Paper posted
I have finished Revision 0 of Wolfram Machines: Design and Implementation, my paper on the theory behind Wolfram Machines.
posted by John Ohno at 3:26 PM
0 comments
Thursday, February 24, 2005
Welcome
Welcome to the Cell Machine blog, a running commentary and list of important events on the Wolfram Machine Project, an effort to create a computer architecture based off of the mathematical concepts presented in Steven Wolfram's book A New Kind of Science.
First off, let me tell you how exactly this works.
The basis for cellular arithmetic, the expanded mathematics used by the project, is the one-dimensional cellular automata. There are 256 different possible rules for a standard binary one-dimensional cellular automata, and they are numbered by their binary representation. With the knowledge of how these are generated, I made the Cell Gate, a boolean algebra equation to give the value of a cell based on an input- and rule- set. I then generated the Condition-Reflexion-Cell (CRC) gate, an extended form able to do both standard cellular arithmetic and traditional arithmetic (for how this is done, check out my paper, Wolfram Machines: Design and Implementation, when it is uploaded). The Cell Unit (CU), a cellular-arithmetic form of the ALU, is used in place of or beside the ALU in the Wolfram Machine's CPU. The memory segmenting, etc., is customized for the purposes of CRC arithmetic; for example, variable word lengths and memory divided up into lines before any larger unit. A Cell Machine, a machine capable of doing CRC arithmetic on the hardware level, is useful for many things, and because of the expanded arithmetic operation set and the much smaller amount of necessary circuitry, it can not only go faster at normal clock speeds (assuming CRC shortcuts are taken), but it can also hypothetically go at a much higher clock speed, and with variable word length, it can transfer different amounts of data per chunk, optimizing the transfer efficiency.
The first landmark was yesterday, when I figured out the most elegant equation for the Cell gate and deduced most of the rules necessary for traditional arithmetic operations with the CRC gate. Today, I plan to work out the rest and incorporate them into the paper.
When the design work is done, I plan to test the design in a minicomputer fashion: I will build a Cell machine out of discrete electronics and test it. This computer will be called the Yggdrasil (unless anyone can come up with a better name), and will likely be the first ever Wolfram Machine in hardware.
For more info, visit the site, http://wolfram-machine.cjb.net , the sourceforge project page, http://sf.net/projects/wolfram-machine , or the discussion forums, http://wolframmachine.proboards36.com .
~John Ohno
First off, let me tell you how exactly this works.
The basis for cellular arithmetic, the expanded mathematics used by the project, is the one-dimensional cellular automata. There are 256 different possible rules for a standard binary one-dimensional cellular automata, and they are numbered by their binary representation. With the knowledge of how these are generated, I made the Cell Gate, a boolean algebra equation to give the value of a cell based on an input- and rule- set. I then generated the Condition-Reflexion-Cell (CRC) gate, an extended form able to do both standard cellular arithmetic and traditional arithmetic (for how this is done, check out my paper, Wolfram Machines: Design and Implementation, when it is uploaded). The Cell Unit (CU), a cellular-arithmetic form of the ALU, is used in place of or beside the ALU in the Wolfram Machine's CPU. The memory segmenting, etc., is customized for the purposes of CRC arithmetic; for example, variable word lengths and memory divided up into lines before any larger unit. A Cell Machine, a machine capable of doing CRC arithmetic on the hardware level, is useful for many things, and because of the expanded arithmetic operation set and the much smaller amount of necessary circuitry, it can not only go faster at normal clock speeds (assuming CRC shortcuts are taken), but it can also hypothetically go at a much higher clock speed, and with variable word length, it can transfer different amounts of data per chunk, optimizing the transfer efficiency.
The first landmark was yesterday, when I figured out the most elegant equation for the Cell gate and deduced most of the rules necessary for traditional arithmetic operations with the CRC gate. Today, I plan to work out the rest and incorporate them into the paper.
When the design work is done, I plan to test the design in a minicomputer fashion: I will build a Cell machine out of discrete electronics and test it. This computer will be called the Yggdrasil (unless anyone can come up with a better name), and will likely be the first ever Wolfram Machine in hardware.
For more info, visit the site, http://wolfram-machine.cjb.net , the sourceforge project page, http://sf.net/projects/wolfram-machine , or the discussion forums, http://wolframmachine.proboards36.com .
~John Ohno
posted by John Ohno at 5:35 AM
0 comments
