IBP is a new lossless bit-packing algorithm with GPU-optimized decompression that speeds up GNN training by 74%, DLRM lookups by 180%, and LLM inference by 24% by reducing CPU-GPU data movement.
Bifeat: Supercharge gnn training via graph feature quantization
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Reducing the GPU Memory Bottleneck with Lossless Compression for ML -- Extended
IBP is a new lossless bit-packing algorithm with GPU-optimized decompression that speeds up GNN training by 74%, DLRM lookups by 180%, and LLM inference by 24% by reducing CPU-GPU data movement.