how to implement Logarithmic Number System in VHDL

how to implement Logarithmic Number System in VHDL

Post by Visw » Thu, 12 Feb 2004 10:55:20



hi

I have some basic questions regarding Logarithmic number system.

First of all how is it represented? Are there any good references to
study this?

I have to design a processor which should be a low power device. This
processor has all the arithmetic operations to be performed on real
numbers like 0.000000984. Which arithmetic is suggestible? ( fixed
point or floating point or LNS ?)

I have to implement it in VHDL and download it to FPGA. how could I
represent all the log of numbers in VHDL? How could I do this using
Xilinx IP core generators? And how could I do this without using
Xilinx Ip core generators?

Any help would be greatly appreciated.

Thanks
Viswan

 
 
 

how to implement Logarithmic Number System in VHDL

Post by Norbert Juff » Thu, 12 Feb 2004 16:52:55



> hi

> I have some basic questions regarding Logarithmic number system.

> First of all how is it represented? Are there any good references to
> study this?

One possible 32-bit format is shown here: http://napier.ncl.ac.uk/HSLA/
Note that that page is part of the website for the European Logarithmic
Microprocessor project. They implemented a "32-bit LNS ALU on a Xilinx
XCV2000-E6 FPGA, clocked at 53MHz. They have obtained a throughput of
one add/subtract operation per 9 clock cycles, and one multiply, divide
or square root per single clock cycle."

-- Norbert

 
 
 

1. logarithmic number representation

Several people have commented that IBM is now experimenting with logarithmic
representations for floating-point numbers.   This is not new; people have
been doing that for a long time.  The proceedings of any IEEE Symposium on
Computer Arithmetic will have several papers on this subject.  The next
symposium is later this month in France.

Nobody has found any worthwhile advantage for logarithmic numbers for general-
purpose computation, but I have heard that they have been used in some
special-purpose embedded systems.

X3J11 adopted a conventional floating-point number representation as part
of its standard for ANSI-C.  I pointed out that there was no way that
<float.h> could be meaningfully defined on a system with logarithmic numbers,
and the committee's response was basically that was OK, those systems just
wouldn't support ANSI-C.
--

David Hough


2. 1605 and Frame relay

3. Books/docs on implementing File-systems

4. Problem transforming an XML file into XSL using the Choose statement.

5. How to implement an email archive system

6. Computing Clusters using QNX

7. Implementing IDE in embedded system

8. 300Mhz Ultra IIi, Ultra-Wide SCSI, pizza-box $3,150

9. Hierarchical Hypercube Multicomp. INs

10. HACF.... now implemented on most systems...

11. 8150, 3270, INS info wanted..

12. how to implement system calls using a SWI

13. Implement a JVM in an embedded system