FW: [sv-ec]E-mail Vote: Closes 12am PST October 26th 2007

Re-resend. I have tried several times to send this.

Shalom 

-----Original Message-----
From: Bresticker, Shalom 
Sent: Tuesday, October 23, 2007 1:56 PM
To: 'sv-ec@server.eda.org'
Subject: RE: [sv-ec]E-mail Vote: Closes 12am PST October 26th 2007

Note that in any case, your sort of formula assumes that the values are
from 0 to 2**M-1. This would in any case not work for the form described
in 18.5 where the value set is explicitly specified.

I also don't think that the current LRM requirement has a nice formula
either.

I assume that the compiler would do maximum precomputation and you would
get the following:

The first P bins contain Q values each and the remaining N-P bins
contain Q+1 values each, 

or the reverse: 

The first P bins contain Q+1 values each and the remaining N-P bins
contain Q+1 values each,

where the compiler would precompute P, Q, Q+1, and P*Q or P*(Q+1).

Then you would get 

bin number = min(VALUE/Q, P + (VALUE-P*Q)/(Q+1)) or
             min(VALUE/(Q+1), P + (VALUE-P*(Q+1))/Q).

Shalom
 
> -----Original Message-----
> From: owner-sv-ec@server.eda.org
> [mailto:owner-sv-ec@server.eda.org] On Behalf Of Bresticker, Shalom
> Sent: Tuesday, October 23, 2007 5:25 AM
> To: Steven Sharp; sv-ec@server.eda-stds.org
> Subject: RE: [sv-ec]E-mail Vote: Closes 12am PST October 26th 2007
> 
> Steven,
> 
> 1. I have no preference between the first bins or the last bins. I 
> chose the last bins only because the existing LRM language also puts 
> the extra values in the last bin.
> 
> 2. Do you have a convenient mathematical expression for the current 
> LRM formula, which has a 'discontinuous' formula?
> 
> Regards,
> Shalom
> 
> > -----Original Message-----
> > From: Steven Sharp [mailto:sharp@cadence.com]
> > Sent: Tuesday, October 23, 2007 1:02 AM
> > To: jonathan.bromley@doulos.com; sv-ec@eda-stds.org; Bresticker, 
> > Shalom
> > Subject: RE: [sv-ec]E-mail Vote: Closes 12am PST October 26th 2007
> > 
> > Shalom,
> > 
> > On 2055, I think it would be simpler mathematically if the extra 
> > values go into the first bins.  The expression for which
> bin a value
> > goes into becomes
> > 
> >   value/((2**M)/N)
> > 
> > instead of something messier like
> > 
> >   (((2**M)/N)-1)-((2**M)-value)/((2**M)/N)
> > 
> > Steven Sharp
> > sharp@cadence.com
---------------------------------------------------------------------
Intel Israel (74) Limited

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.