Tags: FermiQ/OpenLane
Tags
Support Multi-Arch Docker Builds (The-OpenROAD-Project#1075) Tested on Apple M1 and AWS Graviton. + Add support for multiple-architecture building + Support for amd64, ppc64le and arm64v8 added + amd64 and arm64v8 will be automatically built by the CI ~ Major CI Overhauls ~ Added composite actions for the Docker build and setting environment variables ~ pdk build is now its own job, and amd64/aarch64 have a job each ~ matrix -> design_matrix, designs only tested on amd64 for performance reasons ~ run/build base docker images now pushed to Docker Hub based on a hash of dependency lists (to work with the buildx `docker-container` driver) - Remove pandas (overkill for what we're using it for) - Remove `cugr`, `drcu` *for now* (they're x86-64 Linux-only utilities)
Enable PL_ROUTABILITY_DRIVEN and PL_TIME_DRIVEN (The-OpenROAD-Project… …#1092) ORFS enables both of these by default.
Update OR to get pdngen fix (The-OpenROAD-Project#1094)
Make `RUN_SPEF_EXTRACTION` skip STA entirely, remove pandas (The-Open… …ROAD-Project#1089) ~ `RUN_MAGIC` check moved to flow.tcl ~ `RUN_SPEF_EXTRACTION` now skips STA entirely - Remove pandas (it's overkill for what we need + greatly inflates install times)
Update to use the new C++ pdngen in OR (The-OpenROAD-Project#1059) * Update to use the new C++ pdngen in OR designs/inverter: updated die area to be a multiple of site size designs/s44: has no need for a custom cfg so just removed it Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * For spm turn off FP_PDN_AUTO_ADJUST The /8 hack produces an off-grid result. I just manually set the values to something reasonable. Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * Update to the latest OR version Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * Restore default setting of VDD_NET/GND_NET Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * For inverter turn off FP_PDN_AUTO_ADJUST The /8 hack produces an off-grid result. I just manually set the values to something reasonable. Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * Update OR version to get pdn via enclosure fix. Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * Add -extend_to_core_ring to add_pdn_stripe for rings outside the die area This happens on caravel_upw. Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * Update OR for cut enclosure parsing in odb Commit cf8d8dc8091612f7948cfc9f19d6cfdf9913e72d Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * Modifed caravel_upw to have to instances connected to different supplies This is a better test of multi-power pdngen Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * Update pdn scripts for new pdngen in OR Remove the looping over different supplies and run pdngen once with a full configuration using secondary supplies. Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * check FP_PDN_MACRO_HOOKS is set Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * update OR for via env above/below fix Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu> * Update OR to get dpl fix for flipped cells Issue reported in The-OpenROAD-Project/OpenROAD#1840 which was triggered by the OR version update for pdngen. Signed-off-by: Matt Liberty <mliberty@eng.ucsd.edu>
Remove Hard-coded Paths in Sample Designs (The-OpenROAD-Project#1083)
Global Placement Fixes (The-OpenROAD-Project#1039) ~ Move set_routing_layers into own file ~ Don't override global placement routability options ~ Global placement timing and routability driven issues
Distributed runs with OR changelog (The-OpenROAD-Project#1074) ~ Run CI on distributed mode ~ Try to keep track of OR changes between runs
Add Caravel User Project Wrapper to Fastest Test Set (The-OpenROAD-Pr… …oject#1072) + Add caravel_upw as a design, included in the fastest_test_set ~ Alter the clean options in the makefile ~ Improve LVS mismatch message
Small QOL changes (The-OpenROAD-Project#1068) - Change 'Running Resizer ..' to 'Running Placement Resizer..' and other equivalent changes - add the piping part to run_openroad_script function allowing for logging the log file of the openroad script in openlane's run cmds.log - report the relative log file path on openroad script crash - adjust "skipping" messages
PreviousNext