search

The-OpenROAD-Project / OpenROAD

OpenROAD's unified application implementing an RTL-to-GDS Flow. Documentation at https://openroad.readthedocs.io/en/latest/
https://theopenroadproject.org/
BSD 3-Clause "New" or "Revised" License
1.62k stars 561 forks source link

macro placement fails #6055

Closed oharboe closed 2 weeks ago

oharboe commented 3 weeks ago

Describe the bug

with SYNTH_HIERARCHICAL=0, number of instances is half of that of SYNTH_HIERARCHICAL=1, so it was worth trying to turn it off, but macro placement fails. https://github.com/The-OpenROAD-Project/megaboom/pull/188

untar https://drive.google.com/file/d/1wvt-ThCrWKvSeo4jKmxmgpnbLtjV_XRE/view?usp=sharing

$ ./run-me-BoomTile-asap7-1.sh 
OpenROAD v2.0-16664-g9ba039016 
Features included (+) or not (-): +Charts +GPU +GUI +Python
This program is licensed under the BSD-3 license. See the LICENSE file for details.
Components of this program may be licensed under more restrictive licenses which must be honored.
rtl_macro_placer -halo_width 19 -halo_height 19 -report_directory .//objects/asap7/BoomTile/1/rtlmp -target_util 0.24
Floorplan Outline: (0.0, 0.0) (2000.0, 2000.0),  Core Outline: (2.052, 2.16) (1998.0, 1998.0)
    Number of std cell instances: 1046614
    Area of std cell instances: 127052.54
    Number of macros: 72
    Area of macros: 810855.88
    Halo width: 19.00
    Halo height: 19.00
    Area of macros with halos: 1394272.12
    Area of std cell instances + Area of macros: 937908.44
    Core area: 3983592.75
    Design Utilization: 0.24
    Core Utilization: 0.04
    Manufacturing Grid: 1

[ERROR MPL-0040] Failed on cluster (root)_glue_logic_1_0_0
Error: macro_place_util.tcl, 76 MPL-0040
Command exited with non-zero status 1
Elapsed time: 39:50.24[h:]min:sec. CPU time: user 6409.13 sys 687.58 (296%). Peak memory: 16733712KB.

Expected Behavior

Successful macro placement

Environment

OpenROAD v2.0-16664-g9ba039016

To Reproduce

See above.

Relevant log output

No response

Screenshots

No response

Additional Context

No response

oharboe commented 3 weeks ago

Note that if I change the PLACE_DENSITY slightly, then macro placement passes. I tried with 0.22, 0.30, 0.20, 0.25, 0.22: macro placement works in all cases.

maliberty commented 3 weeks ago

Such is the nature of annealing.

AcKoucher commented 3 weeks ago

@oharboe FYI I'm investigating what is going on for #5996 and I think that it might be same problem across multiple issues, including this one.