una-dinosauria / Rayuela.jl

Code for my PhD thesis. Library of quantization-based methods for fast similarity search in high dimensions. Presented at ECCV 18.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Remove unused CUDA kernels

una-dinosauria opened this issue · comments

Some CUDA code is never called from julia, which is confusing, eg

__device__ void _veccost(
float *d_rx, // data to use (X)
float *d_codebooks, // codebooks (C)
unsigned char *d_codes, // the codes (B)
float *d_veccost, // where to save the cost
int m, // number of codebooks
int n) { // number of vectors in X
// FIXME hard-coding 256 entries in each codebook, and 128 dimensions
const int H = 256; // size of each codebook
const int D = 128; // dimensionality of each vector

as reported in #39. We should remove unused code in general.