PackSELL packs delta-encoded indices and values into single words with tunable bit allocation, delivering up to 1.63x faster FP16 SpMV and FP32-accurate performance exceeding FP16 cuSPARSE while reducing memory traffic.
In2025 IEEE 32nd Symp
3 Pith papers cite this work. Polarity classification is still indexing.
years
2026 3verdicts
UNVERDICTED 3representative citing papers
Refinements to error-free transformations plus residue override reduce false reports in floating-point residue computation on most tested benchmarks.
Establishes sufficient more-general conditions for FastTwoSum as an error-free transformation under faithful rounding modes and introduces a configurable ExtractScalar splitting for round-to-odd.
citing papers explorer
-
PackSELL: A Sparse Matrix Format for Precision-Agnostic High-Performance SpMV
PackSELL packs delta-encoded indices and values into single words with tunable bit allocation, delivering up to 1.63x faster FP16 SpMV and FP32-accurate performance exceeding FP16 cuSPARSE while reducing memory traffic.
-
Accurate Residues for Floating-Point Debugging
Refinements to error-free transformations plus residue override reduce false reports in floating-point residue computation on most tested benchmarks.
-
Odd but Error-Free FastTwoSum: More General Conditions for FastTwoSum as an Error-Free Transformation for Faithful Rounding Modes
Establishes sufficient more-general conditions for FastTwoSum as an error-free transformation under faithful rounding modes and introduces a configurable ExtractScalar splitting for round-to-odd.