Sparse matrix operations for AWS Trainium via NKI.
CSR/COO formats, SpMV, SpMM, and integral screening for sparse scientific computing on Trainium. Part of the trnsci scientific computing suite (github.com/trnsci).
trnsparse follows the trnsci 5-phase roadmap. Active work is tracked in phase-labeled GitHub issues:
- Phase 1 — correctness ✅ v0.2.0: NKI SpMM validated on trn1 via densify-then-GEMM (see
trnsci/trnsci#3). - v0.3.0 ✅
BSRMatrix— Trainium-native 128×128 block-sparse format;bsr_spmmNKI kernel. CSR becomes interop; BSR is the compute path. - v0.3.2 ✅
cg_bsr,power_iteration_bsr— iterative solvers over BSR (Python loop; fused kernel gated on NKI capability). - v0.4.0 ✅
screened_spmm— fused Schwarz-screened SpMM in one NKI dispatch. - v0.4.2 ✅ Block-sparse attention —
BSRMatrix+bsr_spmmas the primitive;examples/block_sparse_attention.py+docs/sparse_attention.md. - v0.4.3 ✅ Architecture-friendly alternatives:
chebyshev_bsr/richardson_bsr(fixed-K solvers, no inner products);block_sparse_attention_tiled(two-pass, no O(seq²) intermediate). - Phase 3 — perf: nnz-bucketing, NKI attention kernel pair — parked on NKI indirect DMA gather / hardware validation.
- Phase 4 — multi-chip: sharded BSR across NeuronCores.
- Phase 5 — generation: trn2 DMA bandwidth exploitation.
(No Phase 2 for trnsparse — precision inherited from trnblas.)
Suite-wide tracker: trnsci/trnsci#1.
pip install trnsparseimport torch
import trnsparse
# Dense → sparse
A = torch.randn(100, 100)
A[torch.abs(A) < 1.0] = 0.0
csr = trnsparse.from_dense(A)
# SpMV: y = A @ x
y = trnsparse.spmv(csr, x, alpha=2.0)
# SpMM: C = A @ B
C = trnsparse.spmm(csr, B)
# Integral screening
Q = trnsparse.schwarz_bounds(diagonal_integrals)
mask = trnsparse.screen_quartets(Q, threshold=1e-10)
stats = trnsparse.sparsity_stats(Q)| Operation | Description |
|---|---|
spmv |
Sparse × dense vector (CSR) |
spmm |
Sparse × dense matrix (CSR, PyTorch fallback) |
bsr_spmm |
Block-sparse × dense (BSR-native, Tensor Engine) |
screened_spmm |
Fused Schwarz-screened matmul (one NKI dispatch) |
spmv_symmetric |
Symmetric SpMV (half storage) |
sparse_add |
C = αA + βB |
sparse_scale |
B = αA |
sparse_transpose |
Aᵀ |
cg_bsr |
Conjugate Gradient on BSR matrix |
chebyshev_bsr |
Fixed-K Chebyshev semi-iteration (no inner products) |
richardson_bsr |
Fixed-K Richardson iteration |
power_iteration_bsr |
Dominant eigenpair via power iteration |
jacobi_preconditioner_bsr |
Diagonal preconditioner for cg_bsr |
bsr_diagonal |
Extract main diagonal from BSR matrix |
block_sparse_attention_tiled |
Two-pass sparse attention, no O(seq²) intermediate |
schwarz_bounds |
Schwarz screening bounds |
screen_quartets |
Shell quartet significance mask |
density_screen |
Density-weighted screening |
Apache 2.0 — Copyright 2026 Scott Friedman
trnsci is an independent open-source project. It is not sponsored by, endorsed by, or affiliated with Amazon.com, Inc., Amazon Web Services, Inc., or Annapurna Labs Ltd.
"AWS", "Amazon", "Trainium", "Inferentia", "NeuronCore", "Neuron SDK", and related identifiers are trademarks of their respective owners and are used here solely for descriptive and interoperability purposes. Use does not imply endorsement, partnership, or any other relationship.
All work, opinions, analyses, benchmark results, architectural commentary, and editorial judgments in this repository and on trnsci.dev are those of the project's contributors. They do not represent the views, positions, or commitments of Amazon, AWS, or Annapurna Labs.
Feedback directed at the Neuron SDK or Trainium hardware is good-faith ecosystem commentary from independent users. It is not privileged information, is not pre-reviewed by AWS, and should not be read as authoritative about product roadmap, behavior, or quality.
For official AWS guidance, see aws-neuron documentation and the AWS Trainium product page.