mycpp Code Generation

Measure the speedup from mycpp, and the resource usage.

Source code: oil/mycpp/examples

User Time (milliseconds)

Lower ratios are better.

example name C++ Python C++ : Python
fib_recursive 4 627 0.006
fib_iter 5 651 0.008
modules 1 125 0.010
loops 3 208 0.013
asdl_generated 4 275 0.014
parse 17 553 0.030
scoped_resource 34 720 0.047
tuple_return_value 11 133 0.080
files 5 54 0.096
containers 8 84 0.101
classes 2 14 0.168
length 27 147 0.186
cartesian 64 235 0.273
escape 120 244 0.494
cgi 189 365 0.518
varargs 17 16 1.019
control_flow 147 78 1.881

Max Resident Set Size (MB)

Lower ratios are better. We use MB (powers of 10), not MiB (powers of 2).

example name C++ Python C++ : Python
classes 4.5 10.9 0.41
parse 3.9 7.7 0.51
scoped_resource 3.8 7.3 0.52
modules 3.8 7.3 0.52
cartesian 3.7 7.1 0.52
asdl_generated 3.7 7.1 0.52
escape 3.7 7.0 0.53
cgi 3.7 7.1 0.53
loops 3.8 7.2 0.53
fib_iter 3.8 7.1 0.53
fib_recursive 3.8 7.1 0.53
length 3.8 7.1 0.53
control_flow 3.9 7.2 0.54
files 3.9 7.1 0.55
tuple_return_value 3.9 7.1 0.55
containers 28.5 47.7 0.60
varargs 5.6 7.1 0.79

System Time (milliseconds)

Lower ratios are better.

example name C++ Python C++ : Python
cartesian 0 8 0.000
classes 0 7 0.000
control_flow 0 4 0.000
loops 0 16 0.000
modules 0 4 0.000
scoped_resource 0 4 0.000
containers 4 16 0.264
parse 4 8 0.524
escape 6 8 0.793
tuple_return_value 4 4 0.881
varargs 29 33 0.891
asdl_generated 4 4 0.945
fib_recursive 4 4 0.947
length 4 4 0.982
cgi 0 0 NA
fib_iter 0 0 NA
files 0 0 NA

raw benchmark files