A kernel in C and x86-64 Assembly that performs a dot product between vector A and vector B.

Performance of the Kernels (debug mode)

Each implementation was run 30 times for vector sizes n = 2²⁰, 2²⁴, and 2³⁰, and the average execution time was calculated. This includes a C version and an x86-64 assembly language version.

C Kernel (cfunc.c file)

dot_prod_c(...)

n	ave time
220	0.003267 sec
224	0.051800 sec
230	64.089600 sec

x86-64 Kernel (asmfunc.asm file)

dot_prod_asm(...)

n	ave time
220	0.000967 sec
224	0.015400 sec
230	62.540967 sec

The x86-64 implementation had a lower average execution time than the C implementation for the dot product operation. This suggest that the x86-64 runs faster than the C implementation.

Correctness Check of Dot Product

Dot Product from C vs. Dot Product from x86-64

✔ produces the same results

NOTE: displayed times are an average from 30 consecutive runs

Performance of the Kernels (release mode)

C Kernel (cfunc.c file)

dot_prod_c(...)

n	ave time
220	0.000933 sec
224	0.014800 sec
230	66.016167 sec

x86-64 Kernel (asmfunc.asm file)

dot_prod_asm(...)

n	ave time
220	0.000933 sec
224	0.014533 sec
230	64.533533 sec

Similar to the trials performed in debug mode, the x86-64 implementation once again had a lower average execution time than the C implementation for the dot product operation. This further supports the idea that the x86-64 implementation is a better performing one.

Correctness Check of Dot Product

Dot Product from C vs. Dot Product from x86-64

✔ produces the same results