[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [oc] Memory model consolidation (Behavioral?)




----- Original Message -----
From: "Sam Gladstone" <samg@t-and-t.com>
To: <cores@opencores.org>
Sent: Tuesday, November 06, 2001 11:00 PM
Subject: Re: [oc] Memory model consolidation (Behavioral?)


> Ok, I looked the the tpram.
> I am assuming that you want the behavioral checks to go
> in the that last section. I think I can work to add them into both

Last section, exactly.

> the tpram and dpram.
>
> We can also create a register file memory model as well.
> (I use a model that stitches two dprams together to build a reg file
> in the SXP.  It also has the option of using a FF based reg file as well.)

I think the important is to have models for vendor rams. If certain uses of
RAMs are required in several designs, then we can also put register and
other common uses together in common. But I think it will be hard enough to
maintain vendor models so I would put register file for now.

>
> I guess my next question is, should these models be moved into

Idea is in the reuse. So yes, but we need to work also on infrastructure
that will allow sharing of common blocks...

> the common section? I don't think the tpram is there.

Feel free to put it there.

> I will update the SXP to use these models if we can keep them up to date.
> Here is an idea, can Miha do some type of link to the main common area
> memories?
> That way all repository checkouts will download the same version from the
> same
> place.  tar file downloads would also pull a copy of the link. I think you
> can
> do a link that make a whole copy when it copied to another source.

Exactly and this is already on Miha's TODO list. Basically you will be able
to set up symbolic links in the CVS repository. This way if you update
"your" local version or in common directory, the same file will be updated.

>
> Anyway, I will look into making changes to the tpram to do the extra
> checks and we can put it in the "common" misc section. I can then try
> and make the SXP use that memory model.

Great!

regards,
Damjan


--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml