NVIDIA/nvbench

View on GitHub

CUDA Kernel Benchmarking Library

Stars: 927
Forks: 123
Open beginner issues: 0
Indexed issues: 63
Avg merge: 2d 10h
Merged PRs (30d): 2

Dominant language: Cuda
License: Apache-2.0
Last GitHub push: Aug 31, 2026
Latest indexed: Sep 19, 2026
Contributing guide: No contributing guide
Code of conduct: Code of conduct
Beginner labels: good first issue

Beginner friendly Everything

Open All

63 open issues indexed

cmake version specified in readme does not work Open

Difficulty 2/5 1-3 hours Newbie friendliness 74/100

NVIDIA/nvbench#465 · Opened Aug 31, 2026
Output of --bulk-debug-python generated by robust comparison script should include configuration values Open

NVIDIA/nvbench#460 · 1 assignee · Opened Aug 13, 2026
Provide an option to reset persisting-L2 state between benchmark cells Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

NVIDIA/nvbench#458 · 2 comments · Opened Aug 11, 2026
Add support for int32 parameter axes Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

NVIDIA/nvbench#436 · 3 comments · Opened Jul 25, 2026
Add MVP instructions on addressing benchmark comparison issues Open

NVIDIA/nvbench#416 · 1 assignee · Opened Jul 17, 2026
Spike on tracking quality metrics for nvbench comparison scripts Open

NVIDIA/nvbench#415 · 1 assignee · Opened Jul 17, 2026
NVBench instrumented benchmarks should always include bulk data in JSON output Open

python

NVIDIA/nvbench#411 · 1 assignee · Opened Jul 16, 2026
Integrate CCCL runtime APIs into nvbench Open

blocked

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#410 · 3 comments · Opened Jul 13, 2026
Add script tool to normalize jsonbin sidecar paths in existing NVBench JSON files Open

Difficulty 4/5 3-5 days Newbie friendliness 52/100

NVIDIA/nvbench#407 · 1 comment · Opened Jul 2, 2026
[Skill]: Report and interpret benchmark results Open

NVIDIA/nvbench#390 · 1 assignee · Opened Jun 9, 2026
[Skill]: Choose the right NVBench timing mode Open

NVIDIA/nvbench#389 · 1 assignee · Opened Jun 9, 2026
[Skill]: Write a correct NVBench benchmark for a CUDA kernel or library call Open

NVIDIA/nvbench#388 · 1 assignee · Opened Jun 9, 2026
Add skills to enable agents to use NVBench more effectively Open

NVIDIA/nvbench#387 · 1 assignee · Opened Jun 9, 2026
Revisit handling dependent DLLs on Windows Open

Difficulty 3/5 1-2 days Newbie friendliness 58/100

NVIDIA/nvbench#369 · 1 comment · 1 reaction · Opened May 19, 2026
Compile the list of public API provided by NVBench library Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#355 · Opened May 7, 2026
Update examples/custom_criterion.cu Open

area: examples P2: nice to have type: cleanup

NVIDIA/nvbench#343 · 1 assignee · Opened May 1, 2026
Enable generation of NVBench documentation Open

area: docs blocked

NVIDIA/nvbench#324 · 1 assignee · Opened Mar 10, 2026
Extend comparison script to allow comparison of two distributions of timing samples Open

NVIDIA/nvbench#313 · 1 comment · 2 reactions · 1 assignee · Opened Feb 3, 2026
Running batch benchmarks affects cold benchmarks Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#310 · 1 comment · Opened Feb 2, 2026
Throughput Failed On Mutiple GPUS Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

NVIDIA/nvbench#287 · 2 comments · Opened Nov 23, 2025
Write-only benchmark exceeds 100% bandwidth Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#270 · 5 comments · 1 reaction · Opened Sep 17, 2025
Show batch time comparisons in `nvbench_compare.py` Open

NVIDIA/nvbench#247 · 1 comment · 1 assignee · Opened Jul 17, 2025
Comparison outputs should be colorful when pasted into GitHub Open

Difficulty 4/5 3-5 days Newbie friendliness 38/100

NVIDIA/nvbench#238 · 4 comments · 3 reactions · Opened Jul 1, 2025
Make bandwidth / size display configurable for either decimal / binary units Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#217 · Opened Apr 30, 2025
BWUtil is computed based on GPU Time, even when Batch GPU is available Open

Difficulty 4/5 3-5 days Newbie friendliness 38/100

NVIDIA/nvbench#194 · 1 comment · Opened Nov 28, 2024
Enable CUPTI to measure kernel execution time instead of CUDA Events Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

NVIDIA/nvbench#184 · 1 comment · 4 reactions · Opened Aug 29, 2024
Extend nvbench to measure SOL for compute-bound workloads Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#180 · 1 comment · 3 reactions · Opened Jul 26, 2024
Rename CUPTI methods for consistency with `state` API Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#177 · Opened Jun 27, 2024
Add documentation for CUPTI integration and include `auto_throughput.cu` in the list of examples in the top-level README.md Open

Difficulty 2/5 1-3 hours Newbie friendliness 35/100

NVIDIA/nvbench#176 · Opened Jun 27, 2024
Throughput statistics are not calculated when reads/writes are declared after `state.exec()` Open

Difficulty 3/5 1-2 days Newbie friendliness 45/100

NVIDIA/nvbench#175 · 1 reaction · Opened Jun 27, 2024
Smart selection of the entropy step Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

NVIDIA/nvbench#154 · Opened Jan 9, 2024
Associate parameters with stopping criterion to allow stateful CLI parsing and improved diagnostics Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

NVIDIA/nvbench#153 · Opened Jan 9, 2024
vcpkg upstream Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

NVIDIA/nvbench#152 · 1 comment · Opened Jan 9, 2024
Tracking `nvbench` benchmark results Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

NVIDIA/nvbench#132 · 5 comments · Opened Apr 17, 2023
Project assumes CXX compiler and and CUDACXXHOST compiler are the same Open

Difficulty 3/5 1-2 days Newbie friendliness 42/100

NVIDIA/nvbench#126 · Opened Feb 15, 2023
Automatic cuda graph testing Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

NVIDIA/nvbench#124 · Opened Feb 9, 2023
Document jsonbin output format Open

area: docs type: enhancement

Difficulty 2/5 1-3 hours Newbie friendliness 25/100

NVIDIA/nvbench#122 · Opened Jan 30, 2023
Custom CUPTI metrics Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

NVIDIA/nvbench#121 · 1 comment · Opened Jan 25, 2023
Bazel build rules Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#120 · 7 comments · Opened Jan 25, 2023
Implement and test NVRTC support Open

area: tests P2: nice to have type: enhancement

Difficulty 5/5 Over a week Newbie friendliness 15/100

NVIDIA/nvbench#119 · 1 comment · 1 reaction · Opened Jan 5, 2023
Compact output format Open

P2: nice to have type: enhancement

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#118 · Opened Jan 5, 2023
Extend `nvbench::..._t` aliases to include CUDA specific types Open

P1: should have type: enhancement

Difficulty 3/5 1-2 days Newbie friendliness 35/100

NVIDIA/nvbench#117 · 1 comment · Opened Jan 5, 2023
How can the output metrics be interpreted Open

area: docs type: enhancement

Difficulty 3/5 1-2 days Newbie friendliness 28/100

NVIDIA/nvbench#115 · 4 comments · Opened Dec 19, 2022
[BUG] Memory allocations from flushing L2 can lead to significant delays between benchmark executions Open

P1: should have

Difficulty 4/5 3-5 days Newbie friendliness 48/100

NVIDIA/nvbench#100 · 2 comments · Opened Sep 3, 2022
Specify the kernel name or nvtx range for benchmarking. Open

P2: nice to have type: enhancement

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#92 · 9 comments · Opened Jul 1, 2022
Add example showing how to parse additional arguments Open

area: examples good first issue P2: nice to have type: enhancement

Difficulty 3/5 1-2 days Newbie friendliness 35/100

NVIDIA/nvbench#87 · 6 comments · Opened May 10, 2022
Reasonable enum defaults don't work Open

P1: should have type: bug: functional

Difficulty 2/5 1-3 hours Newbie friendliness 35/100

NVIDIA/nvbench#84 · 2 comments · Opened Apr 7, 2022
Add hidden benchmarks Open

P2: nice to have type: enhancement

Difficulty 4/5 3-5 days Newbie friendliness 35/100

NVIDIA/nvbench#83 · Opened Apr 7, 2022
Add optional types to type axes Open

P2: nice to have type: enhancement

Difficulty 5/5 Over a week Newbie friendliness 35/100

NVIDIA/nvbench#82 · 4 comments · Opened Apr 7, 2022
Add benchmark groups/sets/tags Open

P2: nice to have type: enhancement

Difficulty 5/5 Over a week Newbie friendliness 25/100

NVIDIA/nvbench#81 · 3 comments · Opened Apr 7, 2022

NVIDIA/nvbench

Get new issues in your inbox