That IS an interesting idea. If that bit is 1, we have a larger immediate; if it's zero, then it's all other instructions. We could take it a step further and have a LI that's one bit; it always stores a 31-bit immediate into R0. Not very orthogonal, but it's a good way to get a very long immediate.
Do you think that 23 is short for graphics, because we use 24 a lot? The thing is, we'll probably never manipulate 24-bit RGB. Instead, ARGB (four 8-bit numbers) is pre-processed in the memory system and converted into four floats, then converted back on the way out. This is one reason memory operations have to be queue-based in a later revision. The thing about 31-bit immediates is that we can fit any positive float. And in graphics, those are probably more common than negative floats. Hmmm.... do we have a convenient way to flip the sign? Or maybe for negative numbers, it could be common to sacrifice one bit of precision at the bottom end... load a 31-bit immediate, shift left by one, and be done with it. Good enough for almost all cases and exactly right for half of all possibilities anyhow. We should give this one some serious debate. For the moment, I'm going to finish making the other changes, but I can add that in another revision. On Fri, Mar 22, 2013 at 6:28 PM, Chris Matrakidis <[email protected]>wrote: > > Modifying the ISA that I published earlier, what is the benefit of > having a > > zero register (thereby freeing up the space for the wr flag): > > - We could add one more bit to immediates > > - We could add one more opcode bit > > You can use this bit to identify the unconditional load-immediate (LI) > instruction and 5 bits for the register, leaving 26 bits for the > constant. It seems to me that 23 bit constants are a bit short for > graphics use. > > Best Regards, > > Chris Matrakidis > -- Timothy Normand Miller, PhD Assistant Professor of Computer Science, Binghamton University http://www.cs.binghamton.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)
