Compile Options - 2025.1 English - UG1702

The syn.compile commands specify the default behavior for compilation of the HLS component.

syn.compile.design_size_max_warning

Specifies the design size that triggers a warning related to slow compilation or poor QoR.

syn.compile.design_size_maximum_warning=300000

syn.compile.enable_auto_rewind

When true enables an alternative HLS implementation for pipelined loops which uses automatic loop rewind as described in the Rewinding Pipelined Loops for Performance section of the Vitis High-Level Synthesis User Guide (UG1399) for additional information.

syn.compile.enable_auto_rewind=1

syn.compile.ignore_long_run_time

Disable long runtime warning.

syn.compile.ignore_long_run_time=1

syn.compile.name_max_length <arg>

The name_max_length option will specify the maximum length of function names. If the length is over the threshold, the last part of the name will be truncated.

syn.compile.name_max_length=13

syn.compile.no_signed_zeros

The no_signed_zeros option will ignore the signedness of floating point zero so that compiler can do aggressive optimizations on floating point operations.

syn.compile.no_signed_zeros=1

syn.compile.performance_budgeter

By enabling this option, the top-level performance pragma conducts a comprehensive, design-wide performance analysis. Much like selecting the most efficient route from point A to point B among many possibilities, it considers multiple optimization strategies across the entire design to identify the best path to meet performance targets.

enable: Activates the performance budget for a comprehensive, design-wide performance analysis.

disable: Deactivates the performance budget, preventing any comprehensive analysis.

auto: Automatically enables design-wide performance analysis if a performance pragma is detected; otherwise, the performance budget remains disabled.

syn.compile.pipeline_flush_in_task arg

Specifies that pipelines in hls::tasks will be flushing (flp) by default to reduce the probability of deadlocks in C/RTL Co-simulation. This option applies to pipelines that achieve an II=1 with the default option of ii1. However, you can also specify it as applying always to enable flushing pipelines in either hls::tasks or dataflow, or can be completely disabled using never.

Important: Flushing pipelines (flp) are compatible with the rewind option specified in the PIPELINE pragma or directive when the syn.compile.enable_auto_rewind option is also specified.

always

Always make pipelines flushable in hls::tasks or dataflow regardless of II.

never

Never make pipeline flushable unless specifically overridden by other directives or pragmas.

ii1

Make pipelines that achieve II=1 flushable in hls::tasks. This is the default setting.

syn.compile.pipeline_flush_in_task=always

syn.compile.pipeline_loops arg

Loops with a tripcount equal to or greater than the specified value will be pipelined automatically.

syn.compile.pipeline_loops=20

syn.compile.pipeline_style arg

Set default pipeline style, this is a preference not a hard constraint. Refer to Flushing Pipelines and Pipeline Types in Vitis High-Level Synthesis User Guide (UG1399) for more information. The three styles are: stallable (stp), flushable (flp), and free-running (frp). The default is free-running.

syn.compile.pipeline_style=flp

syn.compile.pragma_strict_mode

Enable errors instead of warnings for unrecognized and improper pragma syntax.

syn.compile.pragma_strict_mode=1

syn.compile.unsafe_math_optimizations

The unsafe_math_optimizations option will ignore the signedness of floating point zero and enable associative floating point operations so that compiler can do aggressive optimizations on floating point operations.

syn.compile.unsafe_math_optimizations=1

Important: Using this option might change the result of any floating point calculations and result in a mismatch in C/RTL co-simulation. Ensure your test bench is tolerant of differences and checks for a margin of difference, not exact values.