RE: ISAC: Analyzed IR and please work on your action items

Chuck,

I've had some email exchange with the submitter on this. It's become clear
from that exchange that he was confusing the use of the word "exponent" in
7.2.7. He was interpreting it as referring to the exponent part of the IEEE
floating-point representation, rather than to the right operand of the "**"
operator. If the exponent part of the IEEE representation of a number is all
0 bits, the number is a denormal. He was misinterpreting the statement in
7.2.7, that an exponent of 0 yields a result of one, as saying that the bit
pattern 00000000000000xxxx...x represents the number 1.0, which is not the
case.

Now that I've clarified that with him, his questions is whether
exponentiation by a large negative exponent should yield 0.0 or a denormal
number. That question is addressed by the proposed resolution of the IR. As
you say, he may not be happy with the resolution, in which case he is at
liberty to propose a language extension to deal with IEEE standard
arithmetic.

Cheers,

PA

--
Dr. Peter J. Ashenden                peter@ashenden.com.au
Ashenden Designs Pty. Ltd.           www.ashenden.com.au
PO Box 640                           VoIP: 0871270078@sip.internode.on.net
Stirling, SA 5152                    Phone (mobile):  +61 414 709 106
Australia

> -----Original Message-----
> From: owner-isac@eda.org [mailto:owner-isac@eda.org] On 
> Behalf Of Chuck Swart
> Sent: Thursday, 27 October 2005 07:26
> To: isac@eda.org
> Subject: ISAC: Analyzed IR and please work on your action items
> 
> 
> Attached is the analysis of IR2070: Support for floating 
> point denormal 
> numbers
> 
> Please review this carefully, since the submitter will not be 
> happy with 
> the conclusion.
> 
> Also, please attempt to deal with your action items before the next 
> meeting, scheduled for:
> 
> 10 November 2005 at 6 pm Pacific Standard Time.
> 
> Chuck Swart
> 
> 
>