Thread overview | |||||||
---|---|---|---|---|---|---|---|
|
January 15, 2021 How do you benchmark? | ||||
---|---|---|---|---|
| ||||
Currently my benchmark game is weak. It consist of writing std.datetime.stopwatch in code, time in terminal and just run executable multiple times. Then eyeball the results. Im not happy with it so I want to ask the D tribe of what they use? I dont want to write my own tools if I dont have to. |
January 15, 2021 Re: How do you benchmark? | ||||
---|---|---|---|---|
| ||||
Posted in reply to welkam | On Friday, 15 January 2021 at 20:26:22 UTC, welkam wrote: > Currently my benchmark game is weak. It consist of writing std.datetime.stopwatch in code, time in terminal and just run executable multiple times. Then eyeball the results. Im not happy with it so I want to ask the D tribe of what they use? I dont want to write my own tools if I dont have to. https://code.dlang.org/packages/chimpfella I use my own library for it. The dub version should be up to date and working although it's a bit rough around the edges. Basically write a benchmark, then the library takes a range to generate a parameter to measure over. |
January 15, 2021 Re: How do you benchmark? | ||||
---|---|---|---|---|
| ||||
Posted in reply to welkam | On Friday, 15 January 2021 at 20:26:22 UTC, welkam wrote: > Currently my benchmark game is weak. It consist of writing std.datetime.stopwatch in code, time in terminal and just run executable multiple times. Then eyeball the results. Im not happy with it so I want to ask the D tribe of what they use? I dont want to write my own tools if I dont have to. Take a look into at https://dlang.org/blog/2020/03/13/tracing-d-applications/ |
January 16, 2021 Re: How do you benchmark? | ||||
---|---|---|---|---|
| ||||
Posted in reply to welkam | On 15.01.21 21:26, welkam via Digitalmars-d wrote: > Currently my benchmark game is weak. It consist of writing std.datetime.stopwatch in code, time in terminal and just run executable multiple times. Then eyeball the results. Im not happy with it so I want to ask the D tribe of what they use? I dont want to write my own tools if I dont have to. If you're just comparing different approaches within the same program, then there's always https://dlang.org/library/std/datetime/stopwatch/benchmark.html As simple as: auto timings = 1000.benchmark!(foo, bar, baz); |
January 16, 2021 Re: How do you benchmark? | ||||
---|---|---|---|---|
| ||||
Posted in reply to oddp | On Saturday, 16 January 2021 at 00:44:08 UTC, oddp wrote:
> On 15.01.21 21:26, welkam via Digitalmars-d wrote:
>> Currently my benchmark game is weak. It consist of writing std.datetime.stopwatch in code, time in terminal and just run executable multiple times. Then eyeball the results. Im not happy with it so I want to ask the D tribe of what they use? I dont want to write my own tools if I dont have to.
>
> If you're just comparing different approaches within the same program, then there's always https://dlang.org/library/std/datetime/stopwatch/benchmark.html
>
> As simple as: auto timings = 1000.benchmark!(foo, bar, baz);
+1 for benchmark, simple as can be really.
|
Copyright © 1999-2021 by the D Language Foundation