". Intel. 2009-04-01. Retrieved 2009-04-10.
  23. ^ a b Teglet, Traian (2008-11-13). "Intel to Add New Nehalem Xeon Processor". softpedia. Retrieved 2008-11-13. Cite error: The named reference "Intel to Add New Nehalem Xeon Processor" was defined multiple times with different content (see the help page).
  24. ^ "Intel® Core™ i7 Processor Extreme Datasheet, Volume 2, (Document Number: 320835-002)" (PDF). Intel. 2008-11-01. Retrieved 2009-03-24.
  25. ^ a b Worrel, Jon (2009-02-09). "Core i7 950 and 975 to replace 940 and 965". Fudzilla. Retrieved 2009-02-10.
  26. ^ "Intel® Xeon® Processor 5500 Series Datasheet Volume 2" (PDF). Intel. March 2009. Retrieved 2009-05-01.
  27. ^ a b c d e f Jansen, Ng (2009-02-10). "Intel Replaces "Havendale" and "Auburndale" With 32nm Die Shrinks". DailyTech. Retrieved 2009-03-01.
  28. ^ http://pc.watch.impress.co.jp/docs/2008/0326/kaigai02.pdf
  29. ^ "VISIO-Intel_CPU_Road Map_.vsd" (PDF). Retrieved 2008-11-24.
  30. ^ [1]
  31. ^ Smalley, Tim (2007-09-19). "Westmere is Nehalem's successor". bit-tech.net.
  32. ^ "Carry-Less Multiplication and Its Usage for Computing The GCM Mode – Intel® Software Network". 2008-04-11. Retrieved 2009-03-01.
  33. ^ a b Fuad Abazovic (2008-09-16). "Westmere 32nm to improve Nehalem features". Fudzilla.com. Retrieved 2009-03-01.
  34. ^ a b c d Ng, Jansen (10 February 2009). ""Gulftown" is the Flagship of 32nm "Westmere" Line". DailyTech. Retrieved 2009-02-10.
  35. ^ "Intel Clarkdale Processor". XTREVIEW. 2009-02-09. Retrieved 2009-03-01.
  36. ^ a b Ng, Jansen (2009-02-10). "Intel Aims for Efficiency With New Server Roadmap". DailyTech. Retrieved 2009-03-01.
  37. ^ a b Bell, Brandon (2009-02-10). "Intel CPU Roadmap 2009–2010". FS Media, Inc. Retrieved 2009-03-01.
  38. ^ "The Intel Tick-Tock Model of Architecture & Silicon Cadence". Intel Corporation.