Nehalem
Intel's 45 nm CPU design behind the first Core i5 and i7 chips
Nehalem is the internal name for a processor design from Intel, first shipped in November 2008 at 45 nm. It powered the debut Core i5 and i7 lines and brought back simultaneous threading while moving the memory controller onto the die.
- First shipped November 2008 at 45 nm
- 731M to 2300M transistors, 1.06 GHz to 3.33 GHz
- Shared L3 cache of 2 MB to 24 MB across all cores
- Hyper-threading and Turbo Boost 1.0 reintroduced
- QuickPath Interconnect replaced the front side bus
What it is
Nehalem is the code name Intel gave to a CPU design that reached customers in November 2008. Built at 45 nm, it sat inside the first Core i5 and i7 desktop and mobile parts as well as Xeon server chips. The design is a large step on from the Core architecture used in Core 2, itself the last chapter of the P6 line that began with Pentium Pro in 1995. Its name comes from a river. Compared with Penryn it reaches higher clock rates without losing efficiency, and it draws less power for the same work.
How it came about
Intel positioned this generation as a performance-focused redesign, which is why individual cores grew larger. The launch in November 2008 covered desktop parts first, with server and mobile versions following under code names such as Gainestown, Beckton and Clarksfield. A 32 nm shrink called Westmere came later, and Sandy Bridge replaced the design completely in January 2011.
Inside the chip
Each core carries 64 KB of level-1 cache split evenly between data and instructions, plus 256 KB of level-2. A shared level-3 pool of 2 to 24 MB serves all cores. Cache lines shrank to 64 bytes. The fetch unit gained a second-level branch predictor with a two-level target buffer and a return stack, and a 512-entry second-level page table cache handles small pages. Execution hardware counts three integer units, two vector units and two address generators per core. Simultaneous threading returned, Turbo Boost 1.0 arrived, and the pipeline runs 20 to 24 stages. Chips use QuickPath Interconnect or DMI instead of a front side bus, and the memory controller plus PCI Express sit on the die.
How it was made
Production used Intel's 45 nm process, with transistor counts spanning 731 million to 2300 million depending on the model. Packaging options include LGA 1156, LGA 1366, LGA 1567 and the mobile μPGA 988. Clock speeds across the family run from 1.06 GHz to 3.33 GHz.
Where it went
Desktop buyers met the design in Core i5-7xx and Core i7-8xx and 9xx parts, including Extreme Edition models. Xeon 3000, 5000 and 7000 series chips carried it into servers and workstations, while mobile Core i7-7xxQM, 8xxQM and 9xxXM parts served laptops. Lynnfield chips pair with a platform controller hub, so no separate northbridge is needed, and they expose 16 PCI Express lanes arranged as one x16 or two x8. Gainestown parts add dual QuickPath links and effectively six memory channels.
Versions and family
Named cores built on the design include Bloomfield, Lynnfield, Clarksfield, Jasper Forest, Gainestown (Nehalem-EP) and Beckton (Nehalem-EX). Aubrurndale and Havendale were planned but cancelled. The 6500 series scales to two sockets and the 7500 series to four or eight. Westmere shrinks the same basic design to 32 nm, and Sandy Bridge is the full successor.
Why it matters
The generation restored simultaneous multi-threading to Intel's mainstream line and put the memory controller and high-speed links on the processor package, ending reliance on a front side bus. Reported gains over Penryn were 10 to 25 percent in single-threaded work and 20 to 100 percent in multi-threaded work at equal power, with about 30 percent less power at equal performance. Clock for clock, each core averaged 15 to 20 percent more throughput.
Problems and reception
SSE4.2 added seven instructions on top of SSE4.1, including POPCNT, and macro-op fusion now applies in 64-bit mode. Latency for atomic operations such as LOCK CMPXCHG was cut by half. Overclocking is available on Bloomfield processors paired with the X58 chipset. Support status is listed as unsupported.
Questions people ask about Nehalem
What exactly is Nehalem?
It is the code name for a processor design from Intel that went on sale in November 2008. Chips built on it appeared as the first Core i5 and i7 desktop and mobile processors, plus Xeon server parts, all made on a 45 nm process.
How does it differ from the Core 2 generation?
It brings back simultaneous threading, shrinks level-2 cache to 256 KB per core while enlarging the shared level-3 to between 2 MB and 24 MB, moves the memory controller onto the package, and swaps the front side bus for QuickPath or DMI links.
Which sockets does it use?
Depending on the model, chips fit LGA 1156, LGA 1366 or LGA 1567 on desktops and servers, while mobile parts use the μPGA 988 package.
What came before and after?
Core, used in Core 2, and Penryn preceded it. Westmere is a 32 nm shrink of the same basic design, and Sandy Bridge replaced it entirely in January 2011.
How much faster was it than Penryn?
Reported figures put single-threaded work 10 to 25 percent ahead and multi-threaded work 20 to 100 percent ahead at the same power, with roughly 30 percent less power for equal performance and a 15 to 20 percent clock-for-clock gain per core.
Good reading and sources
- Nehalem (microarchitecture) on WikipediaEncyclopedia
- More photos on Wikimedia CommonsPhotos
- Wikidata recordOpen data
The facts on this page come from the Wikipedia article "Nehalem (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.