Memory Issue with Rocstar v1.0

[kevinmuellerUTAS]

We are having a memory issue running Rocstar v1.0. While running the ACM test case with either Rocflu or Rocflo, Rocstar's memory usage continues to increase until the memory is filled and the solver crashes. This memory leak has been verified on two different systems: one, a high end Dell CFD Workstation running Redhat Enterprise v7.2, and two, a laptop running Ubuntu 14.04 LTS.

The two builds of Rocstar were used to solve the same Rocflu ACM test case with 4 processors. The used memory for each simulation was charted as a function of time and the rate of memory increase is identical between the two systems. See the attached PDF containing the plots.

Comparison of Rocstar Runs.pdf

Included in the plots are information about the Third Party Libraries and compilers used for preparing

This problem appears to be limited to when the Rocburn module is called - all of the other test cases, Super Seismic Shock, Elastic Piston, and the Shock Panel do not exhibit the memory leak. When the same Rocflu ACM test case was prepared WITHOUT including the '-b' in Rocprep, it also did not suffer from the memory leak.

Please let me know if there is any other information I need to supply in order to help resolve this issue.

Thank you.

[MTCam]

Thank you for the report, Kevin. We really appreciate the detail. It
shouldn't be too terribly hard to track down; Rocburn is pretty small.
We'll look into it.

Cheers,
-Mike

On Thu, Mar 17, 2016 at 1:26 PM, kevinmuellerUTAS [email protected]
wrote:

We are having a memory issue running Rocstar v1.0. While running the ACM
test case with either Rocflu or Rocflo, Rocstar's memory usage continues to
increase until the memory is filled and the solver crashes. This memory
leak has been verified on two different systems: one, a high end Dell CFD
Workstation running Redhat Enterprise v7.2, and two, a laptop running
Ubuntu 14.04 LTS.

The two builds of Rocstar were used to solve the same Rocflu ACM test case
with 4 processors. The used memory for each simulation was charted as a
function of time and the rate of memory increase is identical between the
two systems. See the attached PDF containing the plots.

Comparison of Rocstar Runs.pdf
https://github.com/IllinoisRocstar/Rocstar/files/178441/Comparison.of.Rocstar.Runs.pdf

Included in the plots are information about the Third Party Libraries and
compilers used for preparing

This problem appears to be limited to when the Rocburn module is called -
all of the other test cases, Super Seismic Shock, Elastic Piston, and the
Shock Panel do not exhibit the memory leak. When the same Rocflu ACM test
case was prepared WITHOUT including the '-b' in Rocprep, it also did not
suffer from the memory leak.

Please let me know if there is any other information I need to supply in
order to help resolve this issue.

Thank you.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub
IllinoisRocstar/Rocstar#2

Michael T. Campbell

Computational Physicist

IllinoisRocstar LLC

1800 South Oak Street, Suite 208
Champaign, IL 61820
www.illinoisrocstar.com

(217)390-3287 [email protected] (cell)

[kevinmuellerUTAS]

Great Mike - thank you! Again, please let me know if there's any other information I can provide.
--Kevin

[Hanmeimei521]

Hello, I am using Rocstar to simulate the two-phase flow. I have encountered the problem of selecting 0 (ASCII) or 1 (binary) output in the # FORMATS Section in the.inp file, which gives a "ERROR 1005: Window does not exist". Do you have encountered this problem? ?