IBM Telum

Telum is a microprocessor made by IBM for the IBM z16 series mainframe computers.²³ The processor was announced at the Hot Chips 2021 conference on 23 August 2021.² Telum is IBM's first processor that contains on-chip acceleration for artificial intelligence inferencing while a transaction is taking place.⁴⁵

The chip contains 8 processor cores with a deep superscalar out-of-order pipeline, capable of achieving higher than 5 GHz clock frequency. The cache and chip-interconnection infrastructure provides 32 MB cache per core and can scale to 32 Telum chips.⁶³⁷ The cache design ⁶ allows creating a system where the L2 cache of one core can be used as virtual L3 and L4 caches for another core.³¹ The Telum processor can either be water cooled or air cooled, but water cooling is required for running more than a few Telum processors in a single IBM compute drawer.⁸⁹ Uniquely, the IBM Telum does not thermal throttle by reducing clock speed; instead it inserts sleep state instructions.⁸⁹

Telum adds a new 16-bit floating point format (NNP-Data-Type-1 Format) and several new instructions.¹⁰ The Neural Network Processing Assists (NNPA)¹¹ instruction performs a variety of tensor instructions useful for neural networks.^a

Telum II adds new functions to NNPA.¹²

z-14

z/Architecture Principles of Operation (PDF) (Fourteenth ed.). IBM. May 2022. pp. 26-61 – 26-96. SA22-7832-13. Retrieved March 31, 2025.

z-15

z/Architecture Principles of Operation (PDF) (Fifteenth ed.). IBM. April 2025. SA22-7832-14. Retrieved March 31, 2025.