64 karma · joined July 14, 2020
This is not the first championship in branch prediction, but real-world design constraints have never been seriously considered before (How long does it take to produce a prediction? How much energy does it consume?). We wrote a new C++ library which uses operator overloading to track the latency/energy used by math operations as the predictor predicts. In addition to computation, this library models registers, RAMs, etc. The championship is open to everyone - documentation, a tutorial, etc. are on our website/repository to help you get started!
I read through the ucad website and book for 10 minutes and haven't been able to figure out if there is an analogue to this for ucad?
There are several things that look neat about ucad's language, but I would need to recreate something like openscad's workflow to consider switching.
As you'll find in the results of the recent branch prediction competition, global branch history (a record of which branches have recently been taken) is a piece of context critical to achieving the high accuracy of modern branch predictors: https://ericrotenberg.wordpress.ncsu.edu/cbp2025-workshop-pr...
And this decidedly is not a square not: https://www.thiings.co/things/square-knot
I had read the link and it was not obvious upon reading it why contributing back to Project Gutenberg did not make sense for them. In particular, I did not understand why it would not be desirable to contribute back corrections to the text to the "upstream" and original source so that others could also benefit - I did not see any contradiction between doing so and the goals/benefits stated on their page.
What I would really like is to somehow extract out a representation of the tax logic into something separate from the business of collecting the data from the user, or even calculating the results. I'm not entirely sure what that would look like, maybe it would have to be some domain-specific language (eww) in order to be able to fully express the relationships and dependencies between different fields/forms.
But I think that tax logic portion is the difficult-to-maintain part, and also the portion different projects would benefit the most from sharing.