Skip to content
KernelIndex
Search⌘K

submission 859699

IliassSjm · python · License unknown

Use it

Vendorable · source mirrored · license unknownView source →

No package. Vendor the mirrored source: 13 lines, June 9 Researcher Reciprocity License v1.0.

submission_eigh.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-eigh-859699?include=source"
interfacepython
Compatibility
measured onNVIDIA B200
declared hardwareNVIDIA B200
architecturessm_100
dtypesfp32

Benchmark evidence

1 measurement across 1 GPU, fastest first.

Operation / workload
Hardware
Latency
Rank
Observed
NVIDIA B200
54.2ms
#239 of 286
2026-07-06

Reported · How evidence levels are derived →

Source and license

sourceavailable
revision digestsha256:775ec13fb3650c4d3f25fe0c3761c9ad46a9dbcc28dfaea06e822a5db938ceca
license declaredunknown
license concludedunknown
authorsIliassSjm
imported2026-08-26

Kernel source

submission_eigh.py13 lines
"""eigh baseline submission: the reference itself.

Purpose: passing run + reference benchmark time on B200, to measure headroom
before writing a real solver.
"""
import torch
from task import input_t, output_t


def custom_kernel(data: input_t) -> output_t:
    values, vectors = torch.linalg.eigh(data)  # torch returns (L, Q); task wants (Q, L)
    return vectors, values

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

Best evidence level for this revision: reported

JSON