Jevin Jiang ed4958cb3e [XLA:Mosaic] Add internal scratch VMEM ...

- Make internal scratch size configurable.
- Pass the number of max sublanes allowed in scratch to apply-vector-layout pass.
- Create a helper function to fetch internal scratch VMEM address.

PiperOrigin-RevId: 644184896

2024-06-17 17:31:31 -07:00

..

dialect/tpu

[XLA:Mosaic] Add internal scratch VMEM

2024-06-17 17:31:31 -07:00

gpu

Use constant memory to pass in TMA descriptors to the kernel

2024-06-17 05:31:26 -07:00

python

Fix some load orderings for buildifier

2024-03-27 10:28:57 -07:00

BUILD

Adds rewrite patterns for arith and math operations with bf16 operands/results that are not supported by the underlying hardware.

2024-05-21 11:08:19 -07:00