This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: complex numbers in gcc

From: Richard Henderson <rth at redhat dot com>
To: roy rosen <roy dot 1rosen at gmail dot com>
Cc: gcc at gcc dot gnu dot org
Date: Tue, 17 Aug 2010 09:06:18 -0700
Subject: Re: complex numbers in gcc
References: <AANLkTi=B_Zy3pF6MKjVro6p9XBJoSupU4FC0rO7phqPx@mail.gmail.com>

On 08/17/2010 06:20 AM, roy rosen wrote:
> 2. Is there a way to remove this pass and keep coherency of the
> following passes, so that in the RTL stage I would be able to handle
> complex numbers (CHI/CSI) and not the decomposed code?

By adjusting the rules as to what's legal GIMPLE, yes.  This
may require changes to several passes.

> 3. Is there any other port which have complex arithmetic instructions?

No.  I seem to remember that there used to be one, but
it was removed as unmaintained years ago.

The closest that seems to exist at the moment are vector
instructions that can be used to accelerate arithmetic
over arrays of complex numbers.  But for individual complex
operations those instructions don't help much.

r~

References:
- complex numbers in gcc
  - From: roy rosen

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]