apaszke
Follow
Create your own GitHub profile
Sign up for your own profile on GitHub, the best place to host code, manage projects, and build software alongside 50 million developers.
Sign up
Pinned
512 contributions in the last year
Contribution activity
November 1, 2020
apaszke has no activity
yet for this period.
October 2020
Created a pull request in google-research/dex-lang that received 1 comment
Get rid of AnyValue
We didn't really need it anymore, because in case we know the exact contructor of an ADT statically, we no longer use the sum-as-product representa…
+27
−59
•
1
comment
- Add support for call syntax on Python atoms
- Cache CUDA kernels on the device
- Add support for simple parallel reductions
- Make the kernel language more flexible
- Implement allocation lifting for CUDA kernels
- Implement a parallelism flattening pass
- Fix the GPU backend
- Don't decide about loop parallelization in Imp
- Add support for arrays in export mode
- Add export mode
- Make the reduction used in Inference stronger
- Implement transposition for zero table atoms
- Force dependencies between subsequent blocks in LiveOutput
- Execute `%bench`ed blocks multiple times
- Add more benchmarks
- Start a more comprehensive benchmark suite
- Break out from checking mode when no useful information is propagated
- Check that dependent functions are only applied to reducible expressions
- [WIP] Add complex numbers and some trig functions
- Caustic rendering demo
- Cleanup the interface parsing code.
- Code gen for lists and other dependent data constructors
- Adaptive-step Runge Kutta Example
- Add LU decomposition, linear solves, matrix inversion and determinants.
- Allow implicit, class, and linear lambdas using arrow annotations
- Add an explicit syntax for type annotation expressions.
- Prevent data types and constructors from shadowing existing bindings
- Added syntactic sugar for typeclasses.