amrex::Parser #2063
amrex::Parser #2063
Conversation
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
c05226a to
0d8eecd
Compare
Replace WarpXParser with amrex::Parser. Roundoff errors are expected because of additional optimization in amrex::Parser.
…ge in single precision parser
7ace5f9 to
db8e4ec
Compare
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
There was a problem hiding this comment.
Thank you for generalizing this! Thanks also for updating the docs.
Awesome that this also resolved the recursive function usage. Also looking forward to the temporary variable support with r=sqrt(x*x+z*x); ... expressions 🎉 Do you want to add this in one of the tests?
I quickly checked if #1874 still works and it looks like it's covered in CI.
Added minor inline comments, looks already great to me.
Co-authored-by: Axel Huebl <axel.huebl@plasma.ninja>
|
Maybe we can add the temporary variable to tests later. |
|
As @dpgrote requested, integer is supported in |
|
Not sure why the DPC++ build fails now. 🤔 |
|
Looks like the issue is |
This comment has been minimized.
This comment has been minimized.
|
The issue is |
|
The MCCProcess issue is addressed in #2085. |
|
This pull request fixes 1 alert when merging e3ca37a into 40e36e1 - view on LGTM.com fixed alerts:
|
| @@ -127,7 +127,7 @@ public: | |||
| } | |||
|
|
|||
| // get parser | |||
| HostDeviceParser<m_nvars> reduction_function_parser = getParser(m_parser); | |||
| auto reduction_function_parser = m_parser->compile<m_nvars>(); | |||
There was a problem hiding this comment.
I am wondering if the many specializations in various TUs impact compile-time.
Will take a look when CI ran through just to get an idea.
There was a problem hiding this comment.
Haha, the error bar in CI is about 5min wide, thus it's hard to say. Looks still the same, given that fluctuation :D
Replace WarpXParser with amrex::Parser. Roundoff errors are expected because
of additional optimization in amrex::Parser.