[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [f-cpu] RC5, F-CPU and srotl

On Sun, Apr 07, 2002 at 11:24:56PM +0200, Yann Guidon wrote:
[...]
> But the first problem is still the most annoying : i had expected that
> Michael supported "real" SIMD operation. This comforts me in thinking that
> i have to write my on shift unit.

We already talked about the semantics of SIMD shift/rotate operations.
Quoting myself (from Jan 20, 2001):

	BTW, I encountered another open question: when doing SIMD shifts,
	is the second operand a SIMD operand, too?  Or will all chunks be
	shifted by the same number of bits?  My unit will do the latter,
	but the former might be interesting, too.

And your reply was:

	This question has been raised long ago, during a discussion with
	Thilo Reichelt.  I think that the operation was called "permute"
	or something like that.  In fact that possibilities are so large
	that it's difficult to make a good first decision. In particular,
	the SIMD words are not fixed-size and the extension to 1024 or
	even 65536 bits is not often straight-forward.

	We'll have to complete the Instruction Set Census and examine
	the possibilities.  We have enough free room in the ISA anyway
	so we can add interesting extensions :-)

So I decided to build the simple version first.

[...]
> >         An other thing about the srotl, it currently double the asm code.
> > It mean that, if we have a real srotl operation, we will have the same or
> > better performance than the K7 core (who is actually the best for this
> > algorithm). So my question is what is the cost of having a real srotl
> > instruction ?
> 
> >From my point of view, it would not be really expensive. I wonder
> what Michael thinks about this but after all, the SHL unit is "just
> a bunch of multiplexers"...

I'll have to investigate whether the omega network shifter can perform
that kind of operation at all. If it can, we'll need another set of
input muxes and some additional control logic.

-- 
 Michael "Tired" Riepe <Michael.Riepe@stud.uni-hannover.de>
 "All I wanna do is have a little fun before I die"
*************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/