IBM Telum II Processor
The mainframe processor that proved IBM's whole AI strategy. Telum II does not go in a Power server, but the AI engine inside it is the reason the Spyre card exists, and the reason IBM talks about inference in milliseconds rather than seconds.
Telum II is a mainframe chip. You will not put one in a Power server and it has nothing to do with IBM i. It is in this directory for one reason: it is where IBM's current AI hardware idea was proven, and understanding it explains why the Spyre card behaves the way it does.
The idea it proved
Banks wanted to check whether a card payment was fraudulent while the payment was still happening, not five minutes later in a batch report. That is a brutal deadline. Sending the transaction out to a separate AI server and waiting for a reply was too slow, so IBM put the AI engine on the processor itself.
It worked. Telum shipped with an on-chip AI accelerator, Telum II improved it and added a data processing unit to handle input and output, and the design became IBM's template: put the answering machine as close to the data as physically possible.
The line from here to your Power server
How one design idea reached three different IBM machines
| Chip | Where it runs | What it does for AI |
|---|---|---|
| Telum and Telum II | IBM Z and LinuxONE | AI engine on the processor, scoring inside a live transaction |
| AIU | IBM Research only | Scaled that engine up into a whole 32-core chip |
| Spyre Accelerator | IBM Power 11, IBM Z, LinuxONE | The AIU turned into a PCIe card you can order |
| Matrix Math Acceleration | Every Power 10 and Power 11 core | The same close-to-the-data principle, built into the processor you already own |
Different chips, one consistent argument: keep the model next to the data instead of shipping the data to the model.
Telum II is background for a Power buyer, not a purchasing option. If you are comparing mainframe and Power options in the same project, treat them as separate evaluations.
Related
IBM Artificial Intelligence Unit (AIU)
The research chip that became the card you can order. The AIU is why the Spyre Accelerator has 32 cores and why those cores look so much like the AI engine already inside an IBM Z processor.
IBM Spyre Accelerator for Power
A 75 watt PCIe card that adds more than 300 TOPS of AI inference to a Power 11 server. It is the only IBM AI chip an AS/400 shop can actually order as an add-on, and it comes with a short list of things it refuses to work without.
Power 11 Matrix Math Acceleration
The AI acceleration most AS/400 shops already own and have never switched on. It is inside every core of the processor, it costs nothing extra, and it is the cheapest honest answer to a first AI pilot.
IBM NorthPole
A research chip with no separate memory at all. NorthPole claims 25 times the energy efficiency of a conventional design by refusing to move data, and IBM has already run a Granite model on it.