The Big Processor Guide - Intel Cores
Intel is and has been the 800 lb. gorilla of computer processors. They've been in the microprocessor market since it was born (and they helped deliver it). Intel made the computer processor what it is today: the 8086 essentially established the architecture that would define every x86 processor for decades.
In addition to computer processors, Intel makes a broad variety of other chips. They are in fact the single largest manufacturer of chips on Earth.
Note: All Front Side Bus speeds are listed at the frequency of the clock that drives them; double-pumping is indicated by a "x2" and quad-pumping by an "x4". This page identifies sockets that support those features.
Contents
Desktop
Mainstream
Core i3
For when the Core i5 is still too expensive, but not by much.
Clarkdale: Like the Core i5 Clarkdale, but with a slightly slower uncore clock speed and no Turbo Boost, AES-NI, VT-d, or TXT.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Clarkdale | Core i3 | LGA1156 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.13 GHz | 3.06 GHz (540) |
Core i5
The cheaper (but still not cheap) counterpart to Core i7.
Clarkdale: The first Westmere part on the market, Clarkdale is notable mainly because it has an on-package (but not yet on-die) graphics processor. The integrated memory controller was relocated from the processor die to the die housing the GPU, making that die effectively a full northbridge. The two dice communicate through a QPI link of unspecified speed. We have chosen to describe this as a 32 nm part, although the die with the GPU is actually 45 nm. Unlike Core i5 Lynnfield, Core i5 Clarkdale supports Hyper-Threading. 4 MB of L3 cache shared by the cores. For whatever reason, model numbers ending in 1 lack VT-d and TXT but get a faster clock for the GPU. All models support AES-New Instructions and SSE4.2.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Clarkdale | Core i5 | LGA1156 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.4 GHz | 3.46 GHz (670) |
Lynnfield: Basically the same as the Core i7 Lynnfield, but with Hyper-Threading disabled.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Lynnfield | Core i5 | LGA1156 | 4 | 256 KB | 2.5 GT/s DMI | 45 nm | 2.13 GHz | 2.66 GHz (750) |
Core i7
The "next generation" replacing the Core 2 family, the Core i7 has one of the more nonsensical names to ever emerge out of Intel's marketing department.
Lynnfield: The more economical follow-up to Bloomfield, now with an on-die PCI-Express controller (providing 16 lanes perfect for connecting to a graphics processor... like Larrabee). Some things were cut: the ridiculously fast QPI link is gone, in favor of a slower DMI link to the chipset (since this link no longer needs to service the graphics processor, and there are no plans for a multiprocessor Lynnfield, this is adequate). The on-die memory controller has been trimmed down from 3 channels to 2. Still has 8 MB of L3 cache for the cores to share. Core i7 Lynnfield supports Hyper-Threading: Core i5 Lynnfield does not.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Lynnfield | Core i7 | LGA1156 | 4 | 256 KB | 2.5 GT/s DMI | 45 nm | 2.4 GHz | 2.93 GHz (870) |
Bloomfield: The first Core i7 product and the first product based on the Nehalem microarchitecture, Bloomfield makes up both the mainstream Core i7 and the Core i7 Extreme. 8 MB of L3 cache is shared between the cores. On the LGA1366 socket, Bloomfield supports QPI and sports a memory controller that runs at a lower frequency than the core clock (what has been dubbed the "uncore clock").
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Bloomfield | Core i7 | LGA1366 | 4 | 256 KB | 4.8 GT/s QPI | 45 nm | 2.13 GHz | 3.2 GHz (960) |
Core 2 Quad
As the name implies, the Core 2 Quad is a bigger brother to the Core 2 Duo, technologically similar but with 4 cores instead of 2.
Yorkfield: Penryn-family quad-core part, in effect a die shrink of Kentsfield. Still a "dual die" product. Support for SSE4.1.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Yorkfield | Core 2 Quad | T | 4 | 3 MB | 333 MHz x4 | 45 nm | 3.0 GHz (Q9650) |
Intel | Yorkfield | Core 2 Quad | T | 4 | 1.5 MB | 333 MHz x4 | 45 nm | 2.67 GHz (Q9400) |
Intel | Yorkfield | Core 2 Quad | T | 4 | 1 MB | 333 MHz x4 | 45 nm | 2.66 GHz (Q8400) |
Kentsfield: Two Conroe cores dropped into a single package. Like the Pentium D Presler, this is a "dual die" product.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Kentsfield | Core 2 Quad | T | 4 | 2 MB | 266 MHz x4 | 65 nm | 2.67 GHz (Q6700) |
Core 2 Duo
The return of the king. Core 2 Duo is a step back toward the Pentium III- lower power consumption, better performance.
Wolfdale: A 45 nm die shrink of Conroe, or a desktop adaptation of the laptop Core 2 Duo Penryn, however you prefer to think about it. Like Penryn, it adds SSE4.1 instructions. L2 cache is still shared between the cores, L1 cache is still not.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Wolfdale | Core 2 Duo | T | 2 | 3 MB | 333 MHz x4 | 45 nm | 3.16 GHz (E8500) |
Conroe: The big daddy, Conroe benefits from the strengths of the new Core microarchitecture as well as an increased Front Side Bus speed.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Conroe | Core 2 Duo | T | 2 | 1 MB | 266 MHz x4 | 65 nm | 2.13 GHz (E6400) |
Intel | Conroe | Core 2 Duo | T | 2 | 2 MB | 266 MHz x4 | 65 nm | 2.67 GHz (E6700) |
Intel | Conroe | Core 2 Duo | T | 2 | 2 MB | 333 MHz x4 | 65 nm | 3.0 GHz (E6850) |
Allendale: A design very similar to Conroe, but natively having less L2 cache. The E6300 and E6400 (which have 2 MB of L2 cache) were initally assumed to be Allendale products, but were in fact Conroe chips with half their L2 cache disabled. Confusingly, later E6300 and E6400 chips actually were Allendale products.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Allendale | Core 2 Duo | T | 2 | 1 MB | 200 MHz x4 | 65 nm | 2.2 GHz (E4500) |
Intel | Allendale | Core 2 Duo | T | 2 | 1 MB | 266 MHz x4 | 65 nm | 2.13 GHz (E6400) |
Pentium D
In essence, the Pentium D chips are simply two Pentium 4s in a single package. The idea is to provide the performance benefits of having two processors, but at a lower cost and with less bulk. The reality is somewhat less impressive.
Presler: Two Cedar Mill cores on one chip. Some have refered to this as "double core" rather than "dual core" because the cores are actually separately-manufactured Cedar Mills that are then placed into the same package. Supports EM64T, the No Execute bit, and Vanderpool. Originally slated to support Enhanced Speedstep, but stability concerns forced Intel to disable this feature on the initial run. Does not support Hyper-Threading.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Presler | Pentium D | T | 2 | 2 MB | 200 MHz x4 | 65 nm | 3.4 GHz (950) |
Smithfield: Essentially two Prescott cores on one chip. Supports EM64T and the No Execute bit; does not support Hyper-Threading or Vanderpool.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Smithfield | Pentium D | T | 2 | 1 MB | 200 MHz x4 | 90 nm | 3.2 GHz (840) |
Pentium 4
The long-running, long-suffering leading-edge consumer product from Intel, the Pentium 4 has had more faces than Joan Rivers. Some well-loved, some reviled, they are nothing if not diverse.
Cedar Mill: Die shrink of the Prescott 2M. Supports everything the Prescott 2M 672 does, including SSE1, 2, and 3, EM64T, No Execute, Hyper-Threading, and Vanderpool.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Cedar Mill | Pentium 4 | T | 1 | 2 MB | 200 MHz x4 | 65 nm | 3.6 GHz (661) |
Prescott 2M: A revised Prescott bearing 2 MB of cache (hence the change in the name). Like some previous Prescotts, all Prescott 2M chips support EM64T, the No Execute bit, and Hyper-Threading. The 662 and 672 (only) support Vanderpool virtualization. All others are named 6x0.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Prescott 2M | Pentium 4 | T | 1 | 2 MB | 200 MHz x4 | 90 nm | 3.8 GHz (672) |
Prescott "E": The Prescott was the result of Intel's long and troubled transition to a 90 nm manufacturing process. The Prescott chips were criticized immediately for performing worse than Northwood C chips at the same speeds and for their very high heat output. Intel promised that the chips would show their true colors as they scaled up in speed (and indeed, tests showed they did quite better at speeds of 3.4 GHz and above). All Prescotts support SSE3.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Prescott | Pentium 4 | 478 | 1 | 1 MB | 200 MHz x4 | 90 nm | 3.4 GHz |
Intel | Prescott | Pentium 4 | 478 | 1 | 1 MB | 133 MHz x4 | 90 nm | 2.8 GHz |
Intel | Prescott | Pentium 4 | T | 1 | 1 MB | 200 MHz x4 | 90 nm | 3.8 GHz (571) |
Intel | Prescott | Pentium 4 | T | 1 | 1 MB | 133 MHz x4 | 90 nm | 3.06 GHz (519) |
Northwood "C": The golden age of the Pentium 4. This core substantially outperformed previous Pentium 4s, due mainly to the substantial impact of Front Side Bus speed on NetBurst chips. The real accomplishment of the Northwood C, however, was its much-praised overclocking potential. Boasting the most impressive and consistent overclocking performance since the early-model Celerons, it would gain a reputation (particularly, the 2.4 GHz chip) as the undisputed king of overclocking. Only when AMD's Athlon XP Barton core (and subsequent Mobile variants) appeared would that crown show even a hint of tarnish.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Northwood C | Pentium 4 | 478 | 1 | 512 KB | 200 MHz x4 | 130 nm | 3.4 GHz |
Northwood "B": The first chips to have enabled Hyper-Threading, though it was exclusive to the 3.06 GHz version. Increased Front Side Bus speed would help give these chips a leg up over their predecessors, but they were otherwise rather unremarkable.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Northwood B | Pentium 4 | 478 | 1 | 512 KB | 133 MHz x4 | 130 nm | 3.06 GHz |
Northwood "A": The (much-requested) replacement for Willamette, this core ushered in the beginning of an era when Pentium 4 performance really could impress. With architectural changes, more cache, lower heat output, and resulting better scalability, the Northwood series is hands-down superior to Willamette. The "A" was affixed to model numbers to separate the Northwood from the Willamette; future revisions of the Pentium 4 would be distinguished by different letters.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Northwood A | Pentium 4 | 478 | 1 | 512 KB | 100 MHz x4 | 130 nm | 2.6 GHz |
Willamette: The bastard child. Willamette was mocked at its introduction because at the same clock speeds it would substantially under-perform older Pentium IIIs. Some of this stumbling was aleviated when Socket 423 died a quick death, but Willamette continued to be dogged by performance issues until it was finally phased out. Willamette added SSE2 extensions.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Willamette | Pentium 4 | 423 | 1 | 256 KB | 100 MHz x4 | 180 nm | 2.0 GHz |
Intel | Willamette | Pentium 4 | 478 | 1 | 256 KB | 100 MHz x4 | 180 nm | 2.0 GHz |
Pentium III
Considered by some to be the peak of Intel's processor prowess, the Pentium III continues to be a sought-after chip for all manner of moderate-duty tasks. In later years the architecture would be revived for the Pentium M, and its spiritual successor would appear in the form of Merom. The Pentium III was in many ways just an updated Pentium II, with changes such as a new card packaging that allowed better heatsink contact.
Tualatin: Tualatin is regarded by many as little more than a die shrink. It was eventually made available with increased L2 cache for improved performance.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tualatin | Pentium III | 370 | 1 | 256 KB | 133 MHz | 130 nm | 1.4 GHz |
Intel | Tualatin | Pentium III-S | 370 | 1 | 512 KB | 133 MHz | 130 nm | 1.4 GHz |
Coppermine: Coppermine's two greatest changes were a die shrink and substantially speeding up the L2 cache. Where on the Katmai core the L2 cache had operated at half the frequency of the processor, on the Coppermine it operated at the same frequency. The cache was also brought onto the die itself to facilitate the change (this, in turn, was facilitated by the die shrink).
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Coppermine | Pentium III | Slot 1 | 1 | 256 KB | 100 MHz | 180 nm | 1.0 GHz |
Intel | Coppermine | Pentium III | Slot 1 | 1 | 256 KB | 133 MHz | 180 nm | 1.13 GHz |
Intel | Coppermine | Pentium III-E | 370 | 1 | 256 KB | 100 MHz | 180 nm | 1.1 GHz |
Intel | Coppermine | Pentium III-EB | 370 | 1 | 256 KB | 133 MHz | 180 nm | 1.13 GHz |
Katmai: In truth only slightly different from the Pentium II, Katmai added SSE instructions and gained an improved memory controller for its L1 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Katmai | Pentium III | Slot 1 | 1 | 512 KB | 100 MHz | 250 nm | 600 MHz |
Intel | Katmi | Pentium III-B | Slot 1 | 1 | 512 KB | 133 MHz | 250 nm | 600 MHz |
Pentium II Overdrive
A Pentium II adapted for use in Socket 8 motherboard, the Pentium II Overdrive was meant to provide an upgrade path to Pentium Pro owners.
P6T: Pentium II Deschutes, but on Socket 8.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P6T | Pentium II Overdrive | 8 | 1 | 512 KB | 60 MHz | 250 nm | 300 MHz |
Intel | P6T | Pentium II Overdrive | 8 | 1 | 512 KB | 66 MHz | 250 nm | 333 MHz |
Pentium II
Based on the Pentium Pro P6, the Pentium II was Intel's first slot-and-cartridge type processor. This was a tradeoff decision that allowed engineers to give the processor locally available L2 cache on the same PCB as the processor, at less cost than integrating it into the die.
Deschutes: On a smaller process and available with a faster Front Side Bus, the Deschutes core was a relatively simple but substantial improvement over Klamath. Heat issues were greatly reduced by the die shrink.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Deschutes | Pentium II | Slot 1 | 1 | 512 KB | 66 MHz | 250 nm | 333 MHz |
Intel | Deschutes | Pentium II | Slot 1 | 1 | 512 KB | 100 MHz | 250 nm | 450 MHz |
Klamath: The first Pentium II dealt with substantial heat issues, and suffered under the limitations of its Front Side Bus speed.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Klamath | Pentium II | Slot 1 | 1 | 512 KB | 66 MHz | 350 nm | 300 MHz |
Pentium MMX
The Pentium chip, but with the addition of the MMX instruction set. Double the L1 cache.
P55C: Modified Pentium P54C.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P55C | Pentium MMX | 7 | 1 | 0 KB | 66 MHz | 350 nm | 233 MHz |
Pentium Overdrive
Ostensibly meant to provide an upgrade path for 486 owners wanting a taste of Pentium power, the Pentium Overdrive is largely a footnote in history. Performance issues and poor availability essentially mooted the product.
P54CTB: Essentially the Pentium MMX P55C, but adapted to Socket 5.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P54CTB | Pentium Overdrive | 5 | 1 | 0 KB | 60 MHz | 350 nm | 180 MHz |
Intel | P54CTB | Pentium Overdrive | 5 | 1 | 0 KB | 66 MHz | 350 nm | 200 MHz |
P54CT: Derived from the P54CS. Suffered from many of the same limitations as P5T, but had to also package in its own voltage regulator. 486 motherboards delivered a 5 V supply, and it needed 3.3 V.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P54CT | Pentium Overdrive | 5 | 1 | 0 KB | 50 MHz | 350 nm | 125 MHz |
Intel | P54CT | Pentium Overdrive | 5 | 1 | 0 KB | 60 MHz | 350 nm | 150 MHz |
Intel | P54CT | Pentium Overdrive | 5 | 1 | 0 KB | 66 MHz | 350 nm | 166 MHz |
P5T: Derived from the Pentium P5. Adapting it to work on (some) 486 motherboards was difficult; it needed more L1 cache to boost performance, but still suffered considerably.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P5T | Pentium Overdrive | 4 | 1 | 0 KB | 60 MHz | 350 nm | 120 MHz |
Intel | P5T | Pentium Overdrive | 4 | 1 | 0 KB | 66 MHz | 350 nm | 133 MHz |
Pentium
Intel's take on a "586", the Pentium was a move to create their own proprietary chips, to allow them to differentiate their brand. Where in the days of the 486 several companies made chips that bore the marker "486", only Intel could produce a Pentium. One notable benefit was that they could create separate server/workstation lines to segment the market. The Pentium brand was the central focus of an enormous and very successful marketing push by Intel, and became one of rather few computer-industry brands to truly become a household name.
P54CQS/P54CS: Die shrink of P54C.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P54CQS | Pentium | 5 | 1 | 0 KB | 60 MHz | 350 nm | 120 MHz |
Intel | P54CQS | Pentium | 5 | 1 | 0 KB | 66 MHz | 350 nm | 133 MHz |
Intel | P54CS | Pentium | 5 | 1 | 0 KB | 60 MHz | 350 nm | 150 MHz |
Intel | P54CS | Pentium | 5 | 1 | 0 KB | 66 MHz | 350 nm | 200 MHz |
P54C: The P54C was architecturally the same as the P5, but it ran on a lower voltage (3.3 V versus 5 V), was built on a smaller process, and was a pure CMOS device while the P5 contained some bipolar logic. The net effect was that it produced much less heat.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P54C | Pentium | 5 | 1 | 0 KB | 50 MHz | 600 nm | 100 MHz |
Intel | P54C | Pentium | 5 | 1 | 0 KB | 60 MHz | 600 nm | 90 MHz |
Intel | P54C | Pentium | 5 | 1 | 0 KB | 66 MHz | 600 nm | 100 MHz |
P5: The original Pentium, what some would have liked to see called the 80586 instead. The P5 offered substantial performance gains over the 486, and Intel chose to drop its price below the 486. However, it produced quite a bit of heat for a chip of its time. And in a great embarassment to Intel, some models contained an error that would in rare cases cause it make mistakes in floating point math- the FDIV bug. Intel was eventually forced to recall them to quell the public relations problem.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P5 | Pentium | 4 | 1 | 0 KB | 60 MHz | 800 nm | 60 MHz |
Intel | P5 | Pentium | 4 | 1 | 0 KB | 66 MHz | 800 nm | 66 MHz |
High End
Core i7 Extreme
Another iteration of the same concept, the Core i7 Extreme is very much like the regular Core i7, but with more power.
Gulftown: If you liked its Bloomfield predecessor but wanted more, you got it. 50% more cores (6) and 50% more L3 cache (12 MB), plus all the good stuff that comes with being a Westmere part (a die shrink, AES-NI).
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Gulftown | Core i7 Extreme | LGA1366 | 6 | 256 KB | 6.4 GT/s QPI | 32 nm | 3.3 GHz | 3.33 GHz (980X) |
Bloomfield: Basically the same as the Core i7 Bloomfield, but with an unlocked multiplier, faster QPI, faster uncore (memory controller) clock, and a more extreme name. Like the desktop part, it has 8 MB of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Bloomfield | Core i7 Extreme | LGA1366 | 4 | 256 KB | 6.4 GT/s QPI | 45 nm | 3.3 GHz | 3.33 GHz (975) |
Intel | Bloomfield | Core i7 Extreme | LGA1366 | 4 | 256 KB | 6.4 GT/s QPI | 45 nm | 3.2 GHz | 3.2 GHz (965) |
Core 2 Extreme
Bringing the Extreme Edition concept into the Core era.
Harpertown: The Xeon Harpertown, repurposed as an enthusiast part. A lot like Yorkfield, but with an even faster Front Side Bus. Intended for use with Intel's SkullTrail platform (which is a dual-processor setup), this Core 2 Extreme uses the socket normally reserved for Xeons rather than a typical desktop socket.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Harpertown | Core 2 Extreme | J | 4 | 3 MB | 400 MHz x4 | 45 nm | 3.2 GHz (QX9775) |
Yorkfield: Penryn adapted to the needs of xtreme gamer dudez. More L2 cache than previous Core 2 Extreme products, and a shiny new 45 nm process.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Yorkfield | Core 2 Extreme | T | 4 | 3 MB | 400 MHz x4 | 45 nm | 3.2 GHz (QX9770) |
Intel | Yorkfield | Core 2 Extreme | T | 4 | 3 MB | 333 MHz x4 | 45 nm | 3.0 GHz (QX9650) |
Kentsfield: Two Conroe cores dropped into a single package. Like Pentium D Presler, this is a "dual die" product.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Kentsfield | Core 2 Extreme | T | 4 | 2 MB | 266 MHz x4 | 65 nm | 2.93 GHz (QX6800) |
Intel | Kentsfield | Core 2 Extreme | T | 4 | 2 MB | 333 MHz x4 | 65 nm | 3.0 GHz (QX6850) |
Conroe XE: Functionally identical to Core 2 Duo Conroe, with the exception that its multiplier is not locked.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Conroe XE | Core 2 Extreme | T | 2 | 2 MB | 266 MHz x4 | 65 nm | 2.93 GHz (X6800) |
Pentium Extreme Edition
Keeping up the tradition of the Pentium 4 Extreme series, but now dual core.
Presler: As the Pentium D Presler, but with Hyper-Threading and a faster Front Side Bus.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Presler | Pentium EE | T | 2 | 2 MB | 266 MHz x4 | 65 nm | 3.73 GHz (965) |
Smithfield: As the Pentium D Smithfield, but with Hyper-Threading enabled.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Smithfield | Pentium EE | T | 2 | 1 MB | 200 MHz x4 | 90 nm | 3.2 GHz (840) |
Pentium 4 Extreme Edition
Marketed as suped-up Pentium 4s, but often just retuned Xeons, the P4EE series were pitched as the peak of the enthusiast segment.
Prescott 2M: The same core as used in the regular Pentium 4 series, but with a higher Front Side Bus speed. Supports Hyper-Threading, EM64T, and the No Execute bit.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Prescott 2M | Pentium 4 EE | T | 1 | 2 MB | 266 MHz x4 | 90 nm | 3.7 GHz |
Gallatin: Modified Xeon Gallatin, with 2 MB of L3 cache. Supports Hyper-Threading.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Gallatin | Pentium 4 EE | 478 | 1 | 512 KB | 200 MHz x4 | 130 nm | 3.4 GHz |
Intel | Gallatin | Pentium 4 EE | T | 1 | 512 KB | 200 MHz x4 | 130 nm | 3.4 GHz |
Intel | Gallatin | Pentium 4 EE | T | 1 | 512 KB | 266 MHz x4 | 130 nm | 3.46 GHz |
Value
Pentium
Apparently not satisfied with confusing the world with the "Pentium Dual-Core", Intel decided to reuse the Pentium name with no modifiers at all, in a move sure to confound people for years.
Clarkdale: Much like the Core i3 and i5 CLarkdale, but cut down just a bit more. No Hyper-Threading, less L3 cache (3 MB instead of 4 MB), slower uncore clock, an on-die GPU down-tuned to 533 MHz, and all of the restrictions of Core i3 Clarkdale.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Clarkdale | Pentium | LGA1156 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.0 GHz | 2.8 GHz (G6950) |
Wolfdale: Derived from the Core 2 Wolfdale, but with a slower FSB and reduced L2 cache. Just in case you thought there was any sanity in Intel's marketing department, one of the first products in this series was the Pentium E6300, which might easily be confused with the not-that-much-older Core 2 Duo E6300.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Wolfdale | Pentium | T | 2 | 1.5 MB | 266 MHz x4 | 45 nm | 2.8 GHz (E7400) |
Intel | Wolfdale | Pentium | T | 2 | 1 MB | 266 MHz x4 | 45 nm | 2.8 GHz (E6300) |
Pentium Dual-Core
Quietly introduced not long after the Pentium brand was supposedly retired, the Pentium Dual-Core is essentially a cut-down version of either the Core Duo or Core 2 Duo.
Allendale: Just like the Core 2 Duo Allendale, but with half the L2 cache. Confusingly, often found in laptops despite being identified as a desktop product by Intel.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Allendale | Pentium Dual-Core | T | 2 | 512 KB | 200 MHz x4 | 65 nm | 2.0 GHz (E2180) |
Celeron 400 series
Listed here because they were introduced after the Celeron D products, these are essentially cut-down Core 2 Duos.
Millville: A trimmed down, single-core version of Conroe. Often called Conroe-L. It is unclear which is the actual codename used by Intel. Please note that several Celeron M products use very similar model numbers.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Millville | Celeron 400 series | T | 1 | 512 KB | 200 MHz x4 | 65 nm | 2.0 GHz (440) |
Celeron D
The Celeron D ushered in an impressive recovery for the Celeron series. For years, the Celerons had lagged behind AMD's Durons (which ran at lower clock speeds, used less power, and were cheaper) in performance. And also for years, the Celerons had been designed completely independently of the flagship Pentium series. The Celeron D marked a return to the good old days of the Celeron, when it was basically a down-tuned Pentium.
Cedar Mill 512: A trimmed down Pentium 4 Cedar Mill.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Cedar Mill 512 | Celeron D | T | 1 | 512 KB | 133 MHz x4 | 65 nm | 3.6 GHz (365) |
Prescott: Essentially the same as the Pentium 4 Prescott, but with less L2 cache and a low Front Side Bus.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Prescott | Celeron D | 478 | 1 | 256 KB | 133 MHz x4 | 90 nm | 3.2 GHz (350) |
Intel | Prescott | Celeron D | T | 1 | 256 KB | 133 MHz x4 | 90 nm | 3.33 GHz (355) |
Celeron
The Celerons have taken many forms. Many Pentium cores have been recast as Celerons, usually by reducing their cache and sometimes also their Front Side Bus speeds. Some Celeron cores were designed separately (or at least, mostly separately). After the end of the Pentium III-based Celerons, there was a long dark age that ended only with the introduction of the Celeron D.
Northwood "A": Based on the Pentium 4 Northwood A, but with half the cache- thus it is often called Northwood-128.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Northwood | Celeron | 478 | 1 | 128 KB | 100 MHz x4 | 130 nm | 2.7 GHz |
Willamette: Similar to the Pentium 4 Willamette, but with half the L2 cache. Consequently, often called Willamette-128. Sometimes refered to as the Celeron 4 to distinguish it from previous designs.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Willamette | Celeron | 478 | 1 | 128 KB | 100 MHz x4 | 180 nm | 1.8 GHz |
Tualatin: As was the case for the Celeron Coppermine, the Celeron Tualatin was for all intents and purposes the same as its Pentium III big brother, with less L2 cache than the Pentium III-S version and a slower Front Side Bus. Reportedly, its L2 cache controller had slightly higher latency. The nickname "Tualeron" was coined as a portmanteau of Tualatin and Celeron.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tualatin | Celeron | 370 | 1 | 256 KB | 100 MHz | 130 nm | 1.4 GHz |
Coppermine: Essentially the same as the Pentium III core of the same name, but with reduced L2 cache and a limited Front Side Bus. Often refered to as "Coppermine-128" in honor of its lesser L2 cache, and sometimes called "Celeron II".
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Coppermine | Celeron | 370 | 1 | 128 KB | 66 MHz | 180 nm | 766 MHz |
Intel | Coppermine | Celeron | 370 | 1 | 128 KB | 100 MHz | 180 nm | 1.1 GHz |
Mendocino: Based on the Pentium II, Mendocino sought to rectify the poor performance of Covington by adding L2 cache. Mendocino had on-die L2 cache before the Pentium line did (contemporary Pentium IIs had off-die cache on their PCB). It was in this period that benchmarks began to reveal Celerons could actually outperform their more expensive Pentium brothers; Intel would soon take measures to put a stop to that. Some Mendocino models had names ending in "A" to separate them from Covington chips, and as a result some people continue to refer to them as the "Celeron A" series.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Mendocino | Celeron | Slot 1 | 1 | 128 KB | 66 MHz | 250 nm | 433 MHz |
Intel | Mendocino | Celeron | 370 | 1 | 128 KB | 66 MHz | 250 nm | 533 MHz |
Covington: Based on the Pentium II Deschutes, Covington was distinguished by the fact that it had no L2 cache. Its performance was consequently very poor, and it did not fair well in the market.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Covington | Celeron | Slot 1 | 1 | 0 KB | 66 MHz | 250 nm | 300 MHz |
Laptop
Mainstream
Core i5
Arrandale: Like the Core i7 Mobile Arrandale, but cut down a bit. 3 MB of L3 cache. Model numbers beginning with 3 do not have VT-d, AES-NI, or TXT.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Arrandale | Core i5 | BGA1288 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.4 GHz | 2.53 GHz (540M) |
Intel | Arrandale | Core i5 | PGA988 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.4 GHz | 2.53 GHz (540M) |
Core i3
Arrandale: Like the Core i5 Mobile Arrandale, but cut down a bit more. No Turbo Boost. No parts have VT-d, AES-NI, or TXT.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Arrandale | Core i3 | BGA1288 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.4 GHz | 2.26 GHz (350M) |
Intel | Arrandale | Core i3 | PGA988 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.4 GHz | 2.26 GHz (350M) |
Core i7 Extreme
As with the Core 2 Extreme, we are listing this here as a stop-gap measure. In any event, it is essentially identical to its non-extreme brother, but with an unlocked multiplier.
Clarksfield: As with Core i7 Mobile Clarksfield, but with an unlocked multiplier. 8 MB of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Clarksfield | Core i7 Extreme | PGA988 | 4 | 256 KB | 2.5 GT/s DMI | 45 nm | ? | 2.0 GHz (920XM) |
Core i7
Bringing the oomph of the Core i7 desktop products to laptops.
Arrandale: Based on the desktop Core i5 Clarkdale. Model numbers end in a 1- or 2-letter code that indicates TDP and speed of the GPU. Model numbers ending in E support ECC memory. Model numbers containing a U only support DDR3-800 memory, not DDR3-1066, and have a slower clock for the GPU. 4 MB of L3 cache. The uncore clock is our best guess based on the desktop part.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Arrandale | Core i7 | BGA1288 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.4 GHz | 2.66 GHz (620M) |
Intel | Arrandale | Core i7 | PGA988 | 2 | 256 KB | 2.5 GT/s DMI | 32 nm | 2.4 GHz | 2.66 GHz (620M) |
Clarksfield: The desktop Core i7 Lynnfield, adapted for use in laptops. Model numbers starting with 8 have 8 MB of L2 cache; those starting with 7 have 6 MB. Notice that Intel finally did the smart thing and put an 'M' on the end of model numbers to distinguish them from desktop parts. Hyper-Threading is supported. Intel has not disclosed uncore clock speeds at this time, but our best guess is 2.13 GHz.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Clarksfield | Core i7 | PGA988 | 4 | 256 KB | 2.5 GT/s DMI | 45 nm | ? | 1.73 GHz (820QM) |
Core 2 Extreme
Listed here because it is the first "high end" laptop part we have had to add, the mobile Core 2 Extreme is essentially a mobile Core 2 Duo with an unlocked multiplier.
Penryn XE QC: Core 2 Quad Penryn with an unlocked multiplier.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Penryn XE QC | Core 2 Extreme | P | 4 | 3 MB | 266 MHz x4 | 45 nm | 2.53 GHz (QX9300) |
Penryn XE : Core 2 Duo Penryn with an unlocked multiplier.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Penryn XE | Core 2 Extreme | P | 2 | 3 MB | 266 MHz x4 | 45 nm | 3.06 GHz (X9100) |
Intel | Penryn XE | Core 2 Extreme | P | 2 | 3 MB | 200 MHz x4 | 45 nm | 2.8 GHz (X9000) |
Merom XE: Merom "Refresh" with an unlocked multiplier.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Merom XE | Core 2 Extreme | P | 2 | 2 MB | 200 MHz x4 | 65 nm | 2.8 GHz (X7900) |
Core 2 Quad
Once again a mobile version of the same-named desktop part.
Penryn: Sometimes called Penryn QC (for quad core), this is essentially 2 regular Penryn dies in a "dual die" configuration.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Penryn | Core 2 Quad | P | 4 | 3 MB | 266 MHz x4 | 45 nm | 2.26 GHz (Q9100) |
Intel | Penryn | Core 2 Quad | P | 4 | 1.5 MB | 266 MHz x4 | 45 nm | 2.0 GHz (Q9000) |
Core 2 Duo
Mobile counterpart to the desktop series of the same name, similarly based on the Core microarchitecture.
Penryn: A die shrink of Merom, with the addition of the SSE4.1 instructions.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Penryn | Core 2 Duo | P | 2 | 3 MB | 266 MHz x4 | 45 nm | 3.06 GHz (T9900) |
Intel | Penryn | Core 2 Duo | P | 2 | 3 MB | 200 MHz x4 | 45 nm | 2.6 GHz (T9500) |
Intel | Penryn | Core 2 Duo | P | 2 | 1.5 MB | 233 MHz x4 | 45 nm | 2.66 GHz (P8800) |
Intel | Penryn | Core 2 Duo | P | 2 | 1.5 MB | 200 MHz x4 | 45 nm | 2.4 GHz (T8300) |
Merom: Conroe with lower power dissipation. The parts with a Front Side Bus speed of 200 MHz x4 (or 800 MT/s) are often refered to as "Merom Refresh", and sometimes incorrectly as Santa Rosa (they are a part of the Santa Rosa version of the Centrino Pro platform). Variants with 2 MB of total L2 cache (1 MB per core) are sometimes unofficially called "Merom 2M".
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Merom | Core 2 Duo | M | 2 | 1 MB | 133 MHz x4 | 65 nm | 1.73 GHz (T5300) |
Intel | Merom | Core 2 Duo | M | 2 | 1 MB | 166 MHz x4 | 65 nm | 1.83 GHz (T5600) |
Intel | Merom | Core 2 Duo | M | 2 | 2 MB | 166 MHz x4 | 65 nm | 2.33 GHz (T7600) |
Intel | Merom | Core 2 Duo | P | 2 | 1 MB | 200 MHz x4 | 65 nm | 2.0 GHz (T7250) |
Intel | Merom | Core 2 Duo | P | 2 | 2 MB | 200 MHz x4 | 65 nm | 2.6 GHz (T7800) |
Core 2 Solo
A single-core version of the Core 2 Duo.
Merom: Very much like the Core 2 Duo Merom, but with one core disabled and a reduced operating voltage.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Merom | Core 2 Solo | M | 1 | 1 MB | 133 MHz x4 | 65 nm | 1.2 GHz (U2200) |
Core Duo
Dual-core version of the Core Solo.
Yonah: Adds another processing core to the Core Solo Yonah. The two cores share the same total amount of L2 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Yonah | Core Duo | M | 2 | 1 MB | 166 MHz x4 | 65 nm | 2.16 GHz (T2600) |
Core Solo
Successor to the Pentium M legacy, the Core Solo is Intel's single-core laptop solution.
Yonah: Essentially a revised Pentium M, and confusingly not based on the Core microarchitecture. Incorporates SSE3, LaGrande, Vanderpool, and the NX bit. Does not support the EM64T instructions.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Yonah | Core Solo | M | 1 | 2 MB | 166 MHz x4 | 65 nm | 1.66 GHz (T1300) |
Pentium M
More performance at a lower clock speed, using less power, made the Pentium M series the most popular laptop processor at the dance. The Pentium M dramatically improved laptop battery life and firmly cemented Intel's dominance of the notebook market. The Pentium M also made use of deeper "sleep" states, and more importantly used them more effectively, contributing to its impressive power reduction.
Dothan: A better, faster Banias. With more cache, a smaller process, and lower power use, Dothan proved to be a nice step up from the already impressive Banias. Improved power management including deeper sleep states and active processor speed throttling.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Dothan | Pentium M | 479 | 1 | 2 MB | 100 MHz x4 | 90 nm | 2.1 GHz (765) |
Intel | Dothan | Pentium M | 479 | 1 | 2 MB | 133 MHz x4 | 90 nm | 2.26 GHz (78) |
Banias: Based on the architecture of the Pentium III, Banias came from Intel's design facility in Israel. It introduced a concept known as "Micro-Ops Fusion", whereby the micro-operations to be executed by the processor could be "fused" (that is, certain combinations would be combined into a single operation) before execution.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Banias | Pentium M | 479 | 1 | 1 MB | 100 MHz x4 | 130 nm | 1.7 GHz |
Pentium 4-M
Essentially a Pentium 4 with better power management, the Pentium 4-M series were stuck in the middle of cheap-and-pretty-good Pentium III laptops and Pentium M laptops with remarkable battery life.
Northwood: Based on the Pentium 4 Northwood. Those with 133 MHz Front Side Bus also have Hyper-Threading.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Northwood | Pentium 4-M | 478 | 1 | 512 KB | 100 MHz x4 | 130 nm | 2.6 GHz |
Intel | Northwood | Pentium 4-M | 478 | 1 | 512 KB | 133 MHz x4 | 130 nm | 3.2 GHz |
Mobile Pentium III
A mobile adaptation of the desktop chip.
Tualatin: Based on the Pentium III Tualatin. Slightly different series name. Adds deeper sleep state.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tualatin | Mobile Pentium III-M | 370 | 1 | 512 KB | 100 MHz | 130 nm | 900 MHz |
Intel | Tualatin | Mobile Pentium III-M | 370 | 1 | 512 KB | 133 MHz | 130 nm | 1.33 GHz |
Coppermine: Based on the Pentium III Coppermine.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Coppermine | Mobile Pentium III | 370 | 1 | 256 KB | 100 MHz | 180 nm | 1.0 GHz |
Mobile Pentium II
A mobile adaptation of the desktop chip.
Dixon: Brought the L2 cache on-die. Marketed under the branding "Pentium II Mobile PE", where PE stood for Performance Enhanced.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Dixon | Mobile Pentium II PE | BGA1 | 1 | 256 KB | 66 MHz | 250 nm | 400 MHz |
Intel | Dixon | Mobile Pentium II PE | µPGA1 | 1 | 256 KB | 66 MHz | 250 nm | 400 MHz |
Tonga: Based on the Pentium II Deschutes.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tonga | Mobile Pentium II | MMC1 | 1 | 512 KB | 66 MHz | 250 nm | 300 MHz |
Intel | Tonga | Mobile Pentium II | MMC2 | 1 | 512 KB | 66 MHz | 250 nm | 300 MHz |
Intel | Tonga | Mobile Pentium II | Mini-Cartridge | 1 | 512 KB | 66 MHz | 250 nm | 300 MHz |
Pentium MMX
A mobile adaptation of the desktop chip.
Tillamook: Version of the Pentium MMX P55C made for the mobile market. It used a lower voltage power supply, a smaller process, and had more L1 cache to boost performance. It was meant to be combined with a "mobile module" that provided, among other things, 512 KB of L2 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tillamook | Pentium MMX | 7 | 1 | 512 KB | 66 MHz | 250 nm | 300 MHz |
Value
Pentium Dual-Core
Quietly introduced not long after the Pentium brand was supposedly retired, the Pentium Dual-Core is essentially a cut-down version of either the Core Duo or Core 2 Duo.
Yonah: Just like the Core Duo Yonah, but with half the L2 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Yonah | Pentium Dual-Core | M | 2 | 512 KB | 133 MHz x4 | 65 nm | 1.86 GHz (T2130) |
Celeron M
This series is very much like the Pentium M, but lacks SpeedStep. Celeron M chips are otherwise essentially copies of their Pentium M (later Core 2) brothers, but with half as much L2 cache.
Merom-L: A revision to the Celeron M Merom. This codename definitely is used by Intel. Intel notes a "minor die size decrease", which has been speculated to mean that it is actually manufactured with reduced L2 cache, rather than simply having some L2 cache disabled. It may also be the case that the size reduction is related to the "Merom Refresh" release.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Merom-L | Celeron M | M | 1 | 1 MB | 133 MHz x4 | 65 nm | 1.6 GHz (520) |
Intel | Merom-L | Celeron M | P | 1 | 1 MB | 133 MHz x4 | 65 nm | 2.0 GHz (550) |
Merom: A reduced-cache version of the Core 2 Duo Merom. Sometimes described as "Merom 1024" due to its reduced cache, though we have no idea if this codename is used internally by Intel.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Merom | Celeron M | M | 1 | 1 MB | 133 MHz x4 | 65 nm | 1.73 GHz (530) |
Dothan: Also called Dothan-1024. Pentium M Dothan with reduced cache and no hope of a faster Front Side Bus.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Dothan | Celeron M | 479 | 1 | 2 MB | 100 MHz x4 | 90 nm | 1.7 GHz (390) |
Banias: Also called Banias-512. Like the Pentium M Banias, but with half the L2 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Banias | Celeron M | 479 | 1 | 1 MB | 100 MHz x4 | 130 nm | 1.5 GHz (340) |
Shelton: Also called Banias-0. Based on the Pentium M Banias, but deployed in a low-cost low-power solution Intel sells in some Asian and Latin American markets. As its name implies, it has no L2 cache. Available only in one speed.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Shelton | Celeron M | 479 | 1 | 0 KB | 100 MHz x4 | 130 nm | 1.0 GHz |
Mobile Celeron
Based on the Pentium II, III, or 4 mobiles, the Mobile Celeron is essentially just a cut-down version of those processors. It has never had features such as SpeedStep to reduce power use. L2 cache is generally reduced.
Northwood: Like the Pentium 4-M Northwood (which in turn is like the Pentium 4 Northwood), but with reduced L2 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Northwood | Mobile Celeron | 478 | 1 | 256 KB | 100 MHz x4 | 130 nm | 2.5 GHz |
Tualatin: Based on the Pentium III-M Tualatin (based on the Pentium III Tualatin).
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tualatin | Mobile Celeron | 370 | 1 | 256 KB | 100 MHz | 130 nm | 700 MHz |
Intel | Tualatin | Mobile Celeron | 370 | 1 | 256 KB | 133 MHz | 130 nm | 1.33 GHz |
Coppermine: Based on the Pentium III-M Coppermine (based on the Pentium III Coppermine).
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Coppermine | Mobile Celeron | 370 | 1 | 256 KB | 100 MHz | 130 nm | 900 MHz |
Intel | Coppermine | Mobile Celeron | 370 | 1 | 256 KB | 133 MHz | 130 nm | 933 MHz |
Mendocino: Based on the Celeron Mendocino (based on the Pentium II).
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Mendocino | Mobile Celeron | 370 | 1 | 128 KB | 66 MHz | 250 nm | 500 MHz |
Server
Multi-processor
Xeon MP
Xeon with support for more than two processors in one system.
Nehalem-EX: When you really, absolutely must have up to 256 Nehalem-based processors. Up to 24 MB of shared L3 cache. Hyper-Threading and Turbo Boost supported by most but not all models. Note that most parts are in the Xeon 7500 series, while a few are in the Xeon 6500 series; we suspect that the principle difference is that the 6500 series likely does not support systems with huge numbers of sockets.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Nehalem-EX 24M | Xeon MP | LGA1567 | 8 | 512 KB | 6.4 GT/s QPI | 45 nm | 2.26 GHz (X7560) |
Intel | Nehalem-EX 24M | Xeon MP | LGA1567 | 8 | 512 KB | 5.86 GT/s QPI | 45 nm | 1.86 GHz (L7555) | Intel | Nehalem-EX 18M | Xeon MP | LGA1567 | 8 | 512 KB | 6.4 GT/s QPI | 45 nm | 2.0 GHz (X7550) |
Intel | Nehalem-EX 18M | Xeon MP | LGA1567 | 6 | 512 KB | 6.4 GT/s QPI | 45 nm | 2.0 GHz (E7540) |
Intel | Nehalem-EX 18M | Xeon MP | LGA1567 | 6 | 512 KB | 5.86 GT/s QPI | 45 nm | 2.66 GHz (X7542) |
Intel | Nehalem-EX 12M | Xeon MP | LGA1567 | 6 | 512 KB | 5.86 GT/s QPI | 45 nm | 1.86 GHz (E7530) |
Intel | Nehalem-EX 18M | Xeon MP | LGA1567 | 4 | 512 KB | 4.8 GT/s QPI | 45 nm | 1.86 GHz (E7520) |
Intel | Nehalem-EX 12M | Xeon MP | LGA1567 | 4 | 512 KB | 4.8 GT/s QPI | 45 nm | 1.73 GHz (E6510) |
Dunnington: The last major Xeon release before the next Nehalem microarchitecture, Dunnington is apparently built on the "more is better" philosophy. Dunnington packs as many as 3 Penryn-like processors (a total of 6 cores) into one chip. L2 cache is shared between pairs of cores (but not between the pairs), while up to 16 MB of L3 cache is shared between all cores. Testing by Anandtech has suggested that the L3 cache runs at half the core clock speed. As evidenced by die shots, Dunnington is a "true"/"native" 6-core design, not a multi-chip-module one. Dunnington is the first product out of Intel's Bangalore, India, design team. Please note that in our table below, we have appended a number to the Core name to indicate the amount of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Dunnington 8M | Xeon MP | J | 4 | 1.5 MB | 266 MHz x4 | 45 nm | 2.13 GHz (E7420) |
Intel | Dunnington 12M | Xeon MP | J | 4 | 1.5 MB | 266 MHz x4 | 45 nm | 2.13 GHz (E7430/L7445) |
Intel | Dunnington 16M | Xeon MP | J | 4 | 1.5 MB | 266 MHz x4 | 45 nm | 2.4 GHz (E7440) |
Intel | Dunnington 12M | Xeon MP | J | 6 | 1.5 MB | 266 MHz x4 | 45 nm | 2.4 GHz (E7450) |
Intel | Dunnington 16M | Xeon MP | J | 6 | 1.5 MB | 266 MHz x4 | 45 nm | 2.66 GHz (X7460) |
Tigerton: Essentially an MP version of Clovertown. Tigerton is available in both dual- and quad-core variants. The dual-core variants are essentially equivalent to an MP Woodcrest.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tigerton | Xeon MP | J | 2 | 4 MB | 266 MHz x4 | 65 nm | 2.93 GHz (E7220) |
Intel | Tigerton | Xeon MP | J | 4 | 1 MB | 266 MHz x4 | 65 nm | 1.6 GHz (E7310) |
Intel | Tigerton | Xeon MP | J | 4 | 1.5 MB | 266 MHz x4 | 65 nm | 2.4 GHz (E7330) |
Intel | Tigerton | Xeon MP | J | 4 | 2 MB | 266 MHz x4 | 65 nm | 2.93 GHz (X7350) |
Tulsa: The end of the line for Netburst, Tulsa is essentially Dempsey with more L3 cache (the two cores share 4, 8, or 16 MB total), a more limited Front Side Bus, and MP rather than just DP support.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tulsa | Xeon MP | 604 | 2 | 1 MB | 166 MHz x4 | 65 nm | 3.5 GHz (7150N) |
Intel | Tulsa | Xeon MP | 604 | 2 | 1 MB | 200 MHz x4 | 65 nm | 3.4 GHz (7140M) |
Paxville MP: Multi-processor version of Paxville.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Paxville MP | Xeon MP | 604 | 2 | 1 MB | 166 MHz x4 | 90 nm | 2.66 GHz (7020) |
Intel | Paxville MP | Xeon MP | 604 | 2 | 2 MB | 166 MHz x4 | 90 nm | 3.0 GHz (7030) |
Intel | Paxville MP | Xeon MP | 604 | 2 | 1 MB | 200 MHz x4 | 90 nm | 2.8 GHz (7040) |
Intel | Paxville MP | Xeon MP | 604 | 2 | 2 MB | 200 MHz x4 | 90 nm | 3.0 GHz (7041) |
Potomac: Higher-end version of Cranford with 4 MB or 8 MB of L3 cache. For the purposes of this guide, the core name reflects the amount of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Potomac 4M | Xeon MP | 604 | 1 | 1 MB | 166 MHz x4 | 90 nm | 2.83 GHz |
Intel | Potomac 8M | Xeon MP | 604 | 1 | 1 MB | 166 MHz x4 | 90 nm | 3.33 GHz |
Cranford: Multi-processor version of Xeon DP Nocona.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Cranford | Xeon MP | 604 | 1 | 1 MB | 166 MHz x4 | 90 nm | 3.66 GHz |
Gallatin: Multi-processor version of the Xeon DP Prestonia. For the purposes of this guide, the core name reflects the amount of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Gallatin 1M | Xeon MP | 604 | 1 | 512 KB | 100 MHz x4 | 130 nm | 2.5 GHz |
Intel | Gallatin 2M | Xeon MP | 604 | 1 | 512 KB | 100 MHz x4 | 130 nm | 2.8 GHz |
Intel | Gallatin 3M | Xeon MP | 604 | 1 | 512 KB | 100 MHz x4 | 130 nm | 3.0 GHz |
Foster MP: As the Xeon DP Foster, but with support for more than 2 processors in one system. For the purposes of this guide, the core name reflects the amount of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Foster 512K | Xeon MP | 603 | 1 | 256 KB | 100 MHz x4 | 180 nm | 1.5 GHz |
Intel | Foster 1M | Xeon MP | 603 | 1 | 256 KB | 100 MHz x4 | 180 nm | 1.6 GHz |
Dual processor
Xeon DP
Xeon with support for up to two processors in one system. Older Xeons also supported DP use, but with the introduction of Pentium 4-based Xeons Intel began to differentiate between DP and MP.
Lynnfield: As with Bloomfield and others before it, this is the desktop Core i7 Lynnfield turned into a Xeon. Same feature set as Core i7 Lynnfield, except for the Xeon X3430 which, like the Core i5 Lynnfield, does not support Hyper-Threading. 8 MB of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Uncore clock | Top Speed |
Intel | Lynnfield | Xeon | LGA1156 | 4 | 256 KB | 2.5 GT/s DMI | 45 nm | 2.4 GHz | 2.93 GHz (X3470) |
Bloomfield: Continuing their tradition of turning mainstream parts into single-socket-only Xeons, Intel turned Bloomfield into the 3500 series of Xeons. And continuing our tradition of lumping such Xeons in with the dual-socket parts, we have listed them here. Essentially the same feature set as the desktop Bloomfield or, for that matter, Xeon Nehalem. 8 MB of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Bloomfield | Xeon | LGA 1366 | 4 | 256 KB | 4.8 GT/s QPI | 45 nm | 3.33 GHz (W3580) |
Westmere-EP: A die shrink of Nehalem or Gulftown brought to the server farm, however you prefer to think of it. More cores than Nehalem. Adds AES-NI, Turbo Boost, TXT, and the other goodies that come with Gulftown. 12 MB of shared L3. Parts with a W or X model number have faster QPI links than those with L or E model numbers In true Intel fashion, almost all (but not all) parts support Hyper-Threading and Turbo Boost; L5609 has neither, and is additionally cursed with an even slower QPI link.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Westmere-EP | Xeon | LGA 1366 | 6 | 256 KB | 6.4 GT/s QPI | 32 nm | 3.3 GHz (W5680) |
Intel | Westmere-EP | Xeon | LGA 1366 | 6 | 256 KB | 5.86 GT/s QPI | 32 nm | 2.26 GHz (L5640) |
Intel | Westmere-EP | Xeon | LGA 1366 | 4 | 256 KB | 6.4 GT/s QPI | 32 nm | 3.46 GHz (X5677) |
Intel | Westmere-EP | Xeon | LGA 1366 | 4 | 256 KB | 6.4 GT/s QPI | 32 nm | 2.13 GHz (L5630) |
Intel | Westmere-EP | Xeon | LGA 1366 | 4 | 256 KB | 4.8 GT/s QPI | 32 nm | 1.86 GHz (L5609) |
Nehalem: As usual for Intel, server parts were the last to move to a new microarchitecture. This is perhaps a bit ironic in this case, where Nehalem was actually the "main" product from which desktop and mobile variants were derived, and was the microarchitecture's namesake. The naming system is a bit complicated; see our description on the naming page.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Nehalem | Xeon | LGA 1366 | 4 | 256 KB | 6.4 GT/s QPI | 45 nm | 3.3 GHz (W5590) |
Intel | Nehalem | Xeon | LGA 1366 | 4 | 256 KB | 5.86 GT/s QPI | 45 nm | 2.53 GHz (E5540) |
Intel | Nehalem | Xeon | LGA 1366 | 4 | 256 KB | 4.8 GT/s QPI | 45 nm | 2.13 GHz (E5506) |
Intel | Nehalem | Xeon | LGA 1366 | 2 | 256 KB | 4.8 GT/s QPI | 45 nm | 1.86 GHz (E5502) |
Harpertown: 2 Wolfdales in one package, through the magic of Multi-Chip Modules. There are L, E, and X models; L models use the least power, X models the most. The power savings come from reduced core voltages and, in some cases, slower Front Side Bus speeds.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Harpertown | Xeon | J | 4 | 3 MB | 266 MHz x4 | 45 nm | 2.13 GHz (L5408) |
Intel | Harpertown | Xeon | J | 4 | 3 MB | 333 MHz x4 | 45 nm | 3.33 GHz (X5470) |
Intel | Harpertown | Xeon | J | 4 | 3 MB | 400 MHz x4 | 45 nm | 3.4 GHz (X5492) |
Wolfdale: Penryn for servers. More cache, and a 45 nm process. Adds SSE4.1. Please note that Intel lumps Wofldale into the "Xeon 5400" series, even though the model numbers start with 5200, due to its close relationship with Harpertown.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Wolfdale | Xeon | J | 2 | 3 MB | 266 MHz x4 | 45 nm | 1.86 GHz (E5205) |
Intel | Wolfdale | Xeon | J | 2 | 3 MB | 333 MHz x4 | 45 nm | 3.33 GHz (X5260) |
Intel | Wolfdale | Xeon | J | 2 | 3 MB | 400 MHz x4 | 45 nm | 3.5 GHz (X5270) |
Yorkfield: Just like the Core 2 Quad Yorkfield. Continuing in the same spirit as the Xeon Kentsfield, this is a single-processor-only product.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Yorkfield | Xeon | T | 4 | 1.5 MB | 333 MHz x4 | 45 nm | 2.5 GHz (X3320) |
Intel | Yorkfield | Xeon | T | 4 | 3 MB | 333 MHz x4 | 45 nm | 2.83 GHz (X3360) |
Kentsfield: The same as Core 2 Quad Kentsfield, but with different branding. As with Conroe, not actually a DP product.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Kentsfield | Xeon | T | 4 | 2 MB | 266 MHz x4 | 65 nm | 2.67 GHz (X3230) |
Clovertown: The big-server analogue to Kentsfield, Clovertown is also a quad core "dual die" product. Supports Demand-Based Switching. There is some indication that faster versions of the lower-FSB Clovertown exist, but nothing official from Intel and no really solid information.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Clovertown | Xeon DP | J | 4 | 2 MB | 266 MHz x4 | 65 nm | 1.86 GHz (E5320) |
Intel | Clovertown | Xeon DP | J | 4 | 2 MB | 333 MHz x4 | 65 nm | 3.0 GHz (X5365) |
Allendale: Just like Conroe, Allendale was pulled over to serve double-duty as a Xeon. And just like Conroe, these Xeons are really only single-processor parts.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Conroe | Xeon | T | 2 | 2 MB | 266 MHz x4 | 65 nm | 2.13 GHz (3050) |
Conroe: The desktop chip adapted for use in smaller servers. Please note this is actually not a DP product (it can only be used in single-socket systems), but is placed here to put it in proper context.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Conroe | Xeon | T | 2 | 2 MB | 266 MHz x4 | 65 nm | 2.13 GHz (3050) |
Intel | Conroe | Xeon | T | 2 | 4 MB | 266 MHz x4 | 65 nm | 2.67 GHz (3070) |
Woodcrest: The first chip released with Intel's Core microarchitecture. Better performance and lower power than older Xeons.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Woodcrest | Xeon DP | J | 2 | 2 MB | 266 MHz x4 | 65 nm | 1.87 GHz (5120) |
Intel | Woodcrest | Xeon DP | J | 2 | 2 MB | 333 MHz x4 | 65 nm | 3.0 GHz (5160) |
Sossaman: Server analog to the Core laptop chips, also derived from the Pentium M and also using Socket 479. Built for low power use.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Sossaman | Xeon LV | M | 2 | 1 MB | 166 MHz x4 | 65 nm | 2.0 GHz |
Dempsey: Dual die, Hyperthreading, 64-bit. Die shrink. New socket.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Dempsey | Xeon DP | J | 2 | 2 MB | 166 MHz x4 | 65 nm | 3.0 GHz (5050) |
Intel | Dempsey | Xeon DP | J | 2 | 2 MB | 266 MHz x4 | 65 nm | 3.7 GHz (5080) |
Paxville: Dual core Irwindale, including Hyper-Threading.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Paxville | Xeon DP | 604 | 2 | 2 MB | 200 MHz x4 | 90 nm | 2.8 GHz |
Irwindale: Based on the Pentium 4 Prescott 2M, Irwindale adds the No Execute bit and more L2 cache to Nocona.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Irwindale | Xeon DP | 604 | 1 | 2 MB | 200 MHz x4 | 90 nm | 3.8 GHz |
Nocona: Based on the Pentium 4 Prescott, Nocona brings SSE3 and EM64T instructions to the Xeon family.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Nocona | Xeon DP | 604 | 1 | 1 MB | 200 MHz x4 | 90 nm | 3.6 GHz |
Gallatin: Prestonia with either 1 MB or 2 MB of L3 cache. For the purposes of this guide, the core name reflects the amount of L3 cache.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Gallatin 1M | Xeon DP | 604 | 1 | 512 KB | 133 MHz x4 | 130 nm | 3.2 GHz |
Intel | Gallatin 2M | Xeon DP | 604 | 1 | 512 KB | 133 MHz x4 | 130 nm | 3.2 GHz |
Prestonia: Based on the Pentium 4 Northwood. Similar to Foster, but with support for Hyper-Threading, a die shrink, and reduced operating voltage.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Prestonia | Xeon DP | 603 | 1 | 512 KB | 100 MHz x4 | 130 nm | 3.0 GHz |
Intel | Prestonia | Xeon DP | 604 | 1 | 512 KB | 133 MHz x4 | 130 nm | 3.06 GHz |
Foster: Based on the Pentium 4 Willamette.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Foster | Xeon DP | 603 | 1 | 256 KB | 100 MHz x4 | 180 nm | 2.0 GHz |
Xeon
Following in the footsteps of the Pentium Pro, the Xeon is marketed toward the server and workstation markets. Xeons have been produced based on the same architectures as much of the Pentium line.
Cascades: Another Pentium III Xeon, initially criticized for the fact that its low L2 cache made it indisitinguishable from ordinary Pentium IIIs of its day. Later versions increased the cache but also descreased the Front Side Bus, for a net performance improvement.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Cascades | Xeon | Slot 2 | 1 | 256 KB | 133 MHz | 180 nm | 1 GHz |
Intel | Cascades | Xeon | Slot 2 | 1 | 1 MB | 100 MHz | 180 nm | 700 MHz |
Intel | Cascades | Xeon | Slot 2 | 1 | 2 MB | 100 MHz | 180 nm | 900 MHz |
Tanner: A Pentium III Xeon, this core was essentially Drake with the same changes that appeared in the Katmai Pentium III.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Tanner | Xeon | Slot 2 | 1 | 512 KB | 100 MHz | 250 nm | 550 MHz |
Intel | Tanner | Xeon | Slot 2 | 1 | 1 MB | 100 MHz | 250 nm | 700 MHz |
Intel | Tanner | Xeon | Slot 2 | 1 | 2 MB | 100 MHz | 250 nm | 900 MHz |
Drake: Based on the Pentium Pro P6, with full-speed off-die L2 cache, and released as the Pentium II Xeon. Added MMX instructions, which the P6 lacked.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Drake | Xeon | Slot 2 | 1 | 512 KB | 100 MHz | 250 nm | 450 MHz |
Intel | Drake | Xeon | Slot 2 | 1 | 1 MB | 100 MHz | 250 nm | 450 MHz |
Intel | Drake | Xeon | Slot 2 | 1 | 2 MB | 100 MHz | 250 nm | 450 MHz |
Pentium Pro
The "professional" version of the Pentium, the Pentium Pro was marketed to server and workstation markets with claims of improved performance in such tasks.
P6: Benefiting from valuable (but costly) built-in L2 cache, the P6 was powerful for its time but faced substantial yield problems. L2 cache added greatly to the transistor count of the chip, and therefore made it more expensive and more difficult to produce.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | P6 | Pentium Pro | 8 | 1 | 256 KB | 60 MHz | 350 nm | 180 MHz |
Intel | P6 | Pentium Pro | 8 | 1 | 256 KB | 66 MHz | 350 nm | 200 MHz |
Intel | P6 | Pentium Pro | 8 | 1 | 512 KB | 66 MHz | 350 nm | 200 MHz |
Intel | P6 | Pentium Pro | 8 | 1 | 1 MB | 66 MHz | 350 nm | 200 MHz |
Embedded
Mainstream
Atom
Atom is Intel's attempt to move x86 into the embedded space, similar to AMD's Geode.
Pineview: Pineview is notable mainly for its integrated GPU (GMA 3150) and memory controller, which improved power use but not performance. Part of the Pinetrail platform. Intel describes two versions of Pineview: "Pineview-M", intended for netbooks; and "Pineview-D", intended for desktops; Pineview-M has a lower TDP (probably because Pineview-M gets Enhanced SpeedStep) and a less-capable memory controller. Model numbers starting with N are Pineview-M (why not M?), model numbers starting with D are Pineview-D. No support for Intel Virtualization. Model numbers of the form x5xx have 2 cores; model numbers of the form x4xx have 1 core. Unlike Atom 330, dual-core Pineview is a "true" dual-core part, with both cores on the same die. Intel describes the socket as FCBGA559 or FC-BGA8.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Pineview | Atom | FCBGA559 | 2 | 512 KB | DMI | 45 nm | 1.66 GHz (D510) |
Intel | Pineview | Atom | FCBGA559 | 1 | 512 KB | DMI | 45 nm | 1.83 GHz (N470) |
Silverthorne: Something of a throwback to the days of the original Pentium, Silverthorne is Intel's first in-order (versus out-of-order) design in years. It is the combination of the Bonnell core, L2 cache, and an external bus interface. It is otherwise designed to support the same ISA as Merom, including SSE instructions up to SSSE3 and Intel Virtualization Technology. All versions use a surface-mount BGA437 (also called uFCBGA8) package; no sockets.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Silverthorne | Atom | BGA437 | 2 | 512 KB | 133 MHz x4 | 45 nm | 1.6 GHz (330) |
Intel | Silverthorne | Atom | BGA437 | 1 | 512 KB | 133 MHz x4 | 45 nm | 1.6 GHz (230) |
Diamondville: Diamondville is an even smaller, even lower power version of Silverthorne intended for what Intel calls "Mobile Internet Devices". While Silverthorne uses a standard GTL+ FSB, like most Intel processors, Diamondville uses a lower-power CMOS bus. This saves power, but makes Diamondville incompatible with any chipset except the Poulsbo chipset (UL11/US15), which was developed specifically for it. Diamondville does not support Intel's 64-bit extensions, but generally supports the same ISA as Silverthorne. All but the Z510 support Symmetric Multi-Threading (like the Pentium 4's Hyper-Threading). All versions use a surface-mount BGA package; no sockets. The initial and (when present) ending letter in the model number provide information on the package. The Z515 supports "Burst Mode" (similar to Turbo Boost), which allows it to increase its clock speed to 1.2 GHz for brief periods (the "regular" clock speed has not been specified yet), but not Hyper-Threading.
Manufacturer | Core | Series | Socket | Cores | L2 Cache/Core | FSB | Process Size | Top Speed |
Intel | Diamondville | Atom | BGA437 | 1 | 512 KB | 166 MHz x4 | 45 nm | 1.67 GHz (N280) |
Intel | Diamondville | Atom | BGA441 | 1 | 512 KB | 133 MHz x4 | 45 nm | 2.0 GHz (Z550) |
Intel | Diamondville | Atom | BGA441 | 1 | 512 KB | 100 MHz x4 | 45 nm | 1.2 GHz (Z515) |
Intel | Diamondville | Atom | BGA437 | 1 | 512 KB | 133 MHz x4 | 45 nm | 1.6 GHz (Z530P) |
Intel | Diamondville | Atom | BGA437 | 1 | 512 KB | 100 MHz x4 | 45 nm | 1.1 GHz (Z510P) |