Based on UG904 (v2025.2). For command syntax and property tables see REFERENCE.md; for report_qor_suggestions RTL optimization examples (UG906 before/after) see examples/ug906/ directory.

Implementation Flow Overview

opt_design          ← Logic optimization (REQUIRED)
  ↓
power_opt_design    ← Clock gating power opt (OPTIONAL, not Versal)
  ↓
place_design        ← Placement (REQUIRED)
  ↓
power_opt_design    ← Post-place power opt (OPTIONAL)
  ↓
phys_opt_design     ← Post-place physical opt (OPTIONAL, recommended)
  ↓
route_design        ← Routing (REQUIRED)
  ↓
phys_opt_design     ← Post-route physical opt (OPTIONAL)
  ↓
write_bitstream     ← Bitstream (all except Versal)
write_device_image  ← Device image (Versal only)

Key rule: All implementation commands are re-entrant — they can be run repeatedly on the same design. Each run optimizes the results of the previous run.

opt_design — Logic Optimization

Directive Decision Table

Scenario	Directive	Effect
Default / first try	`Default`

Phase	Option	Default	Description
1	`-retarget`	ON	Retarget primitives across device families
2	`-propconst`	ON	Constant propagation
3	`-sweep`	ON	Remove loadless cells, tie-off, retarget dual-port→single-port RAM
4	`-muxf_remap`	off	Remap MUXF7/F8/F9 to LUT3 for routability
5	`-carry_remap`	off	Remap short CARRY chains to LUTs
6	`-control_set_merge`	off	Merge equivalent control set drivers
7	`-merge_equivalent_drivers`	off	Merge all equivalent drivers (not just control)
8	`-bufg_opt`	ON	Insert BUFG on high-fanout clock/non-clock nets
9	`-shift_register_opt`	ON	SRL fanout opt + SRL↔register transforms
10	`-mbufg_opt`	off	Replace parallel BUFGCEs with MBUFG (Versal)
11	`-dsp_register_opt`	off	Optimize DSP pipeline registers
12	(auto)	ON	Control Set Reduction (CONTROL_SET_REMAP property)
13	`-hier_fanout_limit <N>`	off	Module-based fanout replication
13	`-control_set_opt`	off	Control Set Optimization (auto-selected candidates)
14	`-remap`	off	Combine cascaded LUTs to reduce logic levels
15	`-resynth_remap`	off	Timing-driven re-synthesis + remap
16	`-resynth_area`	off	Re-synthesis for area (reduce LUTs)
17

Scenario	Directive	Designs Benefited
Default	`Default`	All
Deep exploration	`Explore`	All (higher effort detail placement)
Aggressive exploration	`AggressiveExplore`	Timing-critical designs
RAM/DSP dense	`WLDrivenBlockPlacement`	Many BRAM/DSP blocks
RAM/DSP dense	`EarlyBlockPlacement`	RAM/DSP as placement anchors
Timing meets post-place but fails post-route	`ExtraNetDelay_high`	Long-distance nets, high fanout
Timing meets post-place but fails post-route	`ExtraNetDelay_low`	Same, lower pessimism
Congestion	`AltSpreadLogic_high`	High connectivity → congestion
Congestion (moderate)	`AltSpreadLogic_medium/low`	Moderate congestion
SSI congestion	`SSI_SpreadLogic_high/low`	SSI devices
SSI SLR balancing	`SSI_SpreadSLLs`	Balance SLL connections across SLRs
SSI SLR balancing	`SSI_BalanceSLLs`	Balance SLLs between SLRs
SSI SLR balancing	`SSI_BalanceSLRs`	Balance cell count between SLRs
SSI high utilization	`SSI_HighUtilSLRs`	Pack logic closer in each SLR
Extra post-place opt	`ExtraPostPlacementOpt`	All
Alternate timing	`ExtraTimingOpt`	Alternative timing-driven algorithms
ML-predicted best	`Auto_1`	Highest confidence ML prediction
ML-predicted 2nd	`Auto_2`	Second best ML prediction
ML-predicted 3rd	`Auto_3`	Third best ML prediction
Fast iteration

Option	Effect
`-post_place_opt`	Extra timing optimization after placement
`-timing_summary`	Force STA-based timing summary (more accurate, slower)
`-no_timing_driven`	Wirelength-only placement (fastest)
`-no_psip`	Disable Physical Synthesis in Placer
`-no_bufg_opt`	Disable BUFG insertion during placement
`-sll_align_opt`	Align SLL registers for SSI multi-die parts
`-ultrathreads`	Parallel placement across SLRs (UltraScale+ SSI)
`-unplace`	Remove all non-fixed placements

Scenario	Directive	Effect
Default	`Default`	Default optimizations
Deep exploration	`Explore`	Multi-pass + SLR crossing + critical path final phase
Explore + hold fix	`ExploreWithHoldFix`	Explore + hold violation fixing
Explore + aggressive hold	`ExploreWithAggressiveHoldFix`	Explore + aggressive hold fixing
Most aggressive	`AggressiveExplore`	Allows WNS degradation in SLR crossing opt
Alternative replication	`AlternateReplication`	Different critical cell replication algorithm
Fanout-focused	`AggressiveFanoutOpt`	Aggressive fanout optimization
Add retiming	`AddRetime`	Default flow + register retiming
Aggressive + retiming	`AlternateFlowWithRetiming`	Aggressive replication + DSP/BRAM opt + retiming
Fast	`RuntimeOptimized`	Fewest iterations

Option	Description
`-fanout_opt`	Replicate high-fanout net drivers (default post-place)
`-critical_cell_opt`	Replicate cells in failing paths (default post-place)
`-placement_opt`	Re-place critical path cells (default both modes)
`-dsp_register_opt`	Move registers in/out of DSP cells
`-bram_register_opt`	Move registers in/out of BRAM cells
`-uram_register_opt`	Move registers in/out of UltraRAM cells
`-shift_register_opt`	Extract SRL end stages to improve timing
`-restruct_opt`	Swap LUT connections to reduce logic levels
`-lut_opt`	Single LUT movement/replication
`-clock_opt`	Useful clock skew optimization

Option	Description
`-hold_fix`	Fix hold violations above threshold
`-aggressive_hold_fix`	Fix more hold violations
`-sll_reg_hold_fix`	SLL register hold fix (UltraScale+)
`-insert_negative_edge_ffs`	Insert neg-edge FFs to split hold paths

Option	Description
`-retime`	Register retiming (Versal)
`-interconnect_retime`	Interconnect retiming by FF movement (Versal)
`-force_replication_on_nets <nets>`	Force driver replication on specific nets
`-equ_drivers_opt`	Rewire loads to equivalent drivers
`-casc_opt`	LUT cascade optimization (Versal)
`-cell_group_opt`	Critical fanin cone group opt (Versal)
`-bram_enable_opt`	Reverse BRAM power opt on timing-critical enable paths
`-path_groups <args>`	Limit optimization to specific path groups
`-tns_cleanup`	Allow slack degradation if WNS maintained (with -slr_crossing_opt)

Scenario	Directive	Effect
Default	`Default`	Default routing
Explore alternatives	`Explore`	Explore different critical path routes (signoff timing)
Aggressive exploration	`AggressiveExplore`	More aggressive thresholds
No timing relaxation	`NoTimingRelaxation`	Never relax timing goals
More iterations	`MoreGlobalIterations`	Detailed timing analysis all stages
Emphasize delay	`HigherDelayCost`	Trade compile time for delay optimization
Fast	`RuntimeOptimized`	Fewest iterations
Congestion	`AlternateCLBRouting`	Alternate CLB routing algorithms
Fastest	`Quick`	Non-timing-driven, minimum legal routing
QoR-suggested	`RQS`	Uses report_qor_suggestion

Option	Effect
`-tns_cleanup`	Focus on WNS, fix non-critical failing paths. Use before post-route phys_opt
`-preserve`	Preserve existing routes, route remaining. For pre-routing critical nets
`-nets <net_objects>`	Route only specified nets
`-pins <pin_objects>`	Route only specified pins
`-delay`	Route individual nets with smallest delay
`-auto_delay`	Route with timing-constraint-driven budgets (use with -nets/-pins)
`-max_delay <ps>` / `-min_delay <ps>`	Target delay for pin routing (use with -pins)
`-unroute`	Remove routing (entire design or specific nets/pins)
`-timing_summary`	Force STA timing summary (more accurate)
`-finalize`	Complete partially routed connections (ECO flow)
`-eco`	Incremental ECO routing (faster after small changes)
`-ultrathreads`	Parallel routing (faster, slight variation between runs)
`-no_timing_driven`	Disable timing-driven routing (feasibility test only)

# Route top 10 critical nets first with minimum delay
set preRoutes [get_nets -of [get_timing_paths -max_paths 10]]
route_design -nets [get_nets $preRoutes] -delay
# Then route rest preserving critical routes
route_design -preserve

# Basic usage (optimize entire design)
power_opt_design

# Control scope
set_power_opt -include_cells [get_cells inst_A]
set_power_opt -exclude_cells [get_cells inst_B]
set_power_opt -cell_types {BRAM}
set_power_opt -clocks {clk1}

# Non-Project Mode
read_checkpoint -incremental <reference_routed.dcp>
# Automatic incremental (recommended)
read_checkpoint -incremental -auto_incremental <reference.dcp>
# Force incremental even if criteria not met
read_checkpoint -incremental -force_incr <reference.dcp>
# Fix specific objects
read_checkpoint -incremental <ref.dcp> -fix_objects [all_rams]

read_checkpoint -incremental -directive <directive> <ref.dcp>

report_incremental_reuse -file incr_reuse.rpt

report_design_analysis -congestion -file congestion.rpt
report_route_status -file route_status.rpt

Vivado Implementation Decision Guide | Skills Pool

Option	Description
`-routing_opt`	Re-route critical nets/pins (default post-route)
`-slr_crossing_opt`	Optimize inter-SLR paths (default both modes)
`-critical_pin_opt`	Remap LUT pins to faster physical pins

Vivado Implementation Decision Guide

Vivado Implementation Decision Guide

Implementation Flow Overview

opt_design — Logic Optimization

Directive Decision Table

Available Optimizations (18 phases)

place_design — Placement

Directive Decision Table

Key Options

phys_opt_design — Physical Optimization

Two Modes

Directive Decision Table

Key Individual Options

route_design — Routing

Directive Decision Table

Key Options

Pre-routing Critical Nets Pattern

power_opt_design — Power Optimization

Incremental Implementation

Setup

Incremental Directives

Auto Incremental Criteria

Analysis

Congestion Analysis & Resolution

Congestion Levels

Diagnosis

Resolution Decision Tree

Predefined Implementation Strategies (Project Mode)

Performance-focused

Congestion-focused

Area-focused

Taskflow Inbox Triage

Accessibility

Open a Pull Request

Investor Materials

Continuous Agent Loop

Configure Ecc

Directive	Effect
`RuntimeOptimized`	Reuse max, target same WNS as reference (default)
`TimingClosure`	Rip up failing paths, try harder to close timing
`Quick`	No timer, fastest, needs WNS > 1.0 ns