Metraipolis › archive › dc_09c5d41ef2c1
Implementation and Evaluation of BitNet Inference on a CGLA by Signed-Int4 Instructions — arXiv:2609.27453
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.27453
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-08T10:07:10.910Z
- SHA-256 of the archived text
41cf2c4bfb9f9567de57b472769dd35771d33d2d9d739c43525d2fd1f70a3890- Stable URL of this page
https://metraipolis.com/documento/dc_09c5d41ef2c1
How to cite
Implementation and Evaluation of BitNet Inference on a CGLA by Signed-Int4 Instructions — arXiv:2609.27453 — arXiv (en), https://arxiv.org/abs/2609.27453 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_09c5d41ef2c1 (Metraipolis archive dc_09c5d41ef2c1, sha256 41cf2c4bfb9f9567de57b472769dd35771d33d2d9d739c43525d2fd1f70a3890)
BibTeX
@misc{dc09c5d41ef2c1,
title = {Implementation and Evaluation of BitNet Inference on a CGLA by Signed-Int4 Instructions — arXiv:2609.27453},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.27453},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_09c5d41ef2c1; sha256 41cf2c4bfb9f9567de57b472769dd35771d33d2d9d739c43525d2fd1f70a3890},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.27453. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Large language model (LLM) inference transfers model weights and activations for every generated token, making memory traffic and its energy cost part of the decode path. BitNet b1.58 represents its low-bit weights by ternary values and uses integer activations. However, this arithmetic does not match conventional int8 or floating-point general matrix multiplication, and existing BitNet accelerators implement it in specialized datapaths. We instead map this operation to a CPU-Grounded Linear Array (CGLA), a programmable ASIC with explicit direct memory access, local memories, and reusable compiler-visible integer lanes. The mapping adds OP_SMA4 as a reusable signed-int4 multiply-accumulate instruction rather than a BitNet-only datapath. Each ternary weight occupies one signed 4-bit lane. Each int8 activation is split into two signed-int4 fragments and reconstructed by shift-and-add. Frequency scaling of the 145 MHz FPGA measurement to an 840 MHz 28 nm CGLA achieved 0.390 ns per signed-int4 product. We showed that CGLA-offloaded BitNet C++ execution measures 2.52 tokens/s. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-24 in cs.AR). Categorie arXiv: Hardware Architecture (cs.AR). Pagina del lavoro: https://arxiv.org/abs/2609.27453 — 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 2026-10-08T09:14:39.464Z dalla pagina https://arxiv.org/catchup/cs.AR/2026-09-24?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).