Re: Pipe proposal feedback

Shabtay,

Shabtay Matalon wrote:
> 
> However, I am not sure that it is worth the time to define a common 
> method for setting H/W depth of the pipe. How important it is that we 
> define a unified way to set this attribute?

johnS:
The good news is that I've already proposed a mechanism
here and it is pretty simple. You might want to check the
last update to the specification (also the one I handed
out at the face-to-face).

> 
>          3. H/W max width of the pipe – since this needs to be compiled
> 
> [Shabtay] I am not sure why you care about width? Can you explain?

johnS:
I think the single call I've proposed for depth covers this also.

The call specifies depth in elements.

Simply call it for a depth of #elements/message * #messages
and you get a width and depth specified such that an
integral number of messages can be accomodated.

(Note: I'm using Per's proposed terminology of 'messages'
in place of 'transactions' I was using before).

> 
>          3. Note:  Data shaping & S/W depth x width can be configured
>             any time
> 
> [Shabtay] Data shaping – agree. What you mean by SW depth x width and 
> why do you need it?

johnS:
See above, I think the single proposed call for buffer depth
covers this - i.e. both dimensions.

-- johnS

______________________________/\
John Stickley                   \
Mgr., Acceleration Methodologies \
Mentor Graphics - MED             \_
________________________________________________________________