12,2,LR // BEQLR MOVD R9, R6 CMP R6, $8 BLE small // Case for 8 <= n < 64 bytes // Process 32 bytes if available xor32: CMP R6, $32 BLT xor16 ADD $16, R8, R9 LXVD2X (R4)(R8), VS32 LXVD2X (R4)(R9), VS33 LXVD2X (R5)(R8), VS34 LXVD2X (R5)(R9), VS35 XXLXOR VS32, VS34, VS32 XXLXOR VS33, VS35, VS33 STXVD2X VS32, (R3)(R8) STXVD2X VS33, (R3)(R9) ADD $32, R8 ADD $-32, R6 CMP R6, $8 BLE small // Case for 8 <= n < 32 bytes // Process 16 bytes if available xor16: CMP R6, $16 BLT xor8 LXVD2X (R4)(R8), VS32 LXVD2X...

12,2,LR // BEQLR MOVD R9, R6 CMP R6, $8 BLE small // Case for 8 <= n < 64 bytes // Process 32 bytes if available xor32: CMP R6, $32 BLT xor16 ADD $16, R8, R9 LXVD2X (R4)(R8), VS32 LXVD2X (R4)(R9), VS33 LXVD2X (R5)(R8), VS34 LXVD2X (R5)(R9), VS35 XXLXOR VS32, VS34, VS32 XXLXOR VS33, VS35, VS33 STXVD2X VS32, (R3)(R8) STXVD2X VS33, (R3)(R9) ADD $32, R8 ADD $-32, R6 CMP R6, $8 BLE small // Case for 8 <= n < 32 bytes // Process 16 bytes if available xor16: CMP R6, $16 BLT xor8 LXVD2X (R4)(R8), VS32 LXVD2X...

### `tf.data`:

*   Adds new `tf.data.experimental.service.register_dataset` and
    `tf.data.experimental.service.from_dataset_id` APIs to enable one process to
    register a dataset with the tf.data service, and another process to consume
    data from the dataset.
*   Adds support for dispatcher fault tolerance. To enable fault tolerance,
    configure a `work_dir` when running your dispatcher server and set