-
Okay so say you have a list and a number and you want to add that number to each whatever in the list of stuff so you go to use `+` but it be like "sorry buckaroo, but that's cringe you can't concaten…
lyxal updated
2 years ago
-
We use marching squares/ tile based collision resolution, currently via branches.
- [ ] Can we do the same with a lookup table (function pointer valued?) for each tile code?
-
Hi, fantastic work - it's making our code a lot faster! Having said that we got a bit stuck of generating an array/matrix of numbers from binomial distribution for different trials and probabilities. …
-
Clang tip of tree is 120% behind GCC for kernel s128 in TSVC. Compile this input with `-O3 -ffast-math -mcpu=neoverse-v2`:
```
__attribute__((aligned(64))) float x[32000];
__attribute__((aligne…
-
This is a placeholder for discussion. I am not 100% up-to-date with what optimisations are done on the AST level, but catching up :)
Consider the following kernel that is vectorised with vector wid…
-
I've benchmarked the OpenMP task, OpenMP loop and serial versions of the code now on SFP.
All results are with
`-O3 -g -xCORE_AVX512 -fno-omit-frame-pointer -no-inline-min-size -no-inline-max-per-c…
-
| | |
|--------------------|----|
| Bugzilla Link | [PR31691](https://bugs.llvm.org/show_bug.cgi?id=31691) |
| Status | NEW |
| Importance | P normal |
|…
-
Looks like we are 1400% (?!) behind for kernel s231 in TSVC compared to GCC.
Compile this code with `-O3 -mcpu=neoverse-v2 -ffast-math`:
```
__attribute__((aligned(64))) float a[32000],b[32000],…
-
Hi all!
Thanks for the work on this neat module! I've run into a couple of situations where I want to do operations with many regions e.g. if `SkyCoord` falls inside a 'list' of regions. 'Air quot…
-
Hi Kyle
First of all thanks for this great package!
I am having trouble to get the vectorization to work as per your example
I have a pandas date frame with a 'Date' column of (datetimes) and a lo…