submission 71021

Use it

Vendorable · source mirrored · license unknownView source →

Benchmark evidence

1 measurement across 1 GPU, fastest first.

Reported · How evidence levels are derived →

Source and license

Kernel source

import torch

def custom_kernel(data):
    input_tensor, output_tensor = data
    torch.sort(input_tensor, out=(output_tensor, torch.empty_like(input_tensor, dtype=torch.long)))
    return output_tensor

Source code from GPU Mode and the KernelBot dataset · June 9 Researcher Reciprocity License v1.0

Changes from previous submission

Against this author's previous submission submission 71014.

  ⋯ 1 unchanged lines
   
  def custom_kernel(data):
      input_tensor, output_tensor = data
-     output_tensor.copy_(torch.sort(input_tensor, stable=False)[0])
+     torch.sort(input_tensor, out=(output_tensor, torch.empty_like(input_tensor, dtype=torch.long)))
      return output_tensor