[minjukim55]

Enable GPL_PLACE_IOS (#4685) in 20 public designs.

How it was measured:

• GPL_PLACE_IOS off vs on, same OpenROAD binary
• global placement seeds 1-10 (-random_seed), run through GRT
• enabled when setup WS improves or stays within the flag-off seed spread, and placement runtime drops

Means over seeds 1-10, off → on, all at GRT (5_1). fmax = ORFS timing__fmax. GPL wirelength = 3_3 final HPWL, GRT wirelength = 5_1 routed, placement runtime = 3_1..3_5.

design	fmax (GHz)	setup WS	setup TNS	GPL WL	GRT WL	placement runtime
asap7/aes	2.583 → 2.592 (+0.38%)	-7.2 → -5.7 ps	-37.5 → -29.3 ps	+4.4%	+2.5%	-6%
asap7/aes-block	2.193 → 2.214 (+0.96%)	-6.2 → -1.7 ps	-31.8 → -6.2 ps	+0.6%	+2.0%	-16%
asap7/ethmac	3.311 → 3.293 (-0.56%)	-88.0 → -77.7 ps	-1489 → -1311 ps	+0.1%	+0.3%	-5%
asap7/ethmac_lvt	3.567 → 3.559 (-0.21%)	-23.6 → -21.4 ps	-350.2 → -311.7 ps	+4.9%	+1.2%	-12%
asap7/gcd	3.015 → 2.997 (-0.62%)	-46.0 → -45.9 ps	-135.6 → -139.0 ps	+8.9%	+0.9%	-38%
asap7/gcd-ccs	2.976 → 2.957 (-0.63%)	-48.5 → -47.6 ps	-146.2 → -142.1 ps	+8.1%	+1.5%	-37%
asap7/jpeg	1.582 → 1.581 (-0.03%)	-87.2 → -87.4 ps	-22454 → -22574 ps	-1.0%	-1.0%	-5%
asap7/jpeg_lvt	1.696 → 1.692 (-0.25%)	+10.5 → +9.0 ps	0 → 0	+1.5%	+0.6%	-9%
asap7/mock-alu	1.706 → 1.727 (+1.25%)	-286.2 → -279.0 ps	-18362 → -18119 ps	-2.9%	-2.1%	-10%
asap7/picorv32	2.660 → 2.666 (+0.23%)	-164.5 → -164.3 ps	-27437 → -26684 ps	+2.0%	+1.2%	-7%
asap7/tinyRocket	1.439 → 1.438 (-0.08%)	-95.1 → -95.7 ps	-27393 → -22035 ps	+3.1%	+2.2%	-7%
asap7/uart	3.330 → 3.341 (+0.34%)	-30.3 → -29.3 ps	-1120 → -1119 ps	-1.2%	-1.1%	-31%
nangate45/dynamic_node	0.784 → 0.784 (0.00%)	-1.024 → -1.008 ns	-789.5 → -802.4 ns	-5.9%	-4.4%	-1%
nangate45/gcd	2.404 → 2.415 (+0.46%)	-0.166 → -0.164 ns	-7.32 → -7.14 ns	+6.0%	-5.8%	<1 s
nangate45/ibex	0.542 → 0.542 (+0.08%)	+0.002 → +0.002 ns	0 → 0	-5.9%	-4.0%	-8%
nangate45/jpeg	0.905 → 0.904 (-0.09%)	-0.105 → -0.106 ns	-37.2 → -38.1 ns	+2.9%	+2.0%	-7%
nangate45/tpu	0.705 → 0.710 (+0.62%)	-0.418 → -0.409 ns	-345.4 → -342.9 ns	-3.2%	-1.5%	-9%
sky130hs/aes	0.350 → 0.351 (+0.47%)	-0.060 → -0.046 ns	-0.21 → -0.17 ns	+2.9%	+0.5%	-2%
sky130hs/ibex	0.161 → 0.166 (+3.41%)	-0.510 → -0.387 ns	-265.0 → -102.2 ns	-1.1%	+1.8%	-6%
gf180/jpeg	0.166 → 0.166 (-0.16%)	-0.048 → -0.025 ns	-3.46 → -0.09 ns	-0.5%	-0.1%	-4%

Disabled (19, worse with the flag on):

• asap7: aes_lvt, aes-mbff, chi_async_bridge, ibex, riscv32i, tpu
• nangate45: picorv32
• sky130hd: aes, gcd, ibex, riscv32i
• sky130hs: jpeg, riscv32i
• gf180: aes
• ihp-sg13g2: aes
• dropped after CI: asap7/riscv32i-mock-sram (finish TNS), sky130hd/jpeg and gf180/aes-hybrid (GRT antenna diodes up on all 10 seeds), sky130hs/gcd (TNS worse on 8/10 seeds)

Not measured yet (25):

• asap7: coralnpu, cva6, swerv_wrapper
• nangate45: ariane133, ariane136, black_parrot, bp_be_top, bp_fe_top, bp_multi_top, bp_quad, cva6, mempool_group, swerv, swerv_wrapper, tinyRocket
• sky130hd: chameleon, microwatt
• gf180: ibex, riscv32i, uart-blocks
• ihp-sg13g2: gcd, ibex, jpeg, riscv32i, spi

Not applicable (2, pins fixed by FLOORPLAN_DEF/FOOTPRINT): nangate45/aes, ihp-sg13g2/i2c-gpio-expander

[gemini-code-assist]

Code Review

This pull request enables the GPL_PLACE_IOS option by setting export GPL_PLACE_IOS = 1 across various design configuration files (config.mk) for multiple platforms, including asap7, gf180, nangate45, sky130hd, and sky130hs. I have no feedback to provide.

[openroad-ci]

🔍 QoR check

Metrics reflect the PR merge build — i.e. what will land on the target branch.

Commit 7b71b93 · Jenkins build #2 · Baseline: build · View build on dashboard

62 design(s) checked — 2 with regression(s), 0 without a comparable baseline.

❌ asap7/ethmac_lvt base — 1 failing metric(s)

Metric	target	base	delta	limit	band
globalroute__timing__setup__tns	-307.720	-234.642	-31.144%	-294.642	PeriodPadding 20.000%

❌ asap7/uart base — 1 failing metric(s)

Metric	target	base	delta	limit	band
cts__timing__setup__tns	-1163.650	-958.162	-21.446%	-1149.794	PeriodPadding 20.000%

[oharboe]

@minjukim55 Could you post minimum clock period impact? #4583

[minjukim55]

@minjukim55 Could you post minimum clock period impact? #4583

Added fmax