SRR available yet?

I've seen mention about 3 years ago about providing an API so that
cadence customers can load in psfbin files to a c/fortran/whatever
program for additional postprocessing.

Did this ever happen?

I have a tool which currently works with psfascii, but it would be
really nice to not use the ascii version as it gets very big and the
parsing overhead is somewhat painful. I have to believe reading psfbin
would be much faster.

I really don't understand what cadence gains by keeping the format
secret and not providing a library to access it. For example, the
mathworks documented their .mat files years and years ago and provide
libraries now for the new versions.

And no, rewriting the tool in skill is simply not an option. It is an
in house tool written in c++ originally for use with hspice but now we
use it with spectre.

Thanks
-Dan

 

danmc wrote: > I've seen mention about 3 years ago about providing an
API so that > cadence customers can load in psfbin files to a
c/fortran/whatever > program for additional postprocessing. > > Did
this ever happen? > > I have a tool which currently works with
psfascii, but it would be > really nice to not use the ascii version as
it gets very big and the > parsing overhead is somewhat painful. I
have to believe reading psfbin > would be much faster. > > I really
don't understand what cadence gains by keeping the format > secret and
not providing a library to access it. For example, the > mathworks
documented their .mat files years and years ago and provide > libraries
now for the new versions. > > And no, rewriting the tool in skill is
simply not an option. It is an > in house tool written in c++
originally for use with hspice but now we > use it with spectre. No
luck there. You could use the nutbin format. It works and is
documented. It's also messy (single file output, probably no largefile
support). There are matlab libraries for loading psf into matlab
workspace in VSDE. Unfortunately, with recent "enhancements" in Matlab,
it is probably slower than reading ascii. So there your are. I have
complained to the AE's a hundred times, but apparently they have no say
in the decision making process. I managed to talk to someone at R&D,
but they said nothing doing. They don't have any plans of supporting
open formats such as HDF5 either. Satya

 

Dan, please drop me an email from your real work email address (you can decode
mine).

Regards,

Andrew.

 

Can I also drop you my email from work? (In the hope that the reply will
be a full specification of the binary psf storage format so that Svenn
can finally write that commandline waveform postprocessor that he has
alwayw wanted to write to automate his sim documentation)

I have looked at some psfbin files with od and the format is really
messy. Encryption by format, I would say.

 

No, I don't have a full (or even partial) specification of the psf format.

Andrew.