Z15
IBM mainframe processor with on-chip compression and eDRAM cache
The z15 is a microprocessor designed by IBM for its z15 mainframe systems. Announced in September 2019, it features twelve cores, an on-chip nest accelerator for compression, and extensive eDRAM cache levels.
- Announced by IBM on 12 September 2019
- Features 12 cores with two-way multithreading
- Includes on-chip nest accelerator for compression
- Uses eDRAM for all cache levels instead of SRAM
- Successor to the z14 and predecessor to Telum
What it is
The z15 is a central processing unit built by IBM for use in its z15 mainframe computers. It was announced on 12 September 2019. This chip handles complex enterprise workloads using the z/Architecture instruction set. It is notable for integrating a dedicated acceleration unit for data compression directly on the processor die. The design also shifts from traditional static random-access memory to embedded dynamic random-access memory for its cache layers, allowing for larger storage capacities within the same physical footprint.
How it came about
IBM developed this processor to succeed the z14 model. The team focused on improving data handling efficiency and cache capacity. By moving to embedded dynamic memory, they addressed the need for faster access to larger datasets without increasing the chip size disproportionately. The announcement in late 2019 marked the availability of these enhanced capabilities for enterprise server environments.
Inside the chip
The chip contains twelve cores that support two-way simultaneous multithreading, allowing each core to handle two tasks at once. It uses a superscalar, out-of-order pipeline to execute instructions efficiently. A new feature is the on-chip nest accelerator unit, which all cores share to speed up data compression tasks. The cache hierarchy is extensive. Each core has 128 kilobytes of level one instruction cache and 128 kilobytes of data cache. Level two cache provides 4 megabytes for instructions and 4 megabytes for data per core. A shared level three cache offers 256 megabytes. All these levels use embedded dynamic random-access memory rather than static random-access memory.
How it was made
The processor is manufactured using a 14 nanometre process node. IBM designed the chip internally. The use of embedded dynamic random-access memory for all cache levels, including level one through level three, distinguishes its fabrication approach from previous generations that relied on static random-access memory. This material choice enables higher density for the cache structures on the silicon die.
Where it went
This microprocessor powers the IBM z15 mainframe computer systems. These machines are typically deployed in data centres for critical enterprise applications. The enhanced compression capabilities and large cache sizes make it suitable for workloads that require high throughput and low latency for large volumes of transactional data.
Versions and family
The z15 follows the z14 microprocessor. Its successor is the Telum processor. The system controller chip, known as the SCM, works alongside the main processor unit chip. This add-on chip manages the level four cache, which can reach 960 megabytes per drawer in a five-CPC configuration, totaling 4800 megabytes of shared storage.
Why it matters
The shift to embedded dynamic random-access memory allowed IBM to double the level three cache compared to the previous generation. The level four cache also saw a significant increase, rising by 43 per cent to 960 megabytes. These improvements support more efficient data processing for mainframe users, particularly those relying on real-time compression and decompression operations.
Questions people ask about Z15
What architecture does the z15 use?
The z15 implements the CISC z/Architecture. It uses a superscalar, out-of-order pipeline to manage instruction execution across its twelve cores.
How much cache memory does the z15 have?
Each core has 128 KB L1 instruction and data cache, plus 4 MB L2 instruction and data cache. There is a shared 256 MB L3 cache. An external controller adds up to 960 MB of L4 cache.
What is special about the z15 cache technology?
All cache levels, from L1 to L3, use embedded dynamic random-access memory (eDRAM) instead of the traditional static random-access memory (SRAM). This allows for greater density and capacity.
Does the z15 have any special acceleration units?
Yes, it includes an on-chip nest accelerator unit. This unit is shared by all cores and is designed specifically to accelerate data compression tasks.
Good reading and sources
The facts on this page come from the Wikipedia article "IBM z15", 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.