Bart
Guest
|
 Posted:
Sun Jul 17, 2005 12:15 am Post subject:
Re: Where do the gains of OOO architecture actually take eff |
|
|
I was thinking along those lines (regarding it being able to execute
into the next loop iteration) but each iteration depends on the results
of the previous iteration so it seems as though there would be a limit
to how far it could go (less than 1 iteration ahead.) Unless dynamic
scheduling allows for instructions in the pipe which do not have their
operands computed yet?
So, given a loop where each iteration is completely independent of the
previous one, dynamic scheduling + register renaming + branch
prediction do make it possible to put successive iterations into the
pipeline before previous ones are complete, correct? |
|
FAQ
Memberlist
Register
Profile
Log in to check your private messages
Log in