{"id":"arxiv/2604.13433","url":"https://arxiv.org/abs/2604.13433","tags":["cs.DC","math.NA"],"score":0,"title":"PackSELL: A Sparse Matrix Format for Precision-Agnostic High-Performance SpMV","domain":"cs.DC","source":"arxiv","authors":["Kengo Suzuki","Takeshi Iwashita"],"reviews":0,"subject":"cs.DC","summary":"We propose a new sparse matrix format, PackSELL, designed to support diverse data representations and enable efficient sparse matrix-vector multiplication (SpMV) on GPUs. Building on sliced ELLPACK (SELL), PackSELL incorporates delta encoding of column indices and a novel packing scheme that stores each index-delta-value pair in a single word, thereby reducing memory footprint and data movement. This design further enables fine-grained control over the bit allocation between deltas and values, allowing flexible data representations, including non-IEEE formats. Experimental results show that,…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-04-15","reactions":{}}