All Tools
D
Dev ToolsFreeOpen Source
DREAMPLACE
GPU-accelerated VLSI placement on deep-learning toolkits
BSD-3-Clause
ABOUT
Global placement of million-cell ASICs is slow on CPU placers and locked to licensed EDA. DREAMPlace casts placement as a deep-learning training problem so teams run global, legalize, and detailed place on GPU at tens of times CPU speed.
INTEGRATION GUIDE
1. Run GPU global placement on ISPD or industrial netlists
2. Legalize and detailed-place million-cell designs with ABCDPlace
3. Compare open placement quality against RePlAce or NTUPlace
4. Prototype placement research without a closed physical-design license
TAGS
dreamplacevlsiplacementgpuasicedacuda