submission 545306 · KernelIndex

Use it

Vendorable · source mirrored · license unknownView source →

No package. Vendor the mirrored source: 29 lines, June 9 Researcher Reciprocity License v1.0.

sort_v5.py

curl "https://kernelindex.com/api/v1/implementations/kernelbot-sort-v2-545306?include=source"

interfacepython

Compatibility

measured onNVIDIA H100

declared hardwareNVIDIA H100

architecturessm_90

dtypesfp32

Benchmark evidence

1 measurement across 1 GPU, fastest first.

Operation / workload

Hardware

Latency

Rank

Observed

Sortsuite of 5 cases

NVIDIA H100

6.29ms

#10 of 26

2026-03-13

Run →

Reported · How evidence levels are derived →

Source and license

sourceavailable

revision digestsha256:3ed966ff447fa6b97f64266e3dda7757f733ccb834a1f4163b4ad0dc22902ad8

license declaredunknown

license concludedunknown

authorsrajesh0042

imported2026-08-15

Kernel source

sort_v5.py29 lines

import os
os.environ["CUBLAS_WORKSPACE_CONFIG"] = ":4096:8"

import torch
from task import input_t, output_t

# Sort with stable=False for potential speedup
# The reference uses torch.sort(data)[0] which defaults to stable=False
# Pre-allocate both values and indices buffers

_vals_cache = {}
_idx_cache = {}

def _warmup():
    for size in [1024, 4096, 16384, 65536, 262144]:
        x = torch.randn(size, device='cuda', dtype=torch.float32)
        torch.sort(x, stable=False)
    torch.cuda.synchronize()

_warmup()

def custom_kernel(data: input_t) -> output_t:
    data, output = data
    n = data.numel()
    if n not in _idx_cache or _idx_cache[n].device != data.device:
        _idx_cache[n] = torch.empty(n, device=data.device, dtype=torch.int64)
    torch.sort(data, stable=False, out=(output, _idx_cache[n]))
    return output

scrolls · 29 lines total

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 545264.

  ⋯ 3 unchanged lines
  import torch
  from task import input_t, output_t
   
- # Pre-allocate index buffer for sort
+ # Sort with stable=False for potential speedup
+ # The reference uses torch.sort(data)[0] which defaults to stable=False
+ # Pre-allocate both values and indices buffers
+  
+ _vals_cache = {}
  _idx_cache = {}
   
+ def _warmup():
+     for size in [1024, 4096, 16384, 65536, 262144]:
+         x = torch.randn(size, device='cuda', dtype=torch.float32)
+         torch.sort(x, stable=False)
+     torch.cuda.synchronize()
+  
+ _warmup()
+  
  def custom_kernel(data: input_t) -> output_t:
      data, output = data
      n = data.numel()
      if n not in _idx_cache or _idx_cache[n].device != data.device:
          _idx_cache[n] = torch.empty(n, device=data.device, dtype=torch.int64)
-     torch.sort(data, out=(output, _idx_cache[n]))
+     torch.sort(data, stable=False, out=(output, _idx_cache[n]))
      return output

scrolls · 28 diff lines total

Best evidence level for this revision: reported

JSON