Skip to content

Commit 2fdddef

Browse files
Add compute-backends illustration to homepage.
Four-box diagram explains Scalar, Vector, Specific, and GPU paths we optimise on RISC-V boards, with a matching table and arrow-free SVG asset. Co-authored-by: Cursor <cursoragent@cursor.com>
1 parent ed804c7 commit 2fdddef

4 files changed

Lines changed: 60 additions & 0 deletions

File tree

README.md

Lines changed: 13 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -4,6 +4,19 @@ OpenSolvers explores how open-source scientific software runs on real hardware
44

55
We benchmark **scientific libraries** and **applications** on consumer RISC-V boards through the [EESSI](https://www.eessi.io/) stack — from BLAS kernels up to full app runs — swapping fixed OpenBLAS builds via FlexiBLAS without rebuilding downstream code.
66

7+
## What we optimise on the board
8+
9+
Each RISC-V board exposes several compute paths. We benchmark and tune them independently — often swapping backends at runtime via FlexiBLAS rather than rebuilding every app.
10+
11+
![Scalar, Vector, Specific, and GPU compute paths on the board](assets/images/compute-backends.svg)
12+
13+
| Path | What it is | Examples on our boards |
14+
| ---- | ---------- | ---------------------- |
15+
| **Scalar** | Portable C fallbacks — the correctness baseline | `OPENBLAS_CORETYPE=RISCV64_GENERIC`; VisionFive 2 before the U74 kernel |
16+
| **Vector** | ISA vector extensions (RVV) in shared libs | OpenBLAS `RISCV64_ZVL256B` on SpaceMiT X60; the `gemv_n` bug we fixed |
17+
| **Specific** | Silicon-tuned micro-kernels and custom units | U74 **4×4 DGEMM** on VisionFive 2; X60 **IME** (`smt.vmadot`) int8 on RV2 / F3 |
18+
| **GPU** | Discrete or integrated accelerators | On the roadmap — not yet in our RISC-V board benchmarks |
19+
720
Recent highlights on the Orange Pi RV2 (SpaceMiT X60, RVV): fixing an OpenBLAS `gemv_n` bug restores correctness across [BLAS](scientific-libs/blas.html), [LAPACK](scientific-libs/lapack.html), [ELPA](scientific-libs/elpa.html), [HPL](apps/hpl.html), and [Quantum ESPRESSO](apps/qe.html) — with patched RVV reaching **10.53 GFLOP/s** on Linpack, **1.58×** on a dense eigensolve, and **1.31×** on a 64-atom Si DFT SCF.
821

922
## Scientific libs

assets/images/compute-backends.png

968 KB
Loading

assets/images/compute-backends.svg

Lines changed: 46 additions & 0 deletions
Loading

website.md

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -44,6 +44,7 @@ Nav config: `_config.yml` (`navigation`, `navigation_boards`, `navigation_apps`,
4444

4545
| Date | Decision |
4646
|------|----------|
47+
| 2026-07-15 | Homepage: compute-backends illustration (Scalar / Vector / Specific / GPU) |
4748
| 2026-07-14 | Nav layout: 3 stacked rows — Apps (top), Scientific libs (middle), Boards (bottom) |
4849
| 2026-07-14 | Add Quantum ESPRESSO app page from `opensolvers/benchmarks/qe` |
4950
| 2026-07-14 | IME (X60 `smt.vmadot`) section on RV2 and BPI-F3 board pages |

0 commit comments

Comments
 (0)