Triton
Hacker ApprovedRanked 4 of 12 GPU compute frameworks by recommenders
Recommended for writing fast GPU kernels in short Python-like code, often easier than CUDA. Caveat: gains vary a lot by workload.
Recommended for writing fast GPU kernels in short Python-like code, often easier than CUDA. Caveat: gains vary a lot by workload.
20 people gave an opinion
- recommend it
- 18
- criticize it
- 2
- approval share
- 90%
Recommended in 8 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: not recommended
- March 2026: not recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: not recommended
- November 2026: not recommended (month in progress)
Why it carries the badge
- 18 people recommend itNeeds 10 or more, met.
- 9 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 8 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Easier kernel writing than CUDA
“DSLs like Triton that make kernel writing much more ergonomic than anything you would hope to achieve in Rust”
Comment from 16 Sep 2026 in “Nvidia announces native GPU programming in Rust” “Performant Triton programs are usually simpler and shorter than their CUDA equivalents”
Comment from 7 Nov 2025 in “Helion: A high-level DSL for performant and portable ML kernels” “Now you can write very good CUDA kernels with just a few lines of python DSL code.”
Comment from 10 Aug 2026 in “How I use LLMs to learn complex topics” Python-like code that abstracts tile sizes
“If you prefer a python-like approach consider Triton from openai”
Comment from 12 Aug 2026 in “Mojo 1.0 Is Here” “try Triton or a similar DSL that nicely abstract tile sizes etc.”
Comment from 16 Sep 2026 in “Nvidia announces native GPU programming in Rust” Peak performance on Nvidia hardware
“optimized for Nvidia hardware and considered a industry standard”
Comment from 26 Dec 2025 in “CUDA Tile Open Sourced” “it's a path to peak performance for their specific workloads”
Comment from 27 Jun 2026 in “Modern GPU Programming for MLSys”
What people warn about
Each mark is one comment
Results vary by workload
“found that it is very hit or miss how much I will gain over just using torch.compile”
Comment from 7 Jun 2026 in “PyTorch Custom Operation” “how well Triton works varies depending on what you want to do”
Comment from 16 Nov 2025 in “What to Learn? CUDA vs. PyTorch vs. Jax vs. Triton/Pallas”
Other GPU compute frameworks people recommend
The four most recommended other GPU compute frameworks, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| NVIDIA CUDA | 91 | 73 | 55% | 12 | 18 yrs3 owners | — |
| AMD ROCm | 47 | 97 | 33% | 11 | 3 yrs4 owners | — |
| BarraCUDA | 21 | 7 | 75% | 2 | — | — |
| TritonThis page | 18 | 2 | 90% | 8 | — | Approved |
| ZLUDA | 13 | 6 | 68% | 8 | — | Approved |
What people said
All 21 comments that mention it, newest first. Each links to the original comment.
“try Triton or a similar DSL that nicely abstract tile sizes etc.”
“DSLs like Triton that make kernel writing much more ergonomic than anything you would hope to achieve in Rust”
“If you prefer a python-like approach consider Triton from openai”
“We beat cuBLAS, and we beat all of abstractions - Triton, Helion, TLX, Mojo, ThunderKittens, cuteDSL”
“Now you can write very good CUDA kernels with just a few lines of python DSL code.”
“finally have a decent open 3d solution for windows vms”
“I just accepted the ~15% throughput loss for my cloud service”
“it's a path to peak performance for their specific workloads”
“you'd probably write it in triton these days”
“introduces one (several on fact) more DSL, one more tooling ecosystem and solves none of the (long list of) issues related to Python/Numpy programming model”
“found that it is very hit or miss how much I will gain over just using torch.compile”
“effective for less complicated / more self contained type problems”
“The Triton compiler would be another option for kernel writing”
“which apparently make audio models scream”
“It's a fantastic tool, but the learning curve has some sharp edges.”
“tons of tools like Numba, Jax, Triton, etc that you can use to write very fast code”
“optimized for Nvidia hardware and considered a industry standard”
“how well Triton works varies depending on what you want to do”
“One of the main values of Triton is that it significantly expanded the scope of folks who can write kernels”
“Performant Triton programs are usually simpler and shorter than their CUDA equivalents”
“it's ten times harder to debug since it doesn't have the tooling NVIDIA provides”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.