Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!uakari.primate.wisc.edu!pikes!aspen.craycos.com!pmk
From: pmk@craycos.com (Peter Klausler)
Newsgroups: comp.arch
Subject: Re: bizarre instructions
Message-ID: <1991Feb23.223014.12116@craycos.com>
Date: 23 Feb 91 22:30:14 GMT
References: <9102210042.AA12291@ucbvax.Berkeley.EDU> <2998@charon.cwi.nl> <3386.27c55e18@iccgcc.decnet.ab.com>
Organization: Cray Computer Corporation
Lines: 18

In article <2998@charon.cwi.nl>, dik@cwi.nl (Dik T. Winter) writes:
>  >         I would like an instruction which counts the number of ones in
>  > the binary representation of an integer but I find it hard to argue
>  > that this would be widely used.
> Ask Seymour Cray why that instruction was put in the Cray-1 as an
> afterthought.

Almost right. A scalar population count instruction was not new to the CRAY-1,
since the CDC 6600 and descendents all supported it ("CXi Xk", opcode 47).
What was missing on some CRAY-1As and CRAY-1Bs was a vector population count
instruction. The S and M models do support that.

The CRAY-1 (and X-MP/Y-MP followons) *is* missing a vector leading zero count
instruction, though it's there in scalar form. Need a CRAY-2 for the vector
version.

Both operations are quite useful, as has been noted in comp.arch often in the
past.


Brought to you by Super Global Mega Corp .com