Products · Product
CUDA
Raised as a usability problem by 11 people on Hacker News over 11 years (2016–2026), in 12 threads.
No issue was raised by at least 3 people in 3 threads this year.
No issue was raised by at least 3 people in 3 threads in the last 5 years.
Every issue, all time, most people first
Setting up CUDA on Linux is painful and hard to get workingUnreviewed
20162026- User got stuck and found getting CUDA working on Mac confusing2022-09-01
- CUDA install fails with dependency errors and TensorFlow can't find its libraries2021-06-30
- Setting up CUDA from scratch on Linux is painful and hard to get working2020-02-02
Unreviewed issues were grouped by machine: Claude Haiku labelled each comment as a usability complaint, named what it's about and summed up the problem in a line, and a small model grouped similar problems. Nobody has checked them by hand. Expect mistakes.
The newest 12 complaints
- Developers get C++ footguns plus GPU behavior that doesn't match C++ semantics2026-08-11
- Hard to tell which function version (_v2 suffix) to use for a given driver2025-04-04
- Work is pushed to a stream and the thread only learns later whether it worked2024-06-17
- Mixing warp intrinsics with cross-lane ops gets confusing; lane masks must be computed by hand2023-10-15
- User got stuck and found getting CUDA working on Mac confusing2022-09-01
- CUDA install fails with dependency errors and TensorFlow can't find its libraries2021-06-30
- Default download link leads to a release candidate; stable versions buried behind a side link2020-07-08
- Setting up CUDA from scratch on Linux is painful and hard to get working2020-02-02
- Can't find documentation for libcu++, so must dig through SDK samples or headers2019-11-25
- Setting up TensorFlow/Keras with CUDA on Windows is overwhelmingly frustrating2017-03-18
- Their metaphors are too cumbersome to work with compared to MATLAB2016-10-22
- Setup needs matching driver, multi-screen config, and a specific gcc version to work2016-05-15