[Open-graphics] HQ stub in place

[Timothy Normand Miller]

On Wed, Jul 30, 2008 at 4:12 PM, Petter Urkedal <urkedal at nbi.dk> wrote:
> On 2008-07-30, Petter Urkedal wrote:
>> One thing which is not clear to me is what to register at what needs to
>> be combinatorial.  I think the critical part is reads.  The input ports
>> needs to be decoded from the port address (val_or_addr) and fed back to
>> HQ and multiplexed into wb_val_o in one cycle, or is there another way?
>
> Correction:  The output wb_val_o is already muxed from data registered
> in oga1hq_stg4_memio.  If we can afford another level of logic here, we
> can register the read-data before passing it back to HQ.  (It may be
> tight on timing, since wb_val_o goes to register forwarding.)
> Otherwise, we could try to pass it combinatorially as a replacement for
> the current input_port placeholder.

I think we should look towards finding ways to increase the clock
speed on this over time, even if that puts it on a different clock
from the bridge.  As such, we should be very careful about having too
many layers of logic.  Of course, the double-clocked register file is
probably going to kill us anyhow.



-- 
Timothy Normand Miller
http://www.cse.ohio-state.edu/~millerti
Open Graphics Project