Dmitry Babokin

Results 37 issues of Dmitry Babokin

Several warnings are needed for numeric literals not fitting the target type: - integer literals exceeding the target type (example: large integer literal is used to initialize floating point value...

ISPC has too frivolous definition for integer suffixes in parser: `INT_NUMBER (([0-9]+)|(0[xX][0-9a-fA-F]+)|(0b[01]+))[uUlL]*[kMG]?[uUlL]*` ([link](https://github.com/ispc/ispc/blob/fd297fa67ce52ce7d46da137ab3d2281a26d79c7/src/lex.ll#L370)) It has the following problems: 1. it doesn't limit the number of occurrences of [uUlL] symbols, for...

Bugs
Good First Issue

We are using reStructuredText for our documentation, so would be very helpful to render ISPC code snippets with proper syntax highlighting. The "default" way to deal with RST format is...

Features
Good First Issue
Quality of Life

Github uses [Linguist](https://github.com/github/linguist) to highlight the code and detect the programming language to show in the `Languages` chart in the bottom of the right column of every repo. To add...

Features
Quality of Life

[Vundle](http://github.com/VundleVim/Vundle.Vim), the plug-in manager for Vim, needs a repository with a grammar on Github to be able to use it as a plug-in. There are several repos on Guthub with...

Features
Quality of Life

[Opaque Pointers](https://llvm.org/docs/OpaquePointers.html) is a controversial evolution of LLVM with unclear benefits (the declared benefits are more clear/simpler semantics, fewer number of bitcasts), which moves forward on unpredictable schedule (the most...

LLVM bug

It would be good to have a M1 native build in the release artifacts. It can be done right now, when building on M1 macOS. But more user friendly solution...

Features

Doing performance fine tuning, it may be beneficial to have separate code paths if the mask is "all on". Right now there are multiple ways of expressing this paradigm, which...

Good First Issue

[This StackOverflow article](https://stackoverflow.com/questions/36932240/avx2-what-is-the-most-efficient-way-to-pack-left-based-on-a-mask) describes an algorithm for `packed_load_active`/`packed_load_active2` for AVX2. The algorithm produces a vector register, so it will need to be stored with a mask and the number of...

Performance
Good First Issue

Since ISPC 1.17, we have `float16` type (IEEE 754 [half precision](https://en.wikipedia.org/wiki/Half-precision_floating-point_format) type), which is now native ISPC type. This means that it should be supported for by all built-in operations...

Features