Accelerating MTTKRP for sparse tensor decomposition on GPUs

Key Points

• Sparse tensor decomposition can significantly benefit from MTTKRP optimization, enhancing efficiency.
• Performance metrics show a notable speedup using GPUs, improving processing times by up to 50% across benchmarks.
• Analysis employs GPU parallel processing techniques to optimize MTTKRP for large-scale tensor data.
• Findings highlight the potential for large-scale data applications, driving forward data processing capabilities.