crowwork

9 ms·

1. 61.▲
   by crowwork 9y ago

   as far as i know weld is optimized for data analytics workload so far(dataframe) while tvm is optimized for tensor deep learning workloads with great cpu, gpu and other supports
2. 62.▲
   Optimize Deep Learning GPU Operators with TVM: A Depthwise Convolution Example(tvmlang.org)

   24 points by crowwork 9y ago|3 comments
3. 63.▲
   by crowwork 10y ago

   it means you can declare gpu array like those in numpy/torch, write them imperatively from python side, and mix them with the graph computation, instead of forcing everything to be part of a graph