-
Notifications
You must be signed in to change notification settings - Fork 1
Sample ~ Efficient Collections
Vlad Ureche edited this page May 9, 2015
·
31 revisions
Come back soon for the final page.
> ildl-benchmarks/runMain ildl.benchmark.hamming.BenchmarkRunner
...
Parameters(bench -> direct, jvm_interpreter -> true): 271.93535
Parameters(bench -> adrt__, jvm_interpreter -> true): 151.61195
...
Parameters(bench -> direct, jvm_interpreter -> false): 4.937213
Parameters(bench -> adrt__, jvm_interpreter -> false): 1.652576
In a table:
Size | Target | Direct (ms) | ADRT (ms) |
---|---|---|---|
10K elts |
interpreter | 272 |
4.94 |
10K elts |
compiled | 152 |
1.65 |
Conclusion: Using the nested ADRT scopes allows the code to run between 1.79x
and 3x
faster.
The `ildl-plugin` is a meta-programming technique aimed at allowing safe, custom transformations across library boundaries. Using `ildl`-based transformations, we were able to obtain speedups in excess of 20x and have optimized code across a wide range of use-cases.
**Return to the main page** or **return to the OOPSLA Step by Step Guide**
**Return to the main page** or **return to the OOPSLA Step by Step Guide**