submission 512893
JordanNanos · python · License unknown
Use it
Vendorable · source mirrored · license unknownView source →
No package. Vendor the mirrored source: 58 lines, June 9 Researcher Reciprocity License v1.0.
vectoradd_py_submission.py
curl "https://kernelindex.com/api/v1/implementations/kernelbot-vectoradd-v2-512893?include=source"interfacepython
Compatibility
measured onNVIDIA B200
declared hardwareNVIDIA B200
architecturessm_100
dtypesfp16
Benchmark evidence
1 measurement across 1 GPU, fastest first.
Operation / workload
Hardware
Latency
Rank
Observed
Reported · How evidence levels are derived →
Source and license
sourceavailable
revision digestsha256:429b62eae8e1e4a65a54059668d3ed8d7491311c2103a2f499591afe9584ae20
license declaredunknown
license concludedunknown
authorsJordanNanos
imported2026-08-15
Techniques
Extracted from the mirrored source by pattern, never inferred. Each row cites its line.
autotune
@triton.autotune(num-warps = 4
triton.Config({'BLOCK_SIZE': 1024}, num_warps=4, num_stages=2),stages = 2
triton.Config({'BLOCK_SIZE': 1024}, num_warps=4, num_stages=2),Kernel source
vectoradd_py_submission.py58 lines
import triton
import triton.language as tl
import torch
from task import input_t, output_t
@triton.autotune(
configs=[
triton.Config({'BLOCK_SIZE': 1024}, num_warps=4, num_stages=2),
triton.Config({'BLOCK_SIZE': 1024}, num_warps=8, num_stages=2),
triton.Config({'BLOCK_SIZE': 2048}, num_warps=8, num_stages=2),
triton.Config({'BLOCK_SIZE': 2048}, num_warps=16, num_stages=2),
triton.Config({'BLOCK_SIZE': 4096}, num_warps=16, num_stages=2),
triton.Config({'BLOCK_SIZE': 4096}, num_warps=32, num_stages=2),
triton.Config({'BLOCK_SIZE': 8192}, num_warps=32, num_stages=2),
triton.Config({'BLOCK_SIZE': 8192}, num_warps=16, num_stages=2),
triton.Config({'BLOCK_SIZE': 16384}, num_warps=32, num_stages=2),
triton.Config({'BLOCK_SIZE': 32768}, num_warps=32, num_stages=2),
triton.Config({'BLOCK_SIZE': 1024}, num_warps=4, num_stages=4),
triton.Config({'BLOCK_SIZE': 2048}, num_warps=8, num_stages=4),
triton.Config({'BLOCK_SIZE': 4096}, num_warps=16, num_stages=4),
triton.Config({'BLOCK_SIZE': 4096}, num_warps=32, num_stages=4),
triton.Config({'BLOCK_SIZE': 8192}, num_warps=32, num_stages=4),
triton.Config({'BLOCK_SIZE': 16384}, num_warps=32, num_stages=4),
triton.Config({'BLOCK_SIZE': 32768}, num_warps=32, num_stages=4),
triton.Config({'BLOCK_SIZE': 65536}, num_warps=32, num_stages=2),
triton.Config({'BLOCK_SIZE': 65536}, num_warps=32, num_stages=4),
],
key=['n_elements'],
)
@triton.jit
def vector_add_kernel(
a_ptr,
b_ptr,
c_ptr,
n_elements,
BLOCK_SIZE: tl.constexpr,
):
pid = tl.program_id(axis=0)
block_start = pid * BLOCK_SIZE
offsets = block_start + tl.arange(0, BLOCK_SIZE)
mask = offsets < n_elements
a = tl.load(a_ptr + offsets, mask=mask, cache_modifier=".cg")
b = tl.load(b_ptr + offsets, mask=mask, cache_modifier=".cg")
c = a + b
tl.store(c_ptr + offsets, c, mask=mask, cache_modifier=".cg")
def custom_kernel(data: input_t) -> output_t:
A, B, output = data
n_elements = A.numel()
grid = lambda meta: (triton.cdiv(n_elements, meta['BLOCK_SIZE']),)
vector_add_kernel[grid](
A, B, output,
n_elements,
)
return output
scrolls · 58 lines total
Source code from GPU Mode and the KernelBot dataset · June 9 Researcher Reciprocity License v1.0
Best evidence level for this revision: reported
JSON