Home | Libraries | People | FAQ | More |
By running our performance test suite multiple times, we can compare the effect
of different compilers: as might be expected, the differences are generally
small compared to say disabling internal use of long
double
. However, there are still gains
to be main, particularly from some of the commercial offerings:
Table 16.5. Compiler Comparison on Windows x64
Function |
Microsoft Visual C++ version 14.0 |
GNU C++ version 5.3.0 |
GNU C++ version 5.3.0 |
Intel C++ C++0x mode version 1600 |
---|---|---|---|---|
assoc_laguerre |
1.16 |
1.21 |
1.06 |
1.00 |
assoc_legendre |
1.57 |
1.80 |
1.16 |
1.00 |
beta |
1.74 |
3.78 |
2.17 |
1.00 |
beta (incomplete) |
1.44 |
3.03 |
1.84 |
1.00 |
cbrt |
3.21 |
3.43 |
3.07 |
1.00 |
cyl_bessel_i |
1.11 |
2.78 |
1.21 |
1.00 |
cyl_bessel_i (integer order) |
1.29 |
3.16 |
1.22 |
1.00 |
cyl_bessel_j |
1.19 |
2.06 |
1.25 |
1.00 |
cyl_bessel_j (integer order) |
1.36 |
2.39 |
1.47 |
1.00 |
cyl_bessel_k |
1.04 |
9.43 |
1.03 |
1.00 |
cyl_bessel_k (integer order) |
1.26 |
10.49 |
1.10 |
1.00 |
cyl_neumann |
16.76 |
2.01 |
1.00 |
15.35 |
cyl_neumann (integer order) |
1.00 |
1.79 |
1.24 |
1.02 |
digamma |
1.20 |
3.75 |
2.15 |
1.00 |
ellint_1 |
1.87 |
1.99 |
1.32 |
1.00 |
ellint_1 (complete) |
2.18 |
2.57 |
1.57 |
1.00 |
ellint_2 |
1.78 |
2.17 |
1.52 |
1.00 |
ellint_2 (complete) |
2.14 |
2.07 |
1.24 |
1.00 |
ellint_3 |
1.94 |
2.91 |
1.65 |
1.00 |
ellint_3 (complete) |
2.05 |
3.07 |
1.46 |
1.00 |
ellint_rc |
1.58 |
2.68 |
1.55 |
1.00 |
ellint_rd |
1.75 |
1.83 |
1.40 |
1.00 |
ellint_rf |
1.69 |
2.40 |
1.57 |
1.00 |
ellint_rj |
1.91 |
2.96 |
1.43 |
1.00 |
erf |
1.77 |
2.92 |
1.92 |
1.00 |
erfc |
1.29 |
3.41 |
2.06 |
1.00 |
expint |
1.39 |
3.07 |
2.14 |
1.00 |
expint (En) |
1.10 |
2.19 |
1.38 |
1.00 |
expm1 |
1.00 |
3.10 |
2.40 |
1.10 |
gamma_p |
1.29 |
2.64 |
1.71 |
1.00 |
gamma_p_inv |
1.36 |
2.47 |
1.91 |
1.00 |
gamma_q |
1.16 |
2.79 |
1.61 |
1.00 |
gamma_q_inv |
1.37 |
2.38 |
1.89 |
1.00 |
ibeta |
1.34 |
3.13 |
1.83 |
1.00 |
ibeta_inv |
1.51 |
2.81 |
2.27 |
1.00 |
ibetac |
1.29 |
2.94 |
1.73 |
1.00 |
ibetac_inv |
1.41 |
2.80 |
1.78 |
1.00 |
jacobi_cn |
1.36 |
3.72 |
2.28 |
1.00 |
jacobi_dn |
1.82 |
3.68 |
2.27 |
1.00 |
jacobi_sn |
1.31 |
3.73 |
2.43 |
1.00 |
laguerre |
1.00 |
1.15 |
1.04 |
1.10 |
legendre |
1.07 |
1.16 |
1.00 |
1.13 |
legendre Q |
1.00 |
1.20 |
1.01 |
1.07 |
lgamma |
1.00 |
3.07 |
1.82 |
1.05 |
log1p |
1.00 |
2.00 |
1.14 |
1.00 |
polygamma |
1.16 |
1.02 |
2.24 |
1.00 |
sph_bessel |
1.17 |
1.55 |
1.09 |
1.00 |
sph_neumann |
1.08 |
2.07 |
1.00 |
1.01 |
tgamma |
1.31 |
4.03 |
2.49 |
1.00 |
tgamma (incomplete) |
1.26 |
2.52 |
1.73 |
1.00 |
trigamma |
1.35 |
2.88 |
1.41 |
1.00 |
zeta |
1.34 |
3.83 |
2.52 |
1.00 |
Table 16.6. Compiler Comparison on linux
Function |
GNU C++ version 5.3.0 |
GNU C++ version 5.3.0 |
---|---|---|
assoc_laguerre |
1.36 |
1.00 |
assoc_legendre |
2.55 |
1.00 |
beta |
4.71 |
1.00 |
beta (incomplete) |
3.12 |
1.00 |
cbrt |
2.53 |
1.00 |
cyl_bessel_i |
3.47 |
1.00 |
cyl_bessel_i (integer order) |
3.38 |
1.00 |
cyl_bessel_j |
2.69 |
1.00 |
cyl_bessel_j (integer order) |
2.59 |
1.00 |
cyl_bessel_k |
6.86 |
1.00 |
cyl_bessel_k (integer order) |
10.17 |
1.00 |
cyl_neumann |
2.48 |
1.00 |
cyl_neumann (integer order) |
2.20 |
1.00 |
digamma |
1.92 |
1.00 |
ellint_1 |
1.94 |
1.00 |
ellint_1 (complete) |
1.86 |
1.00 |
ellint_2 |
2.32 |
1.00 |
ellint_2 (complete) |
1.71 |
1.00 |
ellint_3 |
2.75 |
1.00 |
ellint_3 (complete) |
2.23 |
1.00 |
ellint_rc |
1.57 |
1.00 |
ellint_rd |
2.23 |
1.00 |
ellint_rf |
1.51 |
1.00 |
ellint_rj |
2.33 |
1.00 |
erf |
2.60 |
1.00 |
erfc |
2.68 |
1.00 |
expint |
2.47 |
1.00 |
expint (En) |
1.86 |
1.00 |
expm1 |
1.00 |
1.10 |
gamma_p |
2.42 |
1.00 |
gamma_p_inv |
2.24 |
1.00 |
gamma_q |
2.53 |
1.00 |
gamma_q_inv |
2.45 |
1.00 |
ibeta |
2.71 |
1.00 |
ibeta_inv |
2.58 |
1.00 |
ibetac |
2.66 |
1.00 |
ibetac_inv |
2.44 |
1.00 |
jacobi_cn |
2.77 |
1.00 |
jacobi_dn |
2.80 |
1.00 |
jacobi_sn |
2.86 |
1.00 |
laguerre |
1.09 |
1.00 |
legendre |
1.16 |
1.00 |
legendre Q |
1.18 |
1.00 |
lgamma |
2.20 |
1.00 |
log1p |
1.00 |
1.00 |
polygamma |
3.93 |
1.00 |
sph_bessel |
1.71 |
1.00 |
sph_neumann |
2.15 |
1.00 |
tgamma |
3.69 |
1.00 |
tgamma (incomplete) |
2.35 |
1.00 |
trigamma |
1.64 |
1.00 |
zeta |
2.71 |
1.00 |