|
|
|
|
|
|
by menaerus
488 days ago
|
|
|
Well, the response really sounded unnatural and llm-ey. If it's not then please take my apology. I write SIMD kernels and the conclusion drawn in the article makes no sense regardless of the fact who wrote it. I don't doubt the observations made in experiments but the hypothesis that the SIMD is slowing down the code. The actual answer is in the disassembly but unfortunately it wasn't shown. |
|
|