Skip to content

run_vtr_task script does not take the max_router_iteration setting in Titan task #1458

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
tangxifan opened this issue Jul 31, 2020 · 6 comments
Assignees

Comments

@tangxifan
Copy link
Contributor

tangxifan commented Jul 31, 2020

The run_vtr_task.pl script does not take the max_router_iteration setting in Titan task config file but give a random number.

Expected Behaviour

In the task config.txt file, user can specify the parameters passed to vpr run, e.g.,
script_params=-starting_stage vpr --route_chan_width 300 --max_router_iterations 400 --router_lookahead map -timeout 172800.
The run_vtr_task will pass the parameters when calling vpr.
If we can the vpr.out log file, it should shows
/usr/bin/env time -v /research/ece/lnis/USERS/tang/research/EDA/vtr-verilog-to-routing/vpr/vpr stratixiv_arch.timing.xml stereo_vision_stratixiv_arch_timing --circuit_file stereo_vision_stratixiv_arch_timing.pre-vpr.blif --route_chan_width 300 --max_router_iterations 400 --router_lookahead map

Current Behaviour

However, when running the titan benchmark, the vpr.out log file shows
/usr/bin/env time -v /research/ece/lnis/USERS/tang/research/EDA/vtr-verilog-to-routing/vpr/vpr stratixiv_arch.timing.xml stereo_vision_stratixiv_arch_timing --circuit_file stereo_vision_stratixiv_arch_timing.pre-vpr.blif --route_chan_width 300 --max_router_iterations 150 --router_lookahead map

Possible Solution

Not sure if this problem occurs only to the titan benchmark configuration file.
Other configuration file may be impacted.

Steps to Reproduce

  • Clone VTR repository git clone https://github.com/verilog-to-routing/vtr-verilog-to-routing.git

  • Go to the repository and compile with make -j

  • Fetch titan benchmark

  • Copy the titan_task_list.txt to vtr_flow/tasks/regression_tests/vtr_reg_weekly/

  • Run titan regression tests using perl vtr_flow/scripts/run_vtr_task.pl -short_task_names -l ./vtr_flow/tasks/regression_tests/vtr_reg_weekly/titan_task_list.txt -j 23

  • Check the vpr.out file for any benchmark vi vtr_flow/tasks/regression_tests/vtr_reg_weekly/vtr_reg_titan/latest/stratixiv_arch.timing.xml/sparcT1_core_stratixiv_arch_timing.blif/common/vpr.out

Context

This is a functional problem of run_vtr_task.pl, which is detected by running the Titan benchmarks when applying RRGraph fix-up #1448

Your Environment

  • VTR revision used: latest master
  • Operating System and version: CentOS 7.5
  • Compiler version: g++-8.4
@vaughnbetz
Copy link
Contributor

Hi Jeff,

Xifan tells me this is holding up his work on re-tuning the Titan architecture; it's also going to hold up others in my group who are doing QoR tuning. Xifan is running run_vtr_task.pl, not vtr_flow.py, so I'm not sure what's changed, but perhaps there is some link?

@vaughnbetz
Copy link
Contributor

vaughnbetz commented Jul 31, 2020

@shadtorrie can you take a look? The recent changes in the related files are by you.

@shadtorrie
Copy link
Contributor

This issue is caused by my changes with run_vtr_flow, I found what was causing it have have now fixed it. Should I make a new pull request to fix this issue?

@vaughnbetz
Copy link
Contributor

Yes, please do. This would be good to get fixed right away.

@shadtorrie
Copy link
Contributor

@vaughnbetz I have submitted the pull request as seen above. @tangxifan could you review the fix as well?

@tangxifan
Copy link
Contributor Author

tangxifan commented Aug 1, 2020

@shadtorrie I just tried your branch by running the titan benchmarks. The problem is fixed. Thanks!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants