*   `tf.keras.optimizers.Optimizer` now points to the new Keras optimizer, and
    old optimizers have moved to the `tf.keras.optimizers.legacy` namespace.
    If you find your workflow failing due to this change,
    you may be facing one of the following issues:

    *   **Checkpoint loading failure.** The new optimizer handles optimizer
        state differently from the old optimizer, which simplies the logic of

dst+16(FP), R3 // Dest pointer MOVD src+24(FP), R4 // Src pointer SETUP_ESPERM(R7) // Set CR{1,2,3}EQ to hold the key size information. CMPU R6, $10, CR1 CMPU R6, $12, CR2 CMPU R6, $14, CR3 MOVD $16, R6 MOVD $32, R7 MOVD $48, R8 MOVD $64, R9 MOVD $80, R10 MOVD $96, R11 MOVD $112, R12 // Load text in BE order P8_LXVB16X(R4, R0, V0) // V1, V2 will hold keys, V0 is a temp. // At completion, V2 will hold the ciphertext. // Load xk[0:3] and xor with text LXVD2X (R0+R5), V1 VXOR V0, V1, V0 // Load xk[4:11]...

dst+16(FP), R3 // Dest pointer MOVD src+24(FP), R4 // Src pointer SETUP_ESPERM(R7) // Set CR{1,2,3}EQ to hold the key size information. CMPU R6, $10, CR1 CMPU R6, $12, CR2 CMPU R6, $14, CR3 MOVD $16, R6 MOVD $32, R7 MOVD $48, R8 MOVD $64, R9 MOVD $80, R10 MOVD $96, R11 MOVD $112, R12 // Load text in BE order P8_LXVB16X(R4, R0, V0) // V1, V2 will hold keys, V0 is a temp. // At completion, V2 will hold the ciphertext. // Load xk[0:3] and xor with text LXVD2X (R0+R5), V1 VXOR V0, V1, V0 // Load xk[4:11]...