Metraipolis › archive › dc_413cc24dad4f

MFSA: A Multi-Format Systolic Array with Native Micro Scaling Support — arXiv:2610.19645

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.19645
License (as archived)
CC0-1.0
Archived
2026-10-15T14:08:16.127Z
SHA-256 of the archived text
11e7079b41ed8259ecab0b235f1e19710dacd282c4d3b7fa693833976f59154c
Stable URL of this page
https://metraipolis.com/documento/dc_413cc24dad4f

How to cite

MFSA: A Multi-Format Systolic Array with Native Micro Scaling Support — arXiv:2610.19645 — arXiv (en), https://arxiv.org/abs/2610.19645 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_413cc24dad4f (Metraipolis archive dc_413cc24dad4f, sha256 11e7079b41ed8259ecab0b235f1e19710dacd282c4d3b7fa693833976f59154c)
BibTeX
@misc{dc413cc24dad4f,
  title = {MFSA: A Multi-Format Systolic Array with Native Micro Scaling Support — arXiv:2610.19645},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.19645},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_413cc24dad4f; sha256 11e7079b41ed8259ecab0b235f1e19710dacd282c4d3b7fa693833976f59154c},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.19645. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

AI workloads increasingly employ low-precision formats such as BF16 / FP16, FP8 variants, FP4, Micro Scaling (MX), and INT modes across models, layers, and execution stages to balance accuracy, memory footprint, and throughput, which in turn requires hardware to support these formats in a flexible and efficient way. We present MFSA, a weight-stationary multi-format systolic array that achieves area efficiency through two architectural choices. At the PE level, MFSA uses dot-product-accumulator-based processing elements extended from an existing open-source trans-precision FPU to support reduced-precision formats (FP32/FP16/BF16/CFP8/FP4/INT16/INT8/INT4) with a shared datapath and proportional throughput scaling. At the array level, MFSA employs Online Block-Scale Rebasing (OBSR) for block-scaled formats, which folds each block scale into the running FP32 partial sum at block boundaries without adding a second cross-block accumulator, thereby reducing the additional overhead of block-scaling support. In TSMC 28 nm at 1 GHz, matched-throughput BF16/FP8/FP4 cascaded FMA-PE baselines are 1.5x/2.5x/4.5x larger than the single-format DPA-PE, the gap growing with subword parallelism. A routed 16x16 MFSA reaches a peak 0.5/1/2/4 TOPS of FP32/BF16/FP8/FP4, and OBSR cuts the incremental MX overhead by 64.2% in area and 67.5% in energy versus a naive MX implementation baseline, without hurting the numerical stability on real LLM traces. Adding support for all other formats based on a BF16+CFP8 systolic array costs 25.7% additional area. OBSR also extends to non-power-of-two block scales: folding the fractional part of an E4M3 scale into the MFPE dot product supports NVFP4 alongside OCP MX for 11.8% area and 7.3% power over a CFP8+FP4 base, less than half the cost of a naive implementation.

Tema degli esploratori: calcolo e infrastrutture (elenco arXiv cs.AR, nuovi lavori del giorno).
Categorie arXiv: Hardware Architecture (cs.AR).
Pagina del lavoro: https://arxiv.org/abs/2610.19645 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-15T13:50:35.380Z dalla pagina https://arxiv.org/list/cs.AR/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).