[Keith]

Currently no. I guess this could be done, though – will have to look into what's involved. At the moment the only thing saved is the window size.

[Keith]

If you installed in GLADE then set GLADE_HOME to GLADE, not GLADEGLADE4_W

Normally on Windows a DLL is searched for in the same directory as the executable, after that is will go and look in C:windowssystem32.

[Keith]

Sorry, not yet. I have only compiled and tested on CentOS 4.7

[Keith]

The latest version has a different via model implementation after finding excessive time taken importing DEF with vias that had many thousands of cut shapes.

Beware of writing out a techfile after importing LEF/DEF in this sort of case – the via may have a logical line length that is very long – currently the limit on DEF line length (length of statement up until terminating ';' character) is 32768 characters, which is much greater than the spec of 2048, but still easily breakable.

[Keith]

Actually it depends on what you mean by assign a techfile to a library.

If you open Glade and create an new empty library, it will have a default techfile created. However the techfile will only have the required system layers.

If you import e.g. GDS2 into a new library, Glade will create a techfile with layers for each GDS2 layer/datatype pair found in the GDS2. They will have random colours assigned, and empty fill patterns.

If you import an existing techfile into a new library, then obviously all the info in that techfile gets assigned to the library.

If you save a library to disk using the File->Save Lib command, the techfile info is stored with the library and gets loaded automatically when you open the library again.

hope that helps.

[Keith]

I've added the min/max area rule to the geomArea command. See the example drc.py file for how to check a contact size. I have uploaded the windows version 4.0.32, other platforms to follow

[Keith]

Actually thinking about it, I could modify the geomArea check to take a max value as well as the minimum value.

So for example if a contact has to be 0.25um on a side, you could specify the min area as 0.0625 and the max as 0.0625, then any contact not exactly 0.25u on a side would fail. Would that make sense?

[Keith]

This could be done – I presume the check would just specify the layer and length of a side of the square, and any shapes on that layer whose edges do not equal the specified length are marked in error?

[Keith]

Hi Paco,

Currently I've got basic MOS devices extracted, the intention is to add other devices too. In terms of parasitics, you mean e.g. parasitic capacitance between nets and ground? A simple area/perimeter extraction would probably be not too difficult.

Going beyond that for more complex parasitic extraction e.g. 2.5D/3D effects is unlikely as this would need significant work.

Glade can currently output Spice netlists – see File->Export->CDL. Currently the netlister just outputs flat netlists from an 'extracted' view generated by running extraction.

– Keith

[Keith]

The latest build fixes some problems introduced in 4.0.30 due to the library browser becoming a dockable widget. One unexpected side effect of this was that certain bindkeys would get assigned weirdly, in particular many keys were mapped to the 'copy' command.

Another fix for a bug that has been in the code for a while corrects an issue where GDS attributes are used (e.g. instance names or net names). These would overwrite instance master cell names resulting in a corrupt database after stream in.

There has been a long standing bug that when displaying multiple windows in MDI mode, switching between them using next/previous could cause the window to go blank (well, black) until a redraw was done. This has been fixed.

I've also updated the OpenGL code to use framebuffer objects in preference to pbuffers if they are available. It seems to be working fine but if you see any weirdness please let me know.

[Keith]

4.0.30 is checked in for Win32, Linux 32 and 64 bit. This is mainly a bugfix release, the most visible change is the library browser is now a dockable widget like the LSW, hierarchy and net browsers.

[Keith]

OK, on further investigation there were some problems specifically with creating instances of pcells from lib1 in cells in lib2.

These have now been fixed and checked into version 4.0.30 available for download today.

[Keith]

Several bugfixes in this version, also a few enhancements. Please check out the change log in the documentation.

[Keith]

Seems to work fine when I tried it:

– import techfile into lib1
– import techfile into lib2
– create new cell in lib1 called test1
– create new cell in lib2 called test2
– in cell test2, create instance of 'lib1 test1' cell

Note that technology files for the two libs should have the same layers etc.

[Keith]

Version 4.0.27 adds a new capability to the LSW – you can change the order of layers.

Originally the order that layers were drawn in was in the order that they were defined in the techfile – or the order they are encountered in GDS/OASIS/LEF etc.

Now you can change this ordering. Left clicking and dragging a layer in the LSW allows to to drop the layer above another layer.

Saving a techfile or saving a library now saves the layers in the current order.

[Author]

Posts