Intel Core microarchitecture
Intel's 2006 multi-core CPU design that replaced NetBurst across desktops, laptops and servers
The Intel Core microarchitecture is a multi-core processor design that Intel launched in mid-2006, developed under the code name Merom. It replaced the power-hungry NetBurst design and carried Intel's desktop, mobile and server chips through the Core 2, Pentium, Celeron and Xeon brands.
- Launched mid-2006; Xeon on 26 June, Core 2 on 27 July
- 14-stage pipeline, under half the length of Prescott's
- Up to four instructions per cycle, against three for P6 and NetBurst
- Macro-Ops Fusion merges two x86 instructions into one micro-op
- Shrunk from 65 nm to 45 nm as the 2007/2008 Tick
What it is
Intel Core is a multi-core CPU microarchitecture that reached the market in mid-2006. It sits in the P6 line that began with the Pentium Pro in 1995, and it is a refreshed successor to the Enhanced Pentium M. It also took over from NetBurst, the design behind the Pentium 4, which drew too much power and ran too hot for the move to chips with two or more cores. The first chips built on it carried the code names Merom for laptops, Conroe for desktops and Woodcrest for servers and workstations. They appeared under the Core 2 name first, then spread to Pentium Dual-Core, Pentium, Celeron and Xeon.
How it came about
NetBurst reached a dead end in early 2004, when the Prescott core needed very high power to hit competitive clock speeds. Intel cancelled the next NetBurst parts, Tejas and Jayhawk, on 7 May 2004. Work on Merom, a 64-bit development of the Pentium M, had been under way since 2001, and Intel chose to extend it to every market instead of only mobile. The result kept the short, efficient pipeline of the Pentium M and gave strong performance without chasing the high clock rates NetBurst had aimed for. Xeon versions arrived on 26 June 2006 and Core 2 parts on 27 July 2006.
Inside the chip
The pipeline runs 14 stages, less than half the length of Prescott's, and the design can in principle retire up to four instructions per cycle against three for P6, Pentium M and NetBurst. Macro-Ops Fusion merges two x86 instructions, such as a compare plus a conditional jump, into one micro-operation, though not in 64-bit mode. Loads can be executed ahead of earlier stores whose addresses are not yet known. Each core has 64 KB of L1 cache, split 32 KB for data and 32 KB for instructions, and cores share L2 cache. All 128-bit SSE instructions run at one cycle throughput instead of two. Hardware virtualisation (VT-x), Intel 64 and SSSE3 are supported, but hyper-threading is absent.
How it was made
The first chips were built on a 65 nm process, with transistor counts from 105 million to 582 million. The 2007/2008 Tick of Intel's Tick-Tock cycle shrank the design to 45 nm, where counts ran from 228 million to 1,900 million. Intel fabricated the parts itself. Packages span Socket M and Socket P for mobile, LGA 775 for desktop, LGA 771 and Socket 604 for servers, plus FCBGA parts. High-end Core 2 and Xeon chips often used multi-chip modules holding two dies to raise cache size or core count.
Where it went
Merom went into mobile systems, Conroe into desktops and Woodcrest into servers and workstations. The same underlying dies were sold across brands: the Allendale core, for example, appeared as Celeron, Pentium, Core 2 and Xeon with different features switched on. Boards needed a compatible chipset from Intel, Nvidia, VIA, SiS or ATI, and Conroe parts also required the VRD 11.0 power delivery specification. The 1600 MT/s Yorkfield XE QX9770 worked with only a small set of boards, and the QX9775 fitted just the Intel D5400XS.
Versions and family
The 65 nm family covers Conroe, Allendale, Merom, Kentsfield, Woodcrest, Clovertown and Tigerton, differing in core count, cache and socket. Conroe-L and Merom-L cut the design to one core and 1 MB of L2 for low cost and low power. At 45 nm the parts are Penryn, Wolfdale and Yorkfield, with SSE4.1 and a new divide/shuffle engine, in 6 MB and 3 MB L2 versions. Dunnington added six cores and an L3 cache for servers. Penryn was the Tick; Nehalem followed as the Tock.
Why it matters
The design let Intel move every product line onto one efficient multi-core platform after NetBurst stalled. At the spring 2006 Intel Developer Forum the company promised 20% more performance than Core Duo for Merom at equal power, 40% more than Pentium D for Conroe using 40% less power, and 80% more than the first dual-core Xeon for Woodcrest at 35% less power. It set up the Tick-Tock rhythm that carried Intel through the 45 nm Penryn shrink and on to Nehalem.
Problems and reception
The architecture is listed as unsupported today. Board support was a common issue: even with the right chipset, some motherboards could not run Conroe because they lacked VRD 11.0, and BIOS updates arrived gradually for Penryn parts. The 45 nm steppings also included a bug fix release, C1/M1, made only for quad-core processors.

Questions people ask about Intel Core microarchitecture
What is the Intel Core microarchitecture?
It is the processor design Intel launched in mid-2006, code-named Merom, that powered the Core 2, Pentium, Celeron and Xeon brands. It is a multi-core design in the P6 family, a refreshed successor to the Enhanced Pentium M and the replacement for NetBurst.
Why did Intel drop NetBurst?
NetBurst used a long pipeline built for high clock rates, which meant high power draw and heat. By early 2004 the Prescott core needed too much power to reach competitive speeds, and that made it unsuitable for dual and multi-core chips. Intel cancelled the follow-on Tejas and Jayhawk on 7 May 2004.
What were Merom, Conroe and Woodcrest?
They were the first three processor lines built on this architecture. Merom was for mobile computing, Conroe for desktops, and Woodcrest for servers and workstations. They were architecturally the same but differed in socket, bus speed and power consumption.
Does it support hyper-threading?
No. Core-based processors carry multiple cores and hardware virtualisation (Intel VT-x), Intel 64 and SSSE3, but they do not have the hyper-threading found in Pentium 4 chips. Hyper-threading returned with the later Nehalem microarchitecture.
How was it manufactured?
The first chips came on a 65 nm process with 105 to 582 million transistors. Intel then shrank the design to 45 nm in the 2007/2008 Tick, raising counts to between 228 million and 1,900 million. Intel fabricated the parts in its own fabs.
Good reading and sources
- Intel InsideIntel, Article
- Checking your browserAnandTech, Article
- Into the Core: Intel's next-generation microarchitectureArs Technica, Article
The facts on this page come from the Wikipedia article "Intel Core (microarchitecture)", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryChip rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.
It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.
Like Intel Core microarchitecture
All CPUs and microprocessorsMore from Intel
All of them
🇺🇸 United StatesYonah
Intel's first 65 nm mobile CPU cores, sold as Core Duo and Core SoloCPU · x86 · 2006
🇺🇸 United StatesMontecito
Intel's first dual-core Itanium 2 processor, launched in July 2006CPU · Other · 2006