On Friday, March 30, 2012 10:56:32 PM UTC-4, dji wrote:
> How will task 3 be marked? I understand that we have to get the pipeline version correct, and working, but is there a implied requirement on performance or throughput improvement? My implementation for Task 1 already finishes 1 instruction every 2 cycles, and it is kind of hard to improve on that, given that there aren't that many architected registers.
You just have to get a correct implementation. Performance enhancement is not a requirement.