Closed bcaddy closed 10 months ago
I looked into this again as part of my MHD debugging. I don't know what I did wrong the first time but initializing the memory did fix the initcheck errors in my MHD branch. Presumably it will with the other builds/branches and I'll include initialization as part of my next MHD PR.
The non-initcheck issues remain.
I ran the Compute Sanitizer on all the various builds and discovered some issues. Most of them appear to be uninitialized memory accesses in the HLLC solver but initializing all the arrays in the integrator with
memset
or a kernel didn't resolve the errors. Below is a table of which builds failed which checks and what initial conditions were used. Note that I pretty much just used Sod for all of them so a more appropriate set of initial conditions could show more errors.See PR #196 for the script to run the compute sanitizer.