relu_tile

void ckernel::relu_tile_init()

Please refer to documentation for any_init.

void ckernel::relu_tile(uint32_t idst)

Performs element-wise computation of relu (0 if negative else 1) on each element of a tile in DST register at index tile_index. The DST register buffer must be in acquired state via acquire_dst call. This call is blocking and is only available on the compute engine.

Return value: None

Argument	Description	Type	Valid Range	Required
tile_index	The index of the tile in DST register buffer to perform the computation on	uint32_t	Must be less than the size of the DST register buffer	True

relu_max_tile

void ckernel::relu_max_tile_init()

Please refer to documentation for any_init.

void ckernel::relu_max_tile(uint32_t idst, uint32_t param0)

Performs element-wise computation of relu max (relu(max(x, upper_limit))) on each element of a tile in DST register at index tile_index. The DST register buffer must be in acquired state via acquire_dst call. This call is blocking and is only available on the compute engine.

Return value: None

Argument	Description	Type	Valid Range	Required
tile_index	The index of the tile in DST register buffer to perform the computation on	uint32_t	Must be less than the size of the DST register buffer	True
upper_limit	Upper limit of relu_min	uint32_t	Greater than 0	True

relu_min_tile

void ckernel::relu_min_tile_init()

Please refer to documentation for any_init.

void ckernel::relu_min_tile(uint32_t idst, uint32_t param0)

Performs element-wise computation of relu min (relu(min(x, lower_limit))) on each element of a tile in DST register at index tile_index. The DST register buffer must be in acquired state via acquire_dst call. This call is blocking and is only available on the compute engine.

Return value: None

Argument	Description	Type	Valid Range	Required
tile_index	The index of the tile in DST register buffer to perform the computation on	uint32_t	Must be less than the size of the DST register buffer	True
lower_limit	Upper limit of relu_min	uint32_t	Greater than 0	True

leaky_relu_tile

void ckernel::leaky_relu_tile_init()

Please refer to documentation for any_init.

void ckernel::leaky_relu_tile(uint32_t idst, uint32_t slope)

Performs element-wise computation of leaky relu (relu(x) + slope*-relu(-x)) on each element of a tile in DST register at index tile_index. The DST register buffer must be in acquired state via acquire_dst call. This call is blocking and is only available on the compute engine.

Return value: None

Argument	Description	Type	Valid Range	Required
tile_index	The index of the tile in DST register buffer to perform the computation on	uint32_t	Must be less than the size of the DST register buffer	True
slope	slope used in leaky relu - will reinterpret unsigned int to float	uint32_t	Greater than 0	True