POWER7
IBM multi-core server processor from 2010
POWER7 is a family of superscalar multi-core microprocessors designed by IBM and released in 2010. It succeeded the POWER6 series, shifting focus from raw clock speed to core count and power efficiency for servers and supercomputers.
- Released in February 2010 by IBM
- Features up to 8 cores with 4 threads each
- Uses 45 nm SOI process with 1.2 billion transistors
- L3 cache uses space-saving embedded DRAM
- Supports TurboCore mode for higher single-thread speed
What it is
POWER7 is a family of microprocessors built by IBM for high-performance computing and enterprise servers. Launched in February 2010, it uses the Power ISA 2.06 instruction set. The design moves away from the high-frequency approach of its predecessor, favouring multiple cores and simultaneous multithreading to improve overall throughput and energy efficiency. It supports up to eight physical cores per chip, with each core handling four threads at once.
How it came about
Development began after IBM secured a DARPA contract in late 2006 to build a petascale supercomputer architecture. This project, known as PERCS, required hardware that could support global shared memory spaces, allowing scientists to program clusters as single systems. IBM collaborated with DARPA to modify addressing hardware for this purpose. The resulting processor was announced in early 2010, meeting the deadline for commercial availability within the HPCS project timeline.
Inside the chip
The architecture features aggressive out-of-order execution, a shift from the in-order design of the POWER6. Each core contains twelve execution units, including floating-point and vector units. A key innovation is the use of embedded DRAM for the L3 cache, which saves space compared to standard SRAM. The chip includes a TurboCore mode that disables half the cores to boost clock speeds on the remaining ones, aiding sequential workloads. It can dispatch six instructions per cycle and issue eight to execution units.
How it was made
IBM fabricated the chips using a 45 nanometre silicon-on-insulator process. The die size is 567 square millimetres and contains approximately 1.2 billion transistors. Clock speeds range from 3.0 to 4.25 gigahertz. The processors are often packaged in quad-chip modules, allowing up to four chips to work together in a single unit. This packaging supports systems with up to 32 ways of multiprocessing.
Where it went
These processors power IBM Power Systems servers, including Express, Enterprise, and High Performance models. Specific systems include the Power 710 through 795 series and various BladeCenter modules. In supercomputing, POWER7 drives the Watson system and the PERCS architecture. The TurboCore feature helps optimise performance for software licensed per core, reducing costs while maintaining speed for sequential tasks.
Versions and family
The initial lineup included chips with four, six, or eight cores. In August 2012, IBM introduced the POWER7+, an updated version made on a 32 nanometre process. This successor offered higher clock speeds up to 4.4 gigahertz and increased L3 cache up to 80 megabytes. The original POWER7 was succeeded by the POWER8 family, which continued the evolution of the Power ISA architecture.
Why it matters
POWER7 marked a strategic pivot for IBM towards parallel processing efficiency rather than pure frequency. It enabled petascale computing capabilities within the DARPA HPCS project. By integrating large eDRAM caches and supporting massive thread counts, it influenced server design for data-intensive workloads. The architecture demonstrated that lower clock speeds with more cores could deliver competitive floating-point performance against contemporary x86 designs.




Questions people ask about POWER7
How many cores does a POWER7 chip have?
A single POWER7 microchip contains either four, six, or eight physical cores. Each core supports four simultaneous threads, allowing a single chip to handle up to 32 threads at once.
What is special about the POWER7 cache?
The L3 cache uses embedded DRAM instead of standard SRAM. This technology requires fewer transistors per cell, allowing IBM to fit a larger 4 MB cache per core into the same area.
What is TurboCore mode?
TurboCore allows an eight-core processor to disable half its cores. The remaining four cores gain access to all memory controllers and cache, running at higher clock speeds for better sequential performance.
Which instruction set does POWER7 use?
It is based on the Power ISA 2.06 architecture. This instruction set supports superscalar execution and simultaneous multithreading, differing significantly from x86 or ARM architectures.
Good reading and sources
- Checking your browserAnandTech, Article
- IBM Has Its PERCSHPCwire, Article
- Power7+ chips debut in fat IBM midrange systemsThe Register, Article
The facts on this page come from the Wikipedia article "POWER7", 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 POWER7
All CPUs and microprocessors
🇺🇸 United StatesPOWER5
Dual-core server microprocessor with simultaneous multithreadingCPU · Power · 2004