Re: Subtleties of Booth's Algorithm Implementation
- From: rickman <gnuarm@xxxxxxxxx>
- Date: Wed, 24 Jun 2009 20:00:13 -0700 (PDT)
On Jun 24, 2:57 pm, Weng Tianxiang <wtx...@xxxxxxxxx> wrote:
On Jun 24, 9:54 am, Mike Treseler <mtrese...@xxxxxxxxx> wrote:
rickman wrote:
I was able to get a schematic from Synplify Pro, but it did not want
to include the text except on part of the drawing. If anyone can tell
me how to get the text to display, I will reprint it. The PDF file
can be found athttp://arius.com/stuff/FPGA/Multiply_16x16.pdf
Don't know about synplify.
Does it have an RTL viewer also?
This incarnation uses 62 LUTs and 42 FFs if I am reading the info
correctly. This includes 32 FFs to hold the output product. The
Multiplicand is not registered, it is assumed that it is held constant
on the input.
That should be ok if the enable is synchronized.
Interesting. Thanks for the posting.
Also, I have not simulated it to be sure it is coded correctly, but I
am pretty confident it is working the way I intend or at least any
mistakes won't change the LUT count much.
I prefer to start with an RTL sim,
but I know that many designers prefer
working on the bench. Good luck.
-- Mike Treseler
Hi Rick,
The file published athttp://arius.com/stuff/FPGA/Multiply_16x16.pdf
has some flaws:
I cannot see the texts after 4 clicks in width direction with 400%
magnification. It means the first 3 clicks in width show normal
drawings, but after that only schematics are shown normally, but no
texts are shown.
Is it normal?
Weng
Yes, I know the print is not very good, but that is the best I could
get out of Synplify. The only way I could get any text at all was to
set the magnification to the lowest level that would still display
text. Then only the portion visible on my screen would print with
text. The rest shows nothing. The large squares are two bit adder
primitives and the smaller rectangles are FFs. The rest is pretty
obvious since they show the gates.
Like I said, if anyone knows how to get Synplify to print the entire
schematic with text, I'll be happy to post that. I can email you the
code if you would like to see that. But like I said, there is nothing
here that isn't pretty obvious other than that you need an extra msb
to preserve sign if you need to support the most negative number (-2**
(n-1)).
Rick
.
- References:
- Subtleties of Booth's Algorithm Implementation
- From: rickman
- Re: Subtleties of Booth's Algorithm Implementation
- From: Mike Treseler
- Re: Subtleties of Booth's Algorithm Implementation
- From: rickman
- Re: Subtleties of Booth's Algorithm Implementation
- From: Mike Treseler
- Re: Subtleties of Booth's Algorithm Implementation
- From: rickman
- Re: Subtleties of Booth's Algorithm Implementation
- From: rickman
- Re: Subtleties of Booth's Algorithm Implementation
- From: Weng Tianxiang
- Re: Subtleties of Booth's Algorithm Implementation
- From: rickman
- Re: Subtleties of Booth's Algorithm Implementation
- From: Mike Treseler
- Re: Subtleties of Booth's Algorithm Implementation
- From: Weng Tianxiang
- Subtleties of Booth's Algorithm Implementation
- Prev by Date: Cable autodetection/programming the Xilinx Virtex2Pro FPGA failing.
- Next by Date: Luxury Living
- Previous by thread: Re: Subtleties of Booth's Algorithm Implementation
- Next by thread: CFP: Optical SuperComputing Workshop 2009
- Index(es):
Relevant Pages
|