Custom Query (126 matches)
Results (79 - 81 of 126)
Ticket | Resolution | Summary | Owner | Reporter |
---|---|---|---|---|
#90 | fixed | MPI dead lock in XIOS | ymipsl | mcastril |
Description |
We are experiencing a repetitive issue with XIOS 1.0 . It appeared using NEMO 3.6 stable and more than 2600 cores, and it seemed to be solved when using Intel 16 compiler and IMPI 5. However, after updating to NEMO 3.6 current stable, the problem appears when using 1920 or more cores. I don't really get how the NEMO revision change could affect to this, but there it is. The problem is just in this line of client.cpp: MPI_Send(buff,buffer.count(),MPI_CHAR,serverLeader,1,CXios::globalComm) ; In the meanwhile the server.cpp is doing MPI_Iprobe continuosly in order to receive all the MPI_Send. What we have observed is that using a high number of cores, around 80-100 of these cores get stucked at the MPI_Send, causing the run to hang and not complete. The fact that with a certain number of cores the issue appears 80% of the times but not always, made us think that could be related with the IMPI implementation. |
|||
#92 | worksforme | Using named constants in filters/operations | ymipsl | ssenesi |
Description |
Assume you want to convert an energy flux in mass flux using an XIOS filter :
<field id="tran" name="tran" unit="kg m-2 s-1"> evptr / 2.5008E+6</field> It would be smarter to use (and re-use) a named constant, such as in :
<variable_definition> <variable id="coeffl" type="float">2.5008E+6</variable> </variable_definition> <field_definition id="field_definition" operation="instant" prec="4"> <field id="tran" name="tran" unit="kg m-2 s-1"> evptr / coeffl </field> </field_definition> However, this does not work |
|||
#95 | fixed | Alternate, sensible, convention for lonvalue_1d when data_dim=1, for a rectilinear domain | ymipsl | ssenesi |
Description |
In domain.cpp, for the case of a rectilinear domain, completeLonLatClient uses lonvalue_1D with an implicit assumption that the distribution also is recti-linear (i.e. that each MPI tasks does process a rectangle in the lat-lon space) This does not apply when data_dim=1 and the distribution follows another scheme. The attached version of domain.cpp, domain_new.cpp, is a proposal for an alternate convention in that case : lonvalue_1d and latvalue_1d are then interpreted as providing arrays which are parallel to the data arrays (which is more intuitive in that data_dim=1 case) |