coriolis/documentation/content/pages/rds/RDSpage.rst

446 lines
19 KiB
ReStructuredText
Raw Normal View History

Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
.. -*- Mode: rst -*-
.. Tools
.. |ocp| replace:: ``ocp``
.. |nero| replace:: ``nero``
.. |ring| replace:: ``ring``
.. |druc| replace:: ``druc``
.. |graal| replace:: ``graal``
.. |cougar| replace:: ``cougar``
.. |cif| replace:: ``cif``
.. |gds| replace:: ``gds``
.. |phseg| replace:: ``phseg``
.. |phvia| replace:: ``phvia``
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
.. RDS file syntax.
.. |MBK_TO_RDS_SEGMENT| replace:: ``MBK_TO_RDS_SEGMENT``
.. |MBK_TO_RDS_VIA| replace:: ``MBK_TO_RDS_VIA``
.. |MBK_TO_RDS_BIGVIA_HOLE| replace:: ``MBK_TO_RDS_BIGVIA_HOLE``
.. |MBK_TO_RDS_BIGVIA_METAL| replace:: ``MBK_TO_RDS_BIGVIA_METAL``
.. |MBK_WIRESETTING| replace:: ``MBK_WIRESETTING``
.. |ALL| replace:: ``ALL``
.. |DRC| replace:: ``DRC``
.. |EXT| replace:: ``EXT``
.. |VW| replace:: ``VW``
.. |LCW| replace:: ``LCW``
.. |RCW| replace:: ``RCW``
.. |ALUx| replace:: ``ALUx``
.. |CALUx| replace:: ``CALUx``
.. |TALUx| replace:: ``TALUx``
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
.. |ALU1| replace:: ``ALU1``
.. |POLY| replace:: ``POLY``
.. |NTIE| replace:: ``NTIE``
.. |PTIE| replace:: ``PTIE``
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
.. |NDIF| replace:: ``NDIF``
.. |PDIF| replace:: ``PDIF``
.. |PWELL| replace:: ``PWELL``
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
.. |NTRANS| replace:: ``NTRANS``
.. |PTRANS| replace:: ``PTRANS``
.. |CONT_DIF_N| replace:: ``CONT_DIF_N``
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
.. |CONT_DIF_P| replace:: ``CONT_DIF_P``
.. |CONT_BODY_N| replace:: ``CONT_BODY_N``
.. |CONT_BODY_P| replace:: ``CONT_BODY_P``
.. |CONT_POLY| replace:: ``CONT_POLY``
.. |CONT_VIA| replace:: ``CONT_VIA``
.. |CONT_VIAx| replace:: ``CONT_VIAx``
.. |C_X_N| replace:: ``C_X_N``
.. |C_X_P| replace:: ``C_X_P``
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
.. |RDS_NDIF| replace:: ``RDS_NDIF``
.. |RDS_NIMP| replace:: ``RDS_NIMP``
.. |RDS_ACTIV| replace:: ``RDS_ACTIV``
.. |RDS_GATE| replace:: ``RDS_GATE``
.. |RDS_POLY| replace:: ``RDS_POLY``
.. |RDS_ALU1| replace:: ``RDS_ALU1``
|newpage|
Symbolic Layout
===============
Symbolic Components
~~~~~~~~~~~~~~~~~~~
A symbolic layout is, in practice, made of only of three objects:
=========================== ============ ===================================================
Object |MBK| Explanation
=========================== ============ ===================================================
Segments |phseg| Oriented segments with a width and an orientation.
VIAs & contacts |phvia| Boils down to just a point.
Big VIAs & Big Contacts |phvia| Point with a width and a height
That is a rectangle of width by height centered
on the VIA coordinates.
=========================== ============ ===================================================
Each of thoses objects is associated to a *symbolic layer* which will
control how the object is translated in many *real rectangles*.
+---------+---------------+-------------+--------------------------------------------+
| |MBK| | Layer Name | Usable By | Usage |
+=========+===============+=============+============================================+
| |phseg| | |NWELL| | Segment | N Well |
| +---------------+-------------+--------------------------------------------+
| | |PWELL| | Segment | P Well |
| +---------------+-------------+--------------------------------------------+
| | |NDIF| | Segment | N Diffusion |
| +---------------+-------------+--------------------------------------------+
| | |PDIF| | Segment | P Diffusion |
| +---------------+-------------+--------------------------------------------+
| | |NTIE| | Segment | N Tie |
| +---------------+-------------+--------------------------------------------+
| | |PTIE| | Segment | P Tie |
| +---------------+-------------+--------------------------------------------+
| | |NTRANS| | Segment | N transistor, in |Alliance|, a transistor |
| | | | is represented as a segment (it's grid). |
| +---------------+-------------+--------------------------------------------+
| | |PTRANS| | Segment | P transistor |
| +---------------+-------------+--------------------------------------------+
| | |POLY| | Segment | Polysilicium |
| +---------------+-------------+--------------------------------------------+
| | |ALUx| | Segment | Metal level *x* |
| +---------------+-------------+--------------------------------------------+
| | |CALUx| | Segment | Metal level *x*, that can be used by the |
| | | | upper hierarchical level as a connector. |
| | | | From the layout point of view it is the |
| | | | same as |ALUx|. |
| +---------------+-------------+--------------------------------------------+
| | |TALUx| | Segment | Blockage for metal level *x*. Will |
| | | | diseappear in the real layout as it is an |
| | | | information for the P&R tools only. |
+---------+---------------+-------------+--------------------------------------------+
| |phvia| | |CONT_BODY_N| | VIA, BIGVIA | Contact to N Well |
| +---------------+-------------+--------------------------------------------+
| | |CONT_BODY_P| | VIA, BIGVIA | Contact to P Well |
| +---------------+-------------+--------------------------------------------+
| | |CONT_DIF_N| | VIA, BIGVIA | Contact to N Diffusion |
| +---------------+-------------+--------------------------------------------+
| | |CONT_DIF_P| | VIA, BIGVIA | Contact to P Diffusion |
| +---------------+-------------+--------------------------------------------+
| | |CONT_POLY| | VIA, BIGVIA | Contact to polysilicium |
| +---------------+-------------+--------------------------------------------+
| | |CONT_VIA| | VIA, BIGVIA | Contact between metal1 and metal2 |
| +---------------+-------------+--------------------------------------------+
| | |CONT_VIAx| | VIA, BIGVIA | Contact between metal *x* and metal *x+1*. |
| | | | The index is the the one of the bottom |
| | | | metal of the VIA. |
| +---------------+-------------+--------------------------------------------+
| | |C_X_N| | VIA | N transistor corner, to build transistor |
| | | | bend. Not used anymore in recent technos |
| +---------------+-------------+--------------------------------------------+
| | |C_X_P| | VIA | P transistor corner, to build transistor |
| | | | bend. Not used anymore in recent technos |
+---------+---------------+-------------+--------------------------------------------+
.. note::
Not all association of object and symbolic layers are meaningful.
For instance you cannot associate a contact to a ``NTRANS`` layer.
.. note::
The symbolic layer associated with blockages is prefixed by a ``T``,
for *transparency*, which may seems silly. It is for historical reasons,
it started as a true transparency, but at some point we had to invert
the meaning (blockage) with the rise of over-the-cell routing, but the
name stuck...
Symbolic Segments
~~~~~~~~~~~~~~~~~
In |Alliance|, segments are oriented (up, down, left, right). This disambiguate
the left or right side when using the ``LCW`` and ``RCW`` rules in the |RDS| file.
It allows to generate, if needed, asymetric object in the real layout file.
|bcenter| |SegmentOrientation| |ecenter|
|newpage|
The RDS File
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
============
The RDS file control how a symbolic layout is transformed into it's real
conterpart.
.. note:: **Unit used inside the RDS file:** all units are expressed in micrometers.
Alliance tools relying on the RDS file, and what layers are active for them:
======================================= ============= ===============================
Tool Name RDS Flags
======================================= ============= ===============================
Layout editor |graal| |ALL|
Design Rule Checker |druc| |ALL|, |DRC|
Electrical extractor |cougar| |ALL|, |EXT|
The symbolic to real layout translator |s2r| |ALL|
======================================= ============= ===============================
Physical Grid & Lambda Value
~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
RDS file: ::
DEFINE PHYSICAL_GRID 0.005
DEFINE LAMBDA 0.09
Tells that the physical grid (founder grid) step is 0.005µm and the lambda has
a value of 0.09µm. That is, one lambda is 18 grid steps.
We can distinguish two kind of |RDS| files:
* The *1µm* kind, odd segment widths and coordinates are allowed, but the ``LAMBDA``
value **must** represent an *even* number of foundry grid step.
* The *2µm* kind, segments widths and coordinates must all be even. And in that case
the ``LAMBDA`` value can be any multiple of the foundry grid.
The |MBK_TO_RDS_SEGMENT| table
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
The |MBK_TO_RDS_SEGMENT| table control the way segments are translated into
real rectangles. Be aware that we are translating *segments* and not *rectangles*.
Segments are defined by their axis (source & target points) and their width.
The geometrical transformations are described according to that model.
Obviously, they are either horizontal or vertical.
The translation method of a symbolic segment is as follow:
1. The segment is translated into one or more physical rectangles.
The generated rectangles depends on the tool which is actually
using |RDS| and the flag for the considered real layer.
For instance, real layers flagged with |DRC| will be generated
for |s2r| (for the |cif| or |gds|) and |druc|, but will not
be shown under |graal|.
2. Translation into one real layer. *First* the source & target coordinates and width
of the symbolic segment are multiplied by the ``LAMBDA`` value to obtain a real
segment. *Then* one of the |VW|, |LCW| or |RCW| transformation is applied to
that segment to get the final real rectangle.
* |VW| for Variable Width, expand the real layer staying centered from the
original one. In those rules, the third number is not used, it is only here
to make the life easier for the parser...
|bcenter| |RDS_VW| |ecenter|
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
* |LCW| or |RCW| for Left/Right Constant Width, create an off-center rectangle
of fixed width relatively to the real segment. Note that the ``SP`` number
is the distance *between the edge* of the real segment and the edge of the
generated real rectangle (*not* from the axis). It is often zero.
|bcenter| |RDS_LCW| |ecenter|
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
|newpage|
Examples: ::
TABLE MBK_TO_RDS_SEGMENT
# (Case 1)
ALU1 RDS_ALU1 VW 0.18 0.09 0.0 ALL
# (Case 2)
NDIF RDS_NDIF VW 0.18 0.0 0.0 ALL \
RDS_ACTIV VW 0.18 0.0 0.0 DRC \
RDS_NIMP VW 0.36 0.36 0.0 DRC
# (Case 3)
NTRANS RDS_POLY VW 0.27 0.00 0.0 ALL \
RDS_GATE VW 0.27 0.00 0.0 DRC \
RDS_NDIF LCW 0.0 0.27 0.0 EXT \
RDS_NDIF RCW 0.0 0.27 0.0 EXT \
RDS_NDIF VW 0.0 0.72 0.0 DRC \
RDS_ACTIV VW 0.0 0.72 0.0 ALL \
RDS_NIMP VW 0.18 1.26 0.0 DRC
END
:fboxtt:`Case 1` the |ALU1| is translated in exacltly one real rectangle of
|RDS_ALU1|, both ends are extended by 0.18µm and it's width is increased
by 0.09µm.
:fboxtt:`Case 2` the |NDIF| will be translated into only one segment
under |graal|, for symbolic visualization. And into three real rectangles
for |s2r| and |druc|.
:fboxtt:`Case 3` the |NTRANS|, associated to a transistor is a little bit
more complex, the generated shapes are different for the extractor |cougar|
in one hand, and for both |druc| & |s2r| in the other hand.
* For the extractor (|EXT| & |ALL| flags) there will be four rectangles
generateds:
1. The gate (|RDS_GATE|)
2. The left diffusion of the transistor (source or drain) (|RDS_NDIF|).
3. The right diffusion of the transistor (drain or source) (|RDS_NDIF|).
4. The active area (|RDS_ACTIV|).
As the extractor must kept separate the source and the drain of the transistor,
they are generated as two offset rectangles, using the |LCW| and |RCW| directives.
* For |s2r| and |druc| (|DRC| and |ALL|), five rectangles are generateds:
1. The poly (|RDS_POLY|).
2. The gate (|RDS_GATE|).
3. The diffusion, as one rectangle that covers both the |LCW| and the |RCW| (|RDS_NDIF|).
4. The active area (|RDS_ACTIV|).
5. The N implantation (|RDS_NIMP|).
In the layout send to the foundry, the source & drain are draw as one rectangle
across the gate area (the transistor being defined by the intersection of both
rectangles).
|newpage|
The |MBK_TO_RDS_VIA| table
~~~~~~~~~~~~~~~~~~~~~~~~~~
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
This table is to translate *default* VIAs into real via. In the symbolic layout
the default VIA is simply a point and a set of layers. All layers are converted
in squares shapes centered on the VIA coordinate. The one dimension given is the
size of the side of that square.
Note that although we are refering to VIAs, which for the purists are between two
metal layers, this table also describe *contacts*.
Example: ::
TABLE MBK_TO_RDS_VIA
CONT_DIF_P RDS_PDIF 0.54 ALL \
RDS_CONT 0.18 ALL \
RDS_ALU1 0.36 ALL \
RDS_ACTIV 0.54 DRC \
RDS_PIMP 0.90 DRC
CONT_POLY RDS_POLY 0.54 ALL \
RDS_CONT 0.18 ALL \
RDS_ALU1 0.36 ALL
CONT_VIA RDS_ALU1 0.45 ALL \
RDS_VIA1 0.27 ALL \
RDS_ALU2 0.45 ALL
END
.. note:: **In CONT_DIF_P** you may see that only three layers will be shown under
|graal|, but five will be generated in the |gds| layout.
The |MBK_TO_RDS_BIGVIA_HOLE| table
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
In |s2r|, when generating BIGVIAs, the matrix of holes they contains is
not draw relative to the position of the BIGVIA itself, but on a grid which
is common througout all the design real layout. This is to allow overlap
between two BIGVIA without risking the holes matrix to be not exactly overlapping.
As a consequence, when visualizing the |gds| file, the holes may not be centerend
inside one individual BIGVIA.
The |MBK_TO_RDS_BIGVIA_HOLE| table define the global hole matrix for the whole
design. The first number is the individual hole side and the second the grid step
(edge to edge). The figure below show the hole generation.
|bcenter| |BIGVIA_1| |ecenter|
Example of BIGVIA overlap:
|bcenter| |BIGVIA_2| |ecenter|
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
Example: ::
TABLE MBK_TO_RDS_BIGVIA_HOLE
CONT_VIA RDS_VIA1 0.27 0.27 ALL
CONT_VIA2 RDS_VIA2 0.27 0.27 ALL
CONT_VIA3 RDS_VIA3 0.27 0.27 ALL
CONT_VIA4 RDS_VIA4 0.27 0.27 ALL
CONT_VIA5 RDS_VIA5 0.36 0.36 ALL
END
.. note:: **BIGVIA demotion.** If the size of the bigvia is too small, there is
a possibility that no hole from the global matrix will be under it.
To avoid that case, if the either side of the BIGVIA is less than
``1.5 * step``, the BIGVIA is demoted to a simple VIA.
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
The |MBK_TO_RDS_BIGVIA_METAL| table
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
This table describe how the metal part of a BIGVIA is expanded (for the hole
part, see the previous table |MBK_TO_RDS_BIGVIA_HOLE|). The rule give for each
metal:
1. The *delta-with* (have to ask Franck).
2. The *overhang*, the length the real rectangle is expanded on each side from
the symbolic rectange.
Example: ::
TABLE MBK_TO_RDS_BIGVIA_METAL
CONT_VIA RDS_ALU1 0.0 0.09 ALL \
RDS_ALU2 0.0 0.09 ALL
CONT_VIA2 RDS_ALU2 0.0 0.09 ALL \
RDS_ALU3 0.0 0.09 ALL
CONT_VIA3 RDS_ALU3 0.0 0.09 ALL \
RDS_ALU4 0.0 0.09 ALL
CONT_VIA4 RDS_ALU4 0.0 0.09 ALL \
RDS_ALU5 0.0 0.09 ALL
CONT_VIA5 RDS_ALU5 0.0 0.09 ALL \
RDS_ALU6 0.0 0.18 ALL
END
The |MBK_WIRESETTING| table
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Bug fixes for MOSIS SCMOS_DEEP support. * Change: In Hurricane, the NetRoutingProperty is moved into Hurricane from Katabatic. Needed for Knik to be able to access thoses informations. * Change: In Hurricane, in RoutingPad::setOnBestComponent(), now in case of identical area, select the component of lowest id. This should not be needed if the component ordering was fully deterministic as it should be (will investigate later). This is to ensure that the choosen component is always the same, especially between save/load of a global routing. * Bug: In Katabatic, in AutoContactHTee::updateTopology(), invalidate the segments only if the topology is valid (no NULL in the cached segments). * Bug: In Katabatic, in GCellTopology::construct(), throw an error if the topology is bad instead of trying to continue (and core dump later... ). * Bug: In Kite, in BuildPowerRails, distinguish the name of the master net in the pad (for vddi, vssi, vdde, vsse, ck, cki & cko) and the name of the net in the *chip* netlist. Must use the later to make comparison as they may differs. * Change: In Knik, in save/load solution, exclude nets that are not globally routed by Knik. That is which NetRoutingProperty is not *Automatic*. * Bug: In Cumulus, in chip.BlockPower take account of the layer width extention to sligthy shrink the connector thus avoiding a notch with standart cell in some cases. * Change: In Cumulus, in chip.ClockTree disable the use of fixed Steiner trees for the leaf clocks, as it seems overconstrained for the router. First move was to lower them in M2/M3 (instead of M3/M4) but that was not sufficent. * New: In Cumulus, RSavePlugin for recursively saving a physical hierarchy. * New: In documentation, first embryo for RDS file. Should have been in Alliance git, but I prefer to keep newest doc in Coriolis.
2014-09-21 09:44:37 -05:00
From a strict standpoint this table shouldn't be here but put in a separate
configuration file, because it contains informations only used by the symbolic
layout tools (|ocp|, |nero|, |ring|).
This table defines the cell gauge the routing pitch and minimal (symbolic)
wire width and minimal spacing for the routers. They are patly redundant.
Example: ::
TABLE MBK_WIRESETTING
X_GRID 10
Y_GRID 10
Y_SLICE 100
WIDTH_VDD 12
WIDTH_VSS 12
TRACK_WIDTH_ALU8 0
TRACK_WIDTH_ALU7 4
TRACK_WIDTH_ALU6 4
TRACK_WIDTH_ALU5 4
TRACK_WIDTH_ALU4 3
TRACK_WIDTH_ALU3 3
TRACK_WIDTH_ALU2 3
TRACK_WIDTH_ALU1 3
TRACK_SPACING_ALU8 0
TRACK_SPACING_ALU7 4
TRACK_SPACING_ALU6 4
TRACK_SPACING_ALU5 4
TRACK_SPACING_ALU4 4
TRACK_SPACING_ALU3 4
TRACK_SPACING_ALU2 4
TRACK_SPACING_ALU1 3
END