TPU v5p

TPU v5p is a training-focused generation. Each chip has 2 TensorCores and 4 SparseCores, and a pod holds 8960 chips connected in a 3D torus.

Use the specification topic to size a slice, the configurations topic to choose a slice shape, and the VM types topic to see what each VM provides.

• TPU v5p specifications
  Per-chip and per-pod specifications for TPU v5p.

• TPU v5p supported configurations
  The slice shapes you can request for TPU v5p, and how many chips and hosts each uses.

• TPU v5p VM types
  The VM types available for TPU v5p and the host resources each one provides.

Related information

TPU v5p specifications

TPU v5p supported configurations

TPU v5p VM types

Create a TPU VM