PCPP replaces all-to-all stale activation exchange in patch-parallel diffusion inference with point-to-point neighbor exchange, cutting communication by about 70% and improving speed-up to 2.36-8.02x, but degrading LPIPS and FID scores.
Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Partially Conditioned Patch Parallelism for Accelerated Diffusion Model Inference
PCPP replaces all-to-all stale activation exchange in patch-parallel diffusion inference with point-to-point neighbor exchange, cutting communication by about 70% and improving speed-up to 2.36-8.02x, but degrading LPIPS and FID scores.