yum-archive/TaSTT-Whisper

High-performance GPGPU inference of OpenAI's Whisper automatic speech recognition (ASR) model

git clone https://git.yummers.dev/yum-archive/TaSTT-Whisper

KonstantinPerformance improvement, `softMax` shader27dfc34

master
302 B6 linesraw
1// This version is for the "dec.probs" shader tag
2// The input tensor has a size [ 51865, 3 ], a very long tensor with just 3 rows.
3// Despite the shader only runs on 3 GPU cores, large count of threads helps substantially, this shader is about 50% faster.
4#define THREADS 1024
5
6#include "softMax.hlsl"