On Thu, Aug 14, 2008 at 2:24 PM, Petter Urkedal <[EMAIL PROTECTED]> wrote: > On 2008-08-14, Timothy Normand Miller wrote: >> One other thing. I noticed that the dequeue signals in the HQ wrapper >> are registered. That means they're a cycle late. That's okay if we >> only dequeue from a given queue every other cycle. A dequeue occurs >> when both valid and deq are asserted on the same cycle, so to make >> this correct, we need to generate them combinatorially. Of course, >> that too can be a problem, in terms of meeting timing. > > Yes, that was a bug. I committed a fix. > > After converting poll_pci() to assembler, we should be able to test the > bridge again the same way you tested it before integrating HQ.
Another goofy thing that we're going to have to think about... The S3, when doing memory access, relies on reads being in even numbers of 32-bit words. So if we turn caching off, but HQ is not running, then a memory read will turn into a single-word request, which the S3 won't handle correctly. I think I may not try to fix that, though. Basically, we can only turn off memory caching if HQ is intercepting traffic. If HQ gets a single word read request from memory (and it doesn't have it cached somewhere), then it must make at least a 2-word request to the S3 and then return the correct half of what comes back. -- Timothy Normand Miller http://www.cse.ohio-state.edu/~millerti Open Graphics Project _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
