Z13
Mainframe microprocessor by IBM using 22 nm process
The z13 is a mainframe microprocessor designed by IBM and announced in January 2015. Built on a 22 nm process at GlobalFoundries, it features up to eight cores running at 5 GHz and introduced vector processing capabilities for improved performance in specialized tasks.
- Runs at clock speeds up to 5.0 GHz
- Contains 3.99 billion transistors on a 22 nm process
- First IBM mainframe chip in single-chip modules
- Adds 32 vector registers and 150 new instructions
- Last z Systems server to support ESA/390 mode
What it is
The z13 is a central processing unit built for IBM mainframe systems. Announced in early 2015, it serves as the computational heart of the z13 server line. IBM designed the chip to handle heavy enterprise workloads, claiming it was the fastest microprocessor available at launch. It executes the z/Architecture instruction set and supports legacy ESA/390 modes, making it suitable for large-scale commercial data processing where reliability and backward compatibility are critical.
How it came about
IBM developed this processor to succeed the zEC12 model. The design team focused on enhancing single-threaded speed while adding significant acceleration for specific data tasks. Announced on January 14, 2015, the chip marked a shift in packaging strategy for IBM mainframes. It was the final system in its lineage to support operating systems running in ESA/390 architecture mode, although existing applications remained unaffected. This transition allowed IBM to modernize the hardware while maintaining support for older software environments.
Inside the chip
Each core uses a superscalar, out-of-order pipeline capable of fetching six instructions and issuing ten per cycle. The chip supports two-way simultaneous multithreading and includes 139 new single instruction multiple data instructions. A new vector facility adds thirty-two 128-bit registers and two independent execution units. Cache hierarchy includes private level one and level two stores for each core, plus a shared 64 megabyte level three cache made of embedded dynamic random access memory. On-board controllers manage DDR3 memory with fault recovery features.
How it was made
GlobalFoundries manufactured the chips at their East Fishkill facility in New York using a 22 nanometre complementary metal oxide semiconductor process on silicon on insulator wafers. The fabrication involved seventeen metal layers. Each processor unit die measures 678 square millimetres and contains approximately 3.99 billion transistors. Unlike previous generations that used multi-chip modules, IBM packaged these processors in single-chip modules. The associated storage controller chips were made on the same process node but with fifteen metal layers.
Where it went
These processors power IBM z13 mainframe computers, which are typically deployed in data centres for critical business operations. The architecture handles high-volume transaction processing, database management, and secure cryptographic tasks. Features like logical partitioning allow a single physical machine to run multiple isolated operating system instances. The inclusion of PCIe gen 3 controllers enables connection to host channel adapters and peripheral devices, supporting complex enterprise infrastructure requirements.
Versions and family
The processor unit chip can be configured with six, seven, or eight active cores depending on the specific system needs. It sits between the zEC12 predecessor and the z14 successor in the product timeline. A companion storage controller chip manages communications between processor clusters and provides an additional 480 megabytes of level four cache per cluster. Two such controllers in a computer drawer yield a total of 960 megabytes of off-die cache shared across the system.
Why it matters
IBM stated the z13 offered a ten percent improvement in general single-threaded computing over the previous generation. Performance gains were significantly higher for specialized tasks utilizing the new vector facilities and compression engines. By moving to single-chip modules, IBM simplified the physical construction of mainframe drawers. The chip maintained support for legacy application states while introducing modern instruction sets for integer, floating-point, and string data operations, extending the viability of mainframe computing for contemporary workloads.
Questions people ask about Z13
What is the clock speed of the z13 processor?
The z13 microprocessor operates at a frequency of 5.0 GHz. This speed is slightly lower than its predecessor, the zEC12, but the chip achieves higher overall performance through architectural improvements and specialized instruction handling.
How many cores does the z13 chip have?
The processor unit chip contains up to eight cores. Depending on the specific configuration required by the system, IBM may enable six, seven, or all eight cores on a single die.
Who manufactured the IBM z13 chip?
While IBM designed the processor, it was manufactured by GlobalFoundries. Production took place at the East Fishkill fabrication plant in New York, which was formerly owned and operated by IBM itself.
What new instruction features did the z13 introduce?
The chip added a vector facility with 32 new 128-bit registers and over 150 new instructions. It also included 139 single instruction multiple data commands, improved cryptography, and data compression capabilities.
Good reading and sources
The facts on this page come from the Wikipedia article "IBM z13", 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 Z13
All CPUs and microprocessorsMore from IBM
All of them
🇺🇸 United StatesTrueNorth
IBM neuromorphic chip mimicking brain structureAI accelerator · Other · 2014