Subject: Re: [sv-ac] OVA donation presentation
From: Stephen Meier (Stephen.Meier@synopsys.com)
Date: Mon Sep 23 2002 - 11:41:15 PDT
Alain:
Thanks for the questions on how variables are handled in OVA.
In the OVA donation (v2.0) the variables are bound to the clock semantics
and are updated at each time the sampling event occurs. This means that
the variable assignments occur outside of the implicit threads and are
referred to by the threads. This supports some level of data integrity
checking,
but is not a complete solution.
You have expressed a valid and important requirement to have variables
which are created once per implicit thread of evaluation. While the
donation does not meet this requirement, Synopsys agree's that it is
an important requirement to include in the language definition and if supported
as a requirement in the ballot we can work to include in DWG.
Steve
--------
At 09:44 AM 9/19/2002 -0700, Alain Raynaud wrote:
>I attended a Synopsys training on OVA, but didn't get an answer to this
>question:
>
>can OVA describe data integrity constraints or checkers, or is it
>restricted to mostly "protocol" checkers?
>
>A simple example (there are more complicated cases): a split-transaction
>bus where requests have tag ids. A protocol checker can check that every
>Request is followed by an Ack. A data checker would also make sure that
>the tag ids match both request and ack.
>
>OVA has a variable construct, but it wasn't clear whether a local copy
>is created for each implicit thread that the temporal expression may be
>spawning.
>
>Thank you for clarifying that point.
>
>I'm assuming such features are part of the sv-ac requirements. To me at
>least, they are critical.
>
>Alain Raynaud
>Tensilica, Inc.
Steve Meier (stephen.meier@synopsys.com) W: 650-584-4476, Cell: 408-393-8246
This archive was generated by hypermail 2b28 : Mon Sep 23 2002 - 11:42:39 PDT