Decision Workspace
sgemm-bi vs burn-cuda vs torsh-backend
Side-by-side comparison of Rust crates
52
sgemm-bi
experimentalv0.1.2
Deterministic, batch-invariant CUDA GEMM engine with a full training triad (forward, dW, dX) in f32 / bf16 / f16, plus an opt-in tensor-core tier that is faster than cuBLAS PEDANTIC. Bit-identical results across runs; fixed reduction order; no atomics; no cuBLAS dependency.
74
burn-cuda
growingv0.22.0
CUDA backend for the Burn framework
61
torsh-backend
growingv0.2.0
Backend abstraction layer for ToRSh
Core Metrics
| sgemm-bi | burn-cuda | torsh-backend | |
|---|---|---|---|
| Health Score | 52 | 74 | 61 |
| Total Downloads | 57 | 1.3M | 2.4K |
| 30d Downloads | 0 | 0 | 0 |
| Dependents | 0 | 368 | 35 |
| Releases | 3 | 28 | 9 |
| Last Updated | 69d ago | 3d ago | 57d ago |
| Age | 3m | 2y 1m | 1y |
Health Breakdown
sgemm-bi
Maintenance
15
Quality
14
Community
6
Popularity
2
Documentation
15
burn-cuda
Maintenance
20
Quality
17
Community
17
Popularity
7
Documentation
13
torsh-backend
Maintenance
19
Quality
13
Community
10
Popularity
4
Documentation
15
Technical Details
| sgemm-bi | burn-cuda | torsh-backend | |
|---|---|---|---|
| Version | 0.1.2 | 0.22.0 | 0.2.0 |
| Stable (≥1.0) | ✗ No | ✗ No | ✗ No |
| License | MIT OR Apache-2.0 | MIT OR Apache-2.0 | Apache-2.0 |
| Dependencies | 2 | 4 | 35 |
| Crate Size | 105KB | 37KB | 712KB |
| Features | 1 | 7 | 14 |
| Yanked % | 0.0% | 0.0% | 0.0% |
| Edition | 2024 | 2024 | 2021 |
| MSRV | 1.94 | — | 1.87 |
| Owners | 1 | 3 (team) | 1 |
Links
Quick Verdict
- •burn-cuda leads with a health score of 74/100, but none of the options score above 80.
- •burn-cuda is depended on by 368 crates — strongest ecosystem trust.