HN Explorer

binarybana

↗ HN profile [ 244 ms ] full profile
Karma
0
Created
()
Submissions
0
  1. Using TVM to generate portable sparse GPU kernels 5x faster than cuBLAS/cuSPARSE (medium.com)
    2 points 0
  2. Open source ML compiler delivers 30% faster BERT on Apple M1 than CoreML (medium.com)
    9 points 1
  3. Google ?= Exxon
    2 points 5