submission 551659 · KernelIndex

Use it

Vendorable · source mirrored · license unknownView source →

No package. Vendor the mirrored source: 22 lines, June 9 Researcher Reciprocity License v1.0.

compiled_out_fn.py

curl "https://kernelindex.com/api/v1/implementations/kernelbot-grayscale-v2-551659?include=source"

interfacepython

Compatibility

measured onNVIDIA A100

declared hardwareNVIDIA A100

architecturessm_80

dtypesfp32

Benchmark evidence

1 measurement across 1 GPU, fastest first.

Operation / workload

Hardware

Latency

Rank

Observed

RGB to grayscalesuite of 6 cases

NVIDIA A100

2.55ms

#27 of 137

2026-03-14

Run →

Reported · How evidence levels are derived →

Source and license

sourceavailable

revision digestsha256:ba7c547dec5e89737c7f97f25dfd257b36936f2c1adde0c21331f23ab87dfda3

license declaredunknown

license concludedunknown

authorsidanbeck

imported2026-08-15

Techniques

Extracted from the mirrored source by pattern, never inferred. Each row cites its line.

autotunereturn torch.compile(f, fullgraph=False, mode='max-autotune-no-cudagraphs')

Kernel source

compiled_out_fn.py22 lines

#!POPCORN leaderboard grayscale_v2
#!POPCORN gpu A100
import torch
_F=None

def _build():
    def f(x, out):
        out[...] = x[...,0]*0.2989 + x[...,1]*0.5870 + x[...,2]*0.1140
        return out
    try:
        return torch.compile(f, fullgraph=False, mode='max-autotune-no-cudagraphs')
    except Exception:
        return f

@torch.inference_mode()
def custom_kernel(data):
    global _F
    rgb,out=data
    if _F is None:
        _F=_build()
    return _F(rgb, out)

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 551588.

  #!POPCORN leaderboard grayscale_v2
  #!POPCORN gpu A100
  import torch
- _F = None
+ _F=None
   
-  
  def _build():
-     def f(x):
-         return x[..., 0] * 0.2989 + x[..., 1] * 0.5870 + x[..., 2] * 0.1140
-  
+     def f(x, out):
+         out[...] = x[...,0]*0.2989 + x[...,1]*0.5870 + x[...,2]*0.1140
+         return out
      try:
-         return torch.compile(f, fullgraph=False, mode="max-autotune-no-cudagraphs")
+         return torch.compile(f, fullgraph=False, mode='max-autotune-no-cudagraphs')
      except Exception:
          return f
   
-  
  @torch.inference_mode()
  def custom_kernel(data):
      global _F
-     rgb, out = data
+     rgb,out=data
      if _F is None:
-         _F = _build()
-     out[...] = _F(rgb)
-     return out
+         _F=_build()
+     return _F(rgb, out)

scrolls · 32 diff lines total

Best evidence level for this revision: reported

JSON