(ICU 04 Item 2 ) ------------------------------------------------ [10/07/04]
Subject: Cadence Tool Interoperability
PLAYS SO-SO WITH OTHERS: It should be no surprise that Cadence tools only
50% of the time nice with other Cadence tools -- it's the same percentage
of Cadence tools playing nice with non-Cadence tools. It just echos their
most-Cadence-tools-are-acquired heritage.
In your experience, do Cadence's point tools play nice with other
Cadence point tools (Y or N)?
Play Nice: ########################### 54%
Trouble: ####################### 46%
Do Cadence point tools play nice with non-Cadence point tools (Y or N)?
Play Nice: ######################## 48%
Trouble: ########################## 52%
Yes on Cadence to/from Cadence. No on Cadence to/from non-Cadence.
Virtuoso & Assura & ADE all play well together.
Hercules + Virtuoso sucks. Calibre + Virtuoso not integrated so well
either. Non-Cadence simulators + ADE absolutely unacceptable. Cadence
has broken third-party simulators in ADE, in the same style as Microsoft
broke RealPlayer.
---- ---- ---- ---- ---- ---- ----
Cadence reading Synopsys design constraints: It's getting better, but
it's been a long and tough battle. VoltageStorm / Hercules Interface
not supported)
---- ---- ---- ---- ---- ---- ----
Before FE, Cadence tools had various problems interfacing with other
tools. With FE, the interface to other tools is good.
---- ---- ---- ---- ---- ---- ----
No. Different versions of SE are/were not compatible DEF read/write with
newer versions.
---- ---- ---- ---- ---- ---- ----
Playing nicer with each other steadily. 4 Years ago, interaction was
terrible. Thanks mostly to the OA jihad, most play pretty well together
today. We recently turned on the triangle of "FE, Virtuoso, CCAR" using
Open Access to get back and forth. Worked surprisingly well. The
Encounter-to-Physical-Compiler works pretty well. Most other connections
to non-Cadence are difficult and getting worse.
---- ---- ---- ---- ---- ---- ----
CeltIC takes in timing window files from PrimeTime, which is good.
---- ---- ---- ---- ---- ---- ----
Yes. Since Verplex LEC takes in RTL and gate-level netlists from DC,
PhysoOpt, or Astro, there isn't much that can go wrong here. CeltIC,
only the SPEF is read into the tool and it does SI analysis. Again,
isn't much that can go wrong.
---- ---- ---- ---- ---- ---- ----
No. Usually the tools are from different companies that Cadence acquired
and they have poor integration. Assura extraction needs lot of post-
processing to run SNA. Similarly to run VoltageStorm, we need mutiple
extractions...
---- ---- ---- ---- ---- ---- ----
No. We have a bad time getting PrimeTime and CeltIC synced up for
scripts and result correlation.
---- ---- ---- ---- ---- ---- ----
Many times rules get dropped in Constraint Manager when you update a
database. Net-in's from Viewlogic often drop information.
---- ---- ---- ---- ---- ---- ----
Cadence's tools all seam to work well with each other although we are
only using the basic tools Composer, Spectre, Virtuoso XL, and Diva.
I did demo the CCAR router which looks like a foreign tool and it
was not as well integrated with the rest.
---- ---- ---- ---- ---- ---- ----
No. We used to use BuildGates and Silicon Ensemble for synthesis, place
and route. There's no easy way to pass the scan chain information created
in BuildGates to SE to do scan chain reordering. Also it took us a while
to figure out the right settings to stream in the gds file generated by SE
back to Virtuoso.
---- ---- ---- ---- ---- ---- ----
I had a trouble with a First Encouter and PhysOpt flow.
---- ---- ---- ---- ---- ---- ----
No -- NC-Sim does not play nicely with Cobalt/Palladium.
|
|
