Performance Comparison Against the Fastest

[wilzbach]

From @burner on June 9, 2016 15:16

I really like the graphs that your benchmark create.
The next step is to compare the performance against what is out there.
It might also be a good idea to compare against strlen to see if you get memory bound, which would be very good.

Copied from original issue: lodo1995/experimental.xml#14

[wilzbach]

From @lodo1995 on June 10, 2016 7:58

Looks like we are not memory bound...
Also, the Cursor is that slow compared to the parser because the benchmark code asks it to parse the attributes of every element.

[wilzbach]

From @rjmcguire on June 10, 2016 8:5

On Fri, Jun 10, 2016 at 9:58 AM, Lodovico Giaretta <[email protected]

wrote:

Looks like we are not memory bound...
Also, the Cursor is that slow compared to the parser because the
benchmark code asks it to parse the attributes of every element.

[image: plot 1]
https://cloud.githubusercontent.com/assets/13088092/15957758/7f0ca828-2ef1-11e6-8ea0-96cb1d1b1c16.png

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
lodo1995/experimental.xml#14 (comment),
or mute the thread
https://github.com/notifications/unsubscribe/ABU8Ca_5SfacOqmz7eJaABKuJZ8_lnuaks5qKRkogaJpZM4IyFhw
.

Nice!
How are you making that chart? Would like to put this in my benchmark code.

[wilzbach]

From @lodo1995 on June 10, 2016 8:12

@rjmcguire My benchmark driver (source/random_benchmark/random_benchmark.d) has an option to output statistical data (min, avg, max, ...) in csv format. Then another program (source/random_benchmark/csvplot.d) takes the csv file and a lot of options (very specific to my benchmark) and outputs a script (containing both code and data) that can be used to feed gnuplot 5.0 and produce an svg graph.

[wilzbach]

From @rjmcguire on June 10, 2016 8:21

On Fri, Jun 10, 2016 at 10:12 AM, Lodovico Giaretta <
[email protected]> wrote:

@rjmcguire https://github.com/rjmcguire My benchmark driver (
source/random_benchmark/random_benchmark.d) has an option to output
statistical data (min, avg, max, ...) in csv format. Then another program (
source/random_benchmark/csvplot.d) takes the csv file and a lot of
options (very specific to my benchmark) and outputs a script (containing
both code and data) that can be used to feed gnuplot 5.0 and produce an svg
graph.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
lodo1995/experimental.xml#14 (comment),
or mute the thread
https://github.com/notifications/unsubscribe/ABU8CRj2yj_tefCh_LZ8V6RKH8Ds7JYBks5qKRyKgaJpZM4IyFhw
.

Thanks. Was hoping you had some thing like:
benchmark!func(50000, "func.png");