cargo / baracuda-transformer-engine
baracuda-transformer-engine 0.0.1-alpha.86
- processes
- 527
- operations
- 17K
- packages
- 60
- compiler runs
- 21
- connections
- 1
- job time
- 25s
- capture
- 219 KB
- records
- 30K
- bytes / record
- 7.3
Safe Rust wrapper for baracuda's port of NVIDIA TransformerEngine's FP8 cast/transpose + delayed-scaling recipe primitives. Provides `Fp8Recipe` (delayed-scaling state with amax history), `Fp8CastPlan` for {f32, f16, bf16} → FP8 with running amax, `Fp8DequantPlan` for FP8 → {f32, f16, bf16}. Cast/recipe subset only — `normalization` / `fused_rope` / `fused_attn` / `fused_softmax` / `activation` / `gemm` skipped (overlap existing baracuda phases). NO cuDNN dep, NO pybind11. On Ada (sm_89) the FP8 wins are bandwidth-saving only (KV cache, weights); FP8 tensor-core math throughput equals BF16. Forward-compatible with Hopper / Blackwell where the compute wins also materialize.
Build behaviournot recorded for this run
Release 0.0.1-alpha.86 run 3748b8d0-d535-4abd-b717-b7f8a7e3b547 · built 2026-10-07
Artifact
- libbaracuda_transformer_engine.rlibfile · retained · product
| Checks | Detections |
|---|---|
| Build behavior | |
| ✓ Network activity | — |
| ✓ Sensitive data access and transfer | — |
| ✓ Cross-process access | — |
| ✓ Build file changes | — |
| ✓ Build script activity | — |
| ✓ Data flow analysis | — |
| ✓ Filesystem writes | — |
| ✓ Context collection | — |
| Supply chain | |
| ✓ Known vulnerabilities | — |
| ✓ Build-time dependency advisories | — |
| ✓ Dependency advisories | — |
| ✓ Declared and observed dependencies | — |
| ✓ Dependency and toolchain versions | — |
| Pipeline | |
| ✓ Pipeline coverage | — |
Detections
0No detections
No abnormal activity detected across 17K flows.
Network
1 peers-
allowedstatic.crates.io:8080
Dependencies
61 packages, 0 with a build script that ran, 0 doing more than compiling| Package | Version | Usage | Build behaviour | |
|---|---|---|---|---|
| baracuda-build | 0.0.1-alpha.86 | observed | ||
| baracuda-core | 0.0.1-alpha.86 | observed | ||
| baracuda-cuda-sys | 0.0.1-alpha.86 | observed | ||
| baracuda-driver direct | 0.0.1-alpha.86 | observed | ||
| baracuda-forge | 0.0.1-alpha.86 | observed | ||
| baracuda-transformer-engine-sys direct | 0.0.1-alpha.86 | observed | ||
| baracuda-types direct | 0.0.1-alpha.86 | observed | ||
| block-buffer | 0.12.1 | observed | ||
| cfg-if | 1.0.4 | observed | ||
| const-oid | 0.10.2 | observed | ||
| cpufeatures | 0.3.1 | observed | ||
| crossbeam-deque | 0.8.7 | observed | ||
| crossbeam-epoch | 0.9.20 | observed | ||
| crossbeam-utils | 0.8.22 | observed | ||
| crunchy | 0.2.4 | observed | ||
| crypto-common | 0.2.2 | observed | ||
| digest | 0.11.3 | observed | ||
| either | 1.18.0 | observed | ||
| float8 | 0.7.0 | observed | ||
| fs2 | 0.4.3 | observed | ||
| glob | 0.3.4 | observed | ||
| half direct | 2.7.1 | observed | ||
| hermit-abi | 0.5.2 | observed | ||
| hybrid-array | 0.4.15 | observed | ||
| itoa | 1.0.18 | observed | ||
| ld-linux-x86-64.so.2 | linked | |||
| libc | 0.2.189 | observed | ||
| libc.so.6 | linked | |||
| libgcc_s.so.1 | linked | |||
| libloading | 0.8.9 | observed | ||
| libm.so.6 | linked | |||
| libmvec.so.1 | linked | |||
| memchr | 2.8.3 | observed | ||
| num_cpus | 1.17.0 | observed | ||
| proc-macro2 | 1.0.107 | observed | ||
| quote | 1.0.47 | observed | ||
| rayon | 1.12.0 | observed | ||
| rayon-core | 1.13.0 | observed | ||
| same-file | 1.0.6 | observed | ||
| serde | 1.0.229 | observed | ||
| serde_core | 1.0.229 | observed | ||
| serde_derive | 1.0.229 | observed | ||
| serde_json | 1.0.151 | observed | ||
| sha2 | 0.11.0 | observed | ||
| syn | 2.0.119 | observed | ||
| syn | 3.0.6 | observed | ||
| thiserror direct | 2.0.20 | observed | ||
| thiserror-impl | 2.0.20 | observed | ||
| typenum | 1.20.1 | observed | ||
| unicode-ident | 1.0.24 | observed | ||
| unpopped-vocab | 0.14.3 | observed | ||
| walkdir | 2.5.0 | observed | ||
| winapi | 0.3.9 | observed | ||
| winapi-i686-pc-windows-gnu | 0.4.0 | observed | ||
| winapi-util | 0.1.11 | observed | ||
| winapi-x86_64-pc-windows-gnu | 0.4.0 | observed | ||
| windows-link | 0.2.1 | observed | ||
| windows-sys | 0.61.2 | observed | ||
| zerocopy | 0.8.56 | observed | ||
| zerocopy-derive | 0.8.56 | observed | ||
| zmij | 1.0.23 | observed |
1 of 1 jobs captured: build-test success · Generated 2026-10-07 · tarp 0.1.0+20261007022833-g16cc1581f632 · run 3748b8d0-d535-4abd-b717-b7f8a7e3b547