Processing cadence result data from psf folder

This is a question about namespaces. In OCEAN data access functions (e.g. v(), i(), getData(), pv() etc), the net name/terminal name can be specified either as a "schematic" name or a "netlist" name. If the name begins with a "/", then it refers to a "schematic" name - i.e. the hierarchical name of something within the Virtuoso database. Otherwise it refers to a name that is in the netlist - in which case it has to refer to spectre's naming rules (i.e. dot "." as the hierarchy separator, and ":" for terminal access or operating point parameter access).

When the schematic is netlisted, you end up with directories called map and amap in the netlist directory which contain the information about how schematic names are mapped to netlist names (in particular this is important if reserved words or illegal characters are used). The names you ask to save get translated via this mapping informaton into corresponding netlist names - so if you ask to save /V2_OUTP_Q/PLUS in OCEAN, it will correspond to a "save V2_OUTP_Q:p" in the netlist. Similarly post-simulation, it uses this mapping information to allow you to access signals using the schematic names (the result database itself only ever contains netlist names).

In order to do that mapping however, OCEAN needs to know where the amap directory is. The simplest way of achieving that is to ensure that you have:

someDirectory
  --> netlist
          --> amap dir
          --> map dir
          --> input.scs
  --> psf

You run the simulation from inside the netlist dir, and then use "spectre -raw ../psf input.scs"

The expectation from the OCEAN access functions is that you have netlist and psf alongside each other. There are ways around that if OCEAN runs the simulation because it can create a runObjFile in the psf dir to tell it where the netlist dir is, but if running spectre standalone that won't be created and it will then fall back to the default assumption of the structure above.

Regards,

Andrew.

This is a question about namespaces. In OCEAN data access functions (e.g. v(), i(), getData(), pv() etc), the net name/terminal name can be specified either as a "schematic" name or a "netlist" name. If the name begins with a "/", then it refers to a "schematic" name - i.e. the hierarchical name of something within the Virtuoso database. Otherwise it refers to a name that is in the netlist - in which case it has to refer to spectre's naming rules (i.e. dot "." as the hierarchy separator, and ":" for terminal access or operating point parameter access).

When the schematic is netlisted, you end up with directories called map and amap in the netlist directory which contain the information about how schematic names are mapped to netlist names (in particular this is important if reserved words or illegal characters are used). The names you ask to save get translated via this mapping informaton into corresponding netlist names - so if you ask to save /V2_OUTP_Q/PLUS in OCEAN, it will correspond to a "save V2_OUTP_Q:p" in the netlist. Similarly post-simulation, it uses this mapping information to allow you to access signals using the schematic names (the result database itself only ever contains netlist names).

In order to do that mapping however, OCEAN needs to know where the amap directory is. The simplest way of achieving that is to ensure that you have:

someDirectory
  --> netlist
          --> amap dir
          --> map dir
          --> input.scs
  --> psf

You run the simulation from inside the netlist dir, and then use "spectre -raw ../psf input.scs"

The expectation from the OCEAN access functions is that you have netlist and psf alongside each other. There are ways around that if OCEAN runs the simulation because it can create a runObjFile in the psf dir to tell it where the netlist dir is, but if running spectre standalone that won't be created and it will then fall back to the default assumption of the structure above.

Regards,

Andrew.