A 28nm hybrid systolic array accelerator with MXINT4 quantization and fused RMSNorm/RoPE units reports 247/117 token/s/mm2 running RetNet 1.3B, claiming over 2.45x/13.5x area-efficiency gains over prior edge LLM accelerators.
Vs-quant: Per-vector scaled quantization for accurate low-precision neural network inference,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
baseline 1
citation-polarity summary
fields
cs.AR 1years
2025 1verdicts
CONDITIONAL 1roles
baseline 1polarities
contest 1representative citing papers
citing papers explorer
-
Hybrid Systolic Array Accelerator with Optimized Dataflow for Edge Large Language Model Inference
A 28nm hybrid systolic array accelerator with MXINT4 quantization and fused RMSNorm/RoPE units reports 247/117 token/s/mm2 running RetNet 1.3B, claiming over 2.45x/13.5x area-efficiency gains over prior edge LLM accelerators.