krl_L2_table_lookup_fast_scan_bs64

Interface Definition

int krl_L2_table_lookup_fast_scan_bs64(int nsq, const uint8_t* codes, const uint8_t* LUT, uint16_t* dis, uint16_t threshold, uint32_t* lt_mask, size_t codes_size, size_t LUT_size, size_t dis_size, size_t lt_mask_size);

Function

An operator designed for 4-bit table lookup, accumulation, filtering, and compression. It is used for processing float-type query vectors individually. It computes Euclidean distances between a query vector and 64 base vectors. Based on the distance comparison rule, if a distance value meets the comparison rule, the bit in lt_mask of the corresponding base vector is set to 1. Otherwise, it is set to 0.

Parameters

Parameter	Data Type	Description	Value Range
nsq	int	Number of subspaces.	An even number in [1, 256]
codes	const uint8_t*	Codebook, which stores indexes of base vectors after krl_pack_codes_4b (blocksize=64) processing.	The size is specified by codes_size.
LUT	const uint8_t*	Distance LUT, which stores the uint8 distances between product quantization (PQ) centroids and query vectors in each subspace.	The size is specified by LUT_size.
dis	uint16_t*	Stores the distance computation result.	The size is specified by dis_size. The memory needs to be allocated in advance.
threshold	const uint16_t	Distance threshold.	The value is 65535.
lt_mask	uint32_t*	Filtered and compressed results.	The size is specified by lt_mask_size. The memory needs to be allocated in advance.
codes_size	size_t	Length of the codebook array.	nsq/2 x 64
LUT_size	size_t	Length of the distance LUT array.	nsq x 16
dis_size	size_t	Size of the distance result array.	The value is 64.
lt_mask_size	size_t	Length of the array storing filtered and compressed results.	The value is 2.

Return Values

Data Type	Description
int	0 is returned on successful execution. In case of failure, an error code is returned to indicate the specific issue:

Parent topic: 4-bit Table Lookup-based Accumulation Interfaces