support other floating-point types

[xmyqsh]

1. support other floating-point types for CPU
2. change function parameter to const reference
3. rename nearest_neighbors_points_cpu.cpp to nearest_neighbor_points_cpu.cpp

Running benchmarks for bm_nearest_neighbor_points/bm_nn_points...

Benchmark Avg Time(μs) Peak Time(μs) Iterations

NN_PYTHON_1_3_1_32 17 50 28602
NN_PYTHON_1_3_1_128 20 54 24513
NN_PYTHON_1_3_128_32 104 149 4822
NN_PYTHON_1_3_128_128 1988 3808 253
NN_PYTHON_1_4_1_32 17 36 28698
NN_PYTHON_1_4_1_128 21 44 24384
NN_PYTHON_1_4_128_32 109 132 4608
NN_PYTHON_1_4_128_128 2058 8013 243
NN_PYTHON_4_3_1_32 22 53 22869
NN_PYTHON_4_3_1_128 27 61 18325
NN_PYTHON_4_3_128_32 2041 7398 245
NN_PYTHON_4_3_128_128 2055 6362 245
NN_PYTHON_4_4_1_32 26 60 19201
NN_PYTHON_4_4_1_128 32 69 15453
NN_PYTHON_4_4_128_32 2084 11528 240
NN_PYTHON_4_4_128_128 2330 10419 215
NN_PYTHON_32_3_1_32 41 85 12251
NN_PYTHON_32_3_1_128 83 138 6022
NN_PYTHON_32_3_128_32 2965 7612 169
NN_PYTHON_32_3_128_128 2588 4914 194
NN_PYTHON_32_4_1_32 44 70 11431
NN_PYTHON_32_4_1_128 88 132 5667
NN_PYTHON_32_4_128_32 2317 13315 216
NN_PYTHON_32_4_128_128 3088 7298 162

Benchmark Avg Time(μs) Peak Time(μs) Iterations

NN_CPU_1_3_1_32 2 34 276333
NN_CPU_1_3_1_128 2 19 231216
NN_CPU_1_3_128_32 20 53 25534
NN_CPU_1_3_128_128 63 98 7963
NN_CPU_1_4_1_32 2 18 273783
NN_CPU_1_4_1_128 2 21 219138
NN_CPU_1_4_128_32 23 45 21636
NN_CPU_1_4_128_128 74 109 6714
NN_CPU_4_3_1_32 2 19 229715
NN_CPU_4_3_1_128 4 20 140817
NN_CPU_4_3_128_32 78 157 6373
NN_CPU_4_3_128_128 246 296 2030
NN_CPU_4_4_1_32 2 19 215370
NN_CPU_4_4_1_128 4 21 126249
NN_CPU_4_4_128_32 90 136 5578
NN_CPU_4_4_128_128 299 375 1675
NN_CPU_32_3_1_32 6 32 80616
NN_CPU_32_3_1_128 17 50 29214
NN_CPU_32_3_128_32 607 681 824
NN_CPU_32_3_128_128 1959 2084 256
NN_CPU_32_4_1_32 9 26 58329
NN_CPU_32_4_1_128 29 61 17353
NN_CPU_32_4_128_32 713 1204 702
NN_CPU_32_4_128_128 2340 2601 214

Benchmark Avg Time(μs) Peak Time(μs) Iterations

NN_CUDA_1_3_1_32 12 92 41091
NN_CUDA_1_3_1_128 12 45 40999
NN_CUDA_1_3_128_32 13 93 39355
NN_CUDA_1_3_128_128 13 44 38039
NN_CUDA_1_4_1_32 13 87 39239
NN_CUDA_1_4_1_128 13 47 39106
NN_CUDA_1_4_128_32 13 86 37276
NN_CUDA_1_4_128_128 14 36 35656
NN_CUDA_4_3_1_32 13 72 39926
NN_CUDA_4_3_1_128 12 32 40978
NN_CUDA_4_3_128_32 14 88 35536
NN_CUDA_4_3_128_128 15 45 33263
NN_CUDA_4_4_1_32 13 125 39217
NN_CUDA_4_4_1_128 13 59 39125
NN_CUDA_4_4_128_32 15 92 32591
NN_CUDA_4_4_128_128 16 57 30676
NN_CUDA_32_3_1_32 12 44 40666
NN_CUDA_32_3_1_128 12 92 40300
NN_CUDA_32_3_128_32 24 94 21261
NN_CUDA_32_3_128_128 24 91 20478
NN_CUDA_32_4_1_32 12 45 40785
NN_CUDA_32_4_1_128 13 74 39771
NN_CUDA_32_4_128_32 28 46 17927
NN_CUDA_32_4_128_128 30 58 16466

[nikhilaravi]

@xmyqsh we are making several changes to this file and function - your changes will cause conflicts so we will not be able to merge this PR.

Also note that in future, when running benchmarks you can paste a screenshot of the generated table as currently the comment is not readable. We also need to know the type of machine that the benchmarks were run on. Thank you!