Add Gradle Wrapper and clean up docs a bit #28

Merged
breandan merged 4 commits from dev into dev 2018-12-28 07:51:48 +03:00
breandan commented 2018-12-28 06:04:54 +03:00 (Migrated from github.com)

After reading through the docs, things are starting to make a bit more sense. It might help to break up some of these classes and interfaces into separate files for easier referencing in Markdown.

After reading through the docs, things are starting to make a bit more sense. It might help to break up some of these classes and interfaces into separate files for easier referencing in Markdown.
breandan (Migrated from github.com) reviewed 2018-12-28 07:36:10 +03:00
breandan (Migrated from github.com) commented 2018-12-28 07:36:10 +03:00

It's not yet clear to me how BufferSpec is supposed to work. Maybe it would help to provide some examples?

It's not yet clear to me how `BufferSpec` is supposed to work. Maybe it would help to provide some examples?
altavir (Migrated from github.com) reviewed 2018-12-28 07:51:37 +03:00
altavir (Migrated from github.com) commented 2018-12-28 07:51:37 +03:00

The performance of ND-Structures is based on performance on underlying linear structures (It is currently called Buffer because all other names are taken). BufferSpec allows to create a non-boxing linear structure of objects of any complexity. Particular test is here: https://github.com/altavir/kmath/blob/dev/kmath-core/src/jvmTest/kotlin/scientifik/kmath/structures/ComplexBufferSpecTest.kt and performance test is here: https://github.com/altavir/kmath/blob/dev/kmath-jmh/src/jmh/kotlin/scientifik/kmath/structures/BufferBenchmark.kt. It removes memory storage overhead. but it should be studied further about boxing-on-read overhead. Also, I will need to port it to multi-platform.

The performance of ND-Structures is based on performance on underlying linear structures (It is currently called `Buffer` because all other names are taken). `BufferSpec` allows to create a non-boxing linear structure of objects of any complexity. Particular test is here: https://github.com/altavir/kmath/blob/dev/kmath-core/src/jvmTest/kotlin/scientifik/kmath/structures/ComplexBufferSpecTest.kt and performance test is here: https://github.com/altavir/kmath/blob/dev/kmath-jmh/src/jmh/kotlin/scientifik/kmath/structures/BufferBenchmark.kt. It removes memory storage overhead. but it should be studied further about boxing-on-read overhead. Also, I will need to port it to multi-platform.
Sign in to join this conversation.
No description provided.