StarOrpheus / aos-vs-soa

Vector of structures vs vector of int copy (quick benchmark)

Preface

Comparison of vector (MyStruct just a wrapper for int) vs vector copy

Results

i9-9980HK - GCC toolchain with stdlibc++

Benchmark	Time	CPU	Iterations	Throughput
BM_copy_vector_of_structures	576979 ns	576024 ns	942	1.69535G/s
BM_copy_vector_of_structures_memcpy	577748 ns	576573 ns	975	1.69374G/s
BM_copy_vector_of_structures_range_reserve	963221 ns	962010 ns	715	1039.49M/s
BM_copy_vector_of_structures_std_copy	567281 ns	566021 ns	1171	1.72531G/s
BM_copy_vector_of_structures_std_copy_back_inserter	738655 ns	737857 ns	780	1.32351G/s
BM_copy_vector_of_ints	366610 ns	366216 ns	1873	2.66663G/s
BM_copy_vector_of_ints_memcpy	366780 ns	366437 ns	1872	2.66502G/s
BM_copy_vector_of_ints_range_reserve	813321 ns	812090 ns	756	1.20253G/s
BM_copy_vector_of_ints_std_copy	384894 ns	384137 ns	1908	2.54223G/s
BM_copy_vector_of_ints_std_copy_back_inserter	798550 ns	797727 ns	819	1.22418G/s

i9-9980HK - Clang toolchain with libc++

Benchmark	Time	CPU	Iterations	Throughput
BM_copy_vector_of_structures	229171 ns	228937 ns	2894	4.26563G/s
BM_copy_vector_of_structures_memcpy	371435 ns	371077 ns	1786	2.6317G/s
BM_copy_vector_of_structures_range_reserve	1117614 ns	1115697 ns	643	896.301M/s
BM_copy_vector_of_structures_std_copy	1135783 ns	1134178 ns	636	881.696M/s
BM_copy_vector_of_structures_std_copy_back_inserter	380907 ns	380080 ns	1874	2.56936G/s
BM_copy_vector_of_ints	218235 ns	217993 ns	3207	4.47979G/s
BM_copy_vector_of_ints_memcpy	388289 ns	387696 ns	1864	2.51888G/s
BM_copy_vector_of_ints_range_reserve	1123663 ns	1122090 ns	551	891.194M/s
BM_copy_vector_of_ints_std_copy	386928 ns	386154 ns	1906	2.52895G/s
BM_copy_vector_of_ints_std_copy_back_inserter	1149359 ns	1147107 ns	620	871.758M/s