IBM Power S1124
The Power S1124 is the largest scale-out IBM Power 11 server, machine type 9824-42A, and IBM lists it for 8, 12, 16 or 24 Spyre AI accelerator cards ... but only when both processor modules are installed. An IBM i shop can keep production on IBM i 7.4 TR12 or later and run model inference in a Red Hat Enterprise Linux partition on the same 4U machine.
What is it?
The Power S1124 is the biggest of IBM's scale-out Power 11 servers. IBM calls it machine type 9824-42A. It is one 4U box with two processor sockets, so it is not built in drawers the way the big Power E1180 is. Think of it as a single large toolbox rather than a chest of drawers. You pick one of six processor features, fill the memory slots you need, and that is the machine. It runs IBM i, AIX and Linux, and it can run them side by side in separate partitions.
What IBM and its configuration data list for the Power S1124
| Item | What is listed |
|---|---|
| Machine type model | 9824-42A |
| Size and sockets | 4U, two processor sockets |
| General availability | 8 July 2025, announcement letter AD25-0036 |
| Cores | 16 to 60 Power 11 cores |
| Processor features | EP3X 16 or 32 core, EP3H 24 or 48 core, EP3Y 30 or 60 core |
| Clock speeds | 3.4 to 4.2 GHz on 16 and 32 core, 3.05 to 4.15 GHz on 24 and 48 core, 2.8 to 3.95 GHz on 30 and 60 core |
| Memory | 64 GB to 8 TB DDR5 across 32 slots |
| Internal storage | Up to 16 NVMe U.2 bays |
| PCIe slots in the body | Four Gen4 x16 or four Gen5 x8, four Gen5 x8 in x16 connectors, two Gen4 x8 |
| Operating systems | IBM i, AIX and Linux |
| IBM i minimum levels | 7.4 TR12, 7.5 TR6, or 7.6 with the GA PTF group |
The 4U size, the 60 core ceiling and the 8 TB memory ceiling are IBM's own figures from its Power S1124 product page. The availability date and announcement letter are from IBM's product lifecycle record for 9824-42A. Processor feature codes, clock speeds, memory slot count, bay count, slot mix and IBM i levels are from the Midland Information Systems configuration page for 9824-42A. U.2 is the connector shape used by the internal flash drives.
How many Spyre cards does it take?
IBM's Spyre Accelerator for Power publication lists 9824-42A at 8, 12, 16 or 24 cards per system. You cannot order one card or three. You order a supported step. Each card holds 32 accelerator cores and 128 GB of its own memory, so the step you pick decides how much AI work the machine can hold at once. Only the Power E1180 goes higher than 24. Everything about this is a Linux-side order, not an IBM i one.
AI accelerator cores at each card step IBM lists for the Power S1124
IBM lists 32 accelerator cores and 128 GB of LPDDR5 memory on each PCIe5 x16 AI Accelerator adapter, feature code ECSE, CCIN 2E1F. So 8 cards carry 1 TB of AI memory and 24 cards carry 3 TB. The card counts are IBM's own supported steps for 9824-42A, not a free choice.
- 9824-42A, 9043-MRU and 9856-42H take 8, 12, 16 or 24 cards
- 9824-22A, 9824-22B and 9856-22H take 8 or 12 cards
- 9080-HEU, the Power E1180, takes 8 to 48 cards in groups of 8 or 12
- One logical partition holds up to 24 cards no matter which machine it runs on
Where do the cards go?
Not in the server body. The S1124 has ten PCIe slots of its own, and storage and network cards want most of them. The Spyre cards go in a separate box: the ENZ0 PCIe4 expansion drawer. IBM describes the ENZ0 as a 4U, 19 inch Gen4 drawer with up to 12 adapter slots. That is why the steps run 8, 12, 16 and 24 ... one drawer covers the first two steps, and 16 or 24 cards needs more than one.
Three things to order before an S1124 can run Spyre
- Both processor modulesIBM says the S1124 and L1124 qualify for Spyre only when populated with two processors. A single-module S1124 is not on the supported list.
- An ENZ0 drawer with the right fanout modulesIBM says the AI accelerator adapter is only supported in ENZ0 PCIe4 expansion drawers with unit type U5B67 and feature code ENZA fanout modules. The cards will not go anywhere else.
- Red Hat software on a Linux partitionIBM requires Red Hat AI Inference Server or Red Hat OpenShift AI on Red Hat Enterprise Linux 9.6, 9.8 or 10.2, with Podman and Quadlet as the container runtime. A drawer holding 12 cards needs Red Hat AI Inference Server 3.5 or later.
So a Spyre quote on an S1124 is never just a card quote. It is cards, plus a drawer with the right fanout modules, plus a second processor module if the machine only has one, plus Red Hat software.
Can IBM i use the cards?
Not directly. IBM's Spyre software for Power runs on Red Hat Enterprise Linux, and IBM i is not on that list. What you get instead is both jobs on one machine. IBM i runs in its own partition at 7.4 TR12, 7.5 TR6, or 7.6 with the GA PTF group. The Linux inference partition runs beside it and owns the cards. Your data never has to leave the building to get an answer from a model.
IBM lists a floor for the partition that drives the cards: at least 512 GB of memory, 10 processor cores, and 600 GB of storage, per model you serve. The S1124 tops out at 8 TB of memory and 60 cores, so the floor fits. But it is real space, and it comes out of the same machine your production work runs on.
- Spyre cards can be added to or removed from a running partition the same way as other PCI devices
- Stop every container using a card before you remove it from the partition
- After adding a card to a running partition, rerun the service report and restart the containers so memory limits are set correctly
Related Terms
Spyre Accelerator
IBM's PCIe AI inference card. On IBM Power 11 it carries 32 cores, 128 GB of memory, and more than 300 TOPS inside a 75 watt envelope.
ENZ0 Expansion Drawer
The PCIe4 expansion drawer IBM documents for supported Spyre Accelerator for Power configurations on Power 11 systems.
IBM Power E1180
The Power E1180 is the high-end IBM Power 11 server, machine type 9080-HEU, and the only Power model IBM lists for more than 24 Spyre AI accelerator cards ... up to 48 per system. An IBM i shop can keep production on IBM i 7.4 TR12 or later and run large AI inference in a Red Hat Enterprise Linux partition on the same machine.
AI Inference Partition
A separate supported operating-system partition used to execute an AI model while an IBM i partition may remain the production application and data source.
Red Hat AI Inference Server
Red Hat's model serving product, built on the open source vLLM engine. IBM lists it as one of two supported ways to drive a Spyre Accelerator on IBM Power.