-
When looking at the issue of multiple Subscribers a topic that came up was
a nested
```
onNext(
onNext(
onNext(
```
stack.
Problem: As onNext cannot be called more than once there …
-
Hey 👋,
I’m playing around with skuba and noticed that it would be useful for me to deploy nodes in parallel to safe some time. When speaking about worker nodes then it should be possible to use a c…
-
Bump parameters need to be fine tuned to either produce reliable bumps when ships touch, or to give the feel of "enough speed" to bump.
Also, a refactoring of stealing has to be done in order for thi…
-
When reading up on the Chase-Lev deque, I came across this: [US7346753B2 - Dynamic Circular Work-Stealing Queue](https://patents.google.com/patent/US7346753B2/en). Since I assume this patent is now ow…
-
# Crossbeam work-stealing queue | Your awesome title
100만개의 일을 만들어서, 10개의 worker들에게 나눠주기. 방식은 work-stealing으로 worker들이 일을 빼앗아 감. ```rust use std::thread; use std::time::Duration; use crossbeam::dequ…
-
Hi, thanks for your nice work on hard label model stealing. I wish to follow your work. However, it seems the teacher model is missing, which may lead to unfair comparison. Can you kindly share the pr…
-
I just noticed that `fold` calls `identity` many more times than number of threads in thread pool. In case `identity` is a non-trivial operation, it might be quite expensive to call it unnecessarily.
…
-
PR #25896 avoids array allocation when moving to a different domain with the same index set, but only for DefaultRectangular arrays. That PR uses the approach of adding an optional method that array i…
-
We currently rely on [`Runtime#availableProcessors`](https://docs.oracle.com/javase/8/docs/api/java/lang/Runtime.html#availableProcessors--) to size the work-stealing threadpool. The JavaDoc warns tha…
-
No excuse for not comparing against our own prior work. This will (hopefully) give some more motivation for why we attempted to do a work-stealing based approach for multi-device. Will need to tweak t…