Investigate why Enso is still slow on sieve example

[wdanilo]

This task is automatically imported from the old Task Issue Board and it was originally created by jaroslavtulach.
Original issue is here.

---

This is a continuation of https://www.pivotaltracker.com/story/show/182485940/comments/235284607 - it was known that

Rewriting the Filter linked list to JavaScript makes the execution 50% faster! Compare: jtulach/sieve@32186dd

it was believed that using unboxed atoms will get us closer to the speed of JavaScript. But it doesn't

My comment concluded: I rewrote the sieve.enso and I know that it is using Layout_Atom_1_1 to store the Filter.Item, but it still takes more than 300ms.

The goal of this task is to investigate why we are still over 300ms and not down to 150ms.

Tasks:

• Check IGV graphs
• Try to rewrite (different) parts of the code to JavaScript
• Nail down where the problem is
• Schedule subsequent work

Blockers:

#182485940 blocked

[Akirathan]

Today, on the latest develop:

Let's keep this issue still open.

[JaroslavTulach]

Things haven't been improved and we may get bitten by this problem when we use JavaScript for something important. Anyway I've just realized that Sieve & JavaScript benchmarks do not seem to be running.

[Akirathan]

Anyway I've just realized that Sieve & JavaScript benchmarks do not seem to be running.

There is a failure in stdlib benchmarks when running Sieve.Java_Script_File, the latest benchmark run with the failure is at https://github.com/enso-org/enso/actions/runs/12226462185/job/34101757695#step:7:8321

Created new issue for that - #11808

GitHub

Benchmark Standard Libraries · enso-org/enso@81b0fcc

Enso Analytics is a self-service data prep and analysis platform designed for data teams. - Benchmark Standard Libraries · 81b0fcc