Hinary Processor-with Processing speed more than exaflops (A PVS Research Consultancy Project)

At some point, we all reach our limits – personally, physically, or in terms of what is technically feasible. When this happens, we can give up, turn back, admit defeat – or we can push on and reach new heights. The new Hinary processor pushes back boundaries with outstanding performance and extraordinary efficiency. Its number crunching qualities are undisputed. This kind of performance requires nothing short of a technology explosion; And maximum throughput. The 1 kHz processor (eight bit wide) delivers 18,000 petaflops of calculation. But without ever overextending itself. There has never been a better opportunity to push the limits; the new Hinary processor. It is a system of using decimal or higher number system in processor; I also designed a sample circuit for representing decimal system electronically using transistors, which was major hurdle in designing such a system.

Increase the amount of data processed and instructions executed are of paramount importance to maximize processor efficiency. In order to increase amount of data processed and instructions executed; I have successfully developed a new concept after spending several years to research. This concept has a very important practical application since current binary processors are nearing end of architectural and physical limit. The new concept use extremely less number of transistors to process same number of data compared to existing processors.

Concept Details & Advantages

Using binary number system for process data is the major bottleneck in manufacturing high capacity future processors. In order to solve this problem I have found a new solution. The new concept is described below

Instead of using binary numbers we can use a quintal (a number system having 4 digits) or Ternary or octal number system for processing data, by this way we can process more data per cycle.

Sample electronic representation of a quintal can be 0v, 2v, 4v and 6v (or 0v, 5v, 10v and 15v). For triplet -5v, 0v and 5v can be used. For example in an 8 bit length data a binary system can process only 2^8 (256) but in a quintal system can process 4^8 (65536) bits of information. Quintal system can process a square of number of data than binary number. This technology can be used in electronic data processing. In the era of Ultra HD Video this concept has very high potential.

If single instructions like “shift” or “add” take a cluster of transistors called a “unit cell”. Then an octal number system based processors can execute 8^n times numbers per cycle using n unit cells compared to 2^n numbers executed by binary computers. For example: - in an 8 bit length data a binary system can process only 2^8 (256) but in a octal system can process 8^8 (16777216 or 16million) bits of information

You can figure out the amazing computing ability of new concept using an example of a pocket calculator processor with 1kHz processor, if an eight bit wide binary Number system (a number system with base 2) can process 2^8= 256; then a 256nary Number system (a number system with base 256) can process 256^8 = 18,446,744,073,709,551,616 or 18,000 petaflops of calculation. So that you can assume a pocket calculator sized device (using the new concept) will be thousands of times faster than today’s fastest supercomputer. One kHz speed means transistors can operate at near threshold voltage reducing power consumption to minimum.

In quintal system only 4 numbers of data segments is required for representing numbers up to 256 but in binary system 8 segments are required. Thus same space (number of transistors) can process more data in quintal system than binary system (number of data processed by binary raised to the power of two using same frequency). Only amount of setting voltage of gates is different.

In the above diagram decimal number 255 is represented in both quintal and binary system. You can clearly see that binary system requires more number of segments or transistors.

The above diagram shows number of data that can processed by a 4 bit wide segment. The quintal system can process 256 whereas binary can only process 16. The difference is 16² =256.

The above diagram shows number of data that can be process ed by a 8 bit wide segment. The quintal system can process 65536 where as binary can only process 256. (256² =65536). Thus it is clear that quintal system can process number of data equal to the number of data processed by binary raised to the power of two using same frequency. Theoretical support for this calculation is that number of data processed by a binary system in n wide system is 2ⁿ and quintal will be 4ⁿ and 4 = 2² ;so 4ⁿ = 2^2*n .

That means this system can cram 1mips to a 1kips processer or 1 terabits to a 1 megabit processer. 500,000 terabits to a 500 gigabit processer module

The quintal form of decimal number 141 is 2031. It is shown in the figure on the right side.

Conversion of Binary to Quintal System can be done electronically for processing. And conversion of Quintal to Binary System can be done electronically.

Salient features of new concept are we can start with decimal number system to make a supercomputer computing power on handheld devices then after some years we can upgrade the system to exponentially higher speeds by using hexadecimal number system, then to number system with 24 as base then to 32 base to 48 base so on to more than 1024 as base of number system. So the new computing system using higher nry number system will be future proof for more than 100 years or even 1000 years. Software vendors as well as others will support these new up gradation since each upgrada5tion will require these software to be upgraded, up gradation frequency can be 4-8 years and system will be backward compatible however old software will be slower to run. For back ward compatibility we can use an emulator that changes number systems to appropriate ones.

In quantum computers higher number system can be used but those number system is limited by number of quantum states or combination of quantum states they cannot exceed more than 4 or 8 how ever in the new concept (Hinary computer) the base of number system can be 1024 or more only limited by dynamic range or noise floor. And above all practical quantum computers are 3 or 4 decades away. Quantum computing have been proven in the lab, using bulky expensive equipment instead of a cheap silicon chip.

Today’s tablets or smartphones require cloud computing to deliver full functionality; roughly one cabinet server is required per 100 tablets, not to mention bandwidth consumed. However using new concept all the processing can be done on tablet itself; at todays’ supercomputing power ( PetaFLOPS) speed.

Viability of Concept

In a grounded-emitter transistor circuit, as the base voltage rises the base and collector current rise exponentially, and the collector voltage drops because of the collector load resistor. The relevant equations:

VRC = ICE × RC, the voltage across the load (the lamp with resistance RC)

VRC + VCE = VCC, the supply voltage shown as 6V

If VCE could fall to 0 (perfect closed switch) then Ic could go no higher than VCC / RC, even with higher base voltage and current. The transistor is then said to be saturated. Hence, values of input voltage can be chosen such that the output is either completely off,[23] or completely on. The transistor is acting as a switch, and this type of operation is common in digital circuits where only "on" and "off" values are relevant. In my concept input voltage is chosen such a way that there are two in between stage between completely on and completely off state of output or totally 4 stages of voltages.

Theory

In this new processor, in a quintal cell processing device, which processes more than one voltage level per cell (4 voltage levels), the amount of current flow is sensed (rather than simply its presence or absence), in order to determine more precisely the level of voltage on the transistors.

To easily understand the speed this new concept provides in a octal (number system with base 8) versus binary system visualize octal as 8 storied Vehicle and binary as double storied Vehicle travelling through a particular station ( station equivalent to an operation in computer) passing one Vehicle per second . Then in 4 lanes in one seconds double storied Vehicle commutes 2 X 2 X 2 X 2 or 2 ^ 4=16; but 8 storied Vehicle commutes 8 X 8 X 8 X 8 or 8 ^ 4 = 4096 operations . There is multiplication sign since 2 and 8 are bases of that number system for example in a 4 digit decimal number system can transfer 10 X 10 X 10 X 10 = 10000(0-9999), these 4 digits are represented by 4 lanes. In 4 seconds (or 4 cycles) the binary Vehicle will commute 2^4 + 2^4 + 2^4 + 2^4 =16+16+16+16=64; and octal Vehicle will commute 8 ^ 4 + 8 ^ 4 + 8 ^ 4 + 8 ^ 4 = 4096+4096+4096+4096 = 16384 so octal Vehicle bill be commuting 16384/64 = 256 faster. So in a real world example of a 64bit 1MHz processor; the binary system can calculate 2^64 X 1048576 = 1.9342813113834066795298816e+25; but the octal processor can calculate 8^64 X 1048576 = 6.5820182292848241686198767302294e+63 an trillion of trillion of trillion times faster (36 zeros or decimal spaces after 1 – actually 38 zeros).

Error - Correcting Codes

For correcting error in storage either parity bit or modified version of Hamming Error-correcting code can be used. Modified version of Hamming error correcting code is explained below.

We will discuss the code with the help of Venn diagrams, for simplicity we will confine only to 4bit data. Data is filled in intersecting inner compartments. Parity bit is added in non intersecting part of circle in such a way that total of the each number in a circle is a multiple of 4. If sums of two circles have errors and value is one less than ideal one is added to the data in the intersection of two circles. If the difference is two, two is added to the data.

In the second diagram 2circles are in error so we can find out that error is in the data in intersection so we can correct them. In the third diagram error is in two circles (first and third) but we cannot correct it by only changing the data in intersection. To correct it we have to change data in 3 intersections as shown in diagram colored red. We first change the data in such a way that sum of numbers in the intersections in second circle is 4 and sums of first and third circle differ to correct value by a common number.

You can upgrade the processer by using higher number system like hexadecimal (16) or even 256nry number system

You can use other quantum numbers in addition to spin quantum numbers to change number system into Hinary for example quintal system instead of binary system in quantum computers to implement the above technology for maximum processor efficiency and compactness.

Effects, once the proposal is taken

Ø Near 100% market share is possible since this is a nascent and novel technology

Ø Boosts brand equity exponentially.

Ø Can be a founding partner of ground breaking and world changing technology.

Ø Huge ROI.

Ø Opens new doors to unimaginable opportunities…

I strongly feel right now is the best time to develop and implement my concept HINARY PROCESSOR WITH SPEED MORE THAN PETAFLOPS since binary system computing is already saturated we cannot shrink processors below 6nm and Quantum computer is not ready for commercial applications.

X-------------------X

Inventor: Diji N J

Nedunghayil house

Vennala P.O.

Ernakulam

Kerala

India

Ph: +91-04844063025

Mobile: +91-7736419388

Contact: delvezone@gmail.com

© Diji N J and https://sites.google.com/site/hinaryprocessor/, 2009-2017. Unauthorized use and/or duplication of this material without express and written permission from this site’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Diji N J and https://sites.google.com/site/hinaryprocessor/ with appropriate and specific direction to the original content.