1.
Pipat Methavanitpong (M1)
November 19, 2012

2.
OUTLINE
Me
Fractional-order Sinusoidal Oscillator
Faster Microprocessors

3.
ME <=
 Name: Pipat Methavanitpong  Skills: have experienced many
 Nationality: Thailand programming languages both H/L level
 Graduated from: SIIT, Thammasat  [Experience does not mean
University in 2012 proficiency]
 Electronic and Communication  MATLAB, SIMULINK, OrCAD,
Engineering LabView
 Senior Project – Fractional-order  68HC11, 8086, Arduino, PLC
Sinusoidal Oscillator  C, C#, Java, Groovy, SQL, HTML,
 Work Experience: none CSS, PHP, VHDL, Flex
 Internship Experience: YES!!  Goal: Develop faster CPU than others in
 NECTEC Integrated Circuit the market
Development Section  My Current Work: Support Surachai-san
 Basic SystemC Syntax developing Dalvik extension to Lab’s
 Silicon Craft TCT processor
 Basic SRAM Schematic

4.
FRACTIONAL-ORDER SINUSOIDAL OSCILLATOR
Simple
MATH:

5.
FRACTIONAL-ORDER SINUSOIDAL OSCILLATOR
 What I did
 Follow Elwakil’s work
 he provides generalization of design of n-
fractional-order devices oscillator
 The only HOPE for my graduation!!
 Literature review on Fractional-order
devices
 Implement this knowledge in my advisor’s
Current Tunable Sinusoidal Oscillator ’87
 Result – It works and oscillates faster
 BUT, still have not fully understood what
fractional-order calculus is
 Very complex calculation
S. Pookaiyaudom, B. Srisuchinwong, and W. Kurutach,
“A Current-Tunable Sinusoidal Oscillator”, IEEE
Transactions on Instrumentation and Measurement, Vol.
IM-36, No. 3, September, pp. 725-729, Sep 1987.

6.
FRACTIONAL-ORDER SINUSOIDAL OSCILLATOR
 WHY none in market
 Creation of these devices is
NOT FEASIBLE
 Realization from a mesh of
recursive R and C structure
 Require LARGE area to
make it near ideal
performance
5-level stage becomes this
mess

7.
FRACTIONAL-ORDER SINUSOIDAL OSCILLATOR
1 level (LPF)
8 levels
12 levels
More levels -> More bandwidth

8.
FRACTIONAL-ORDER SINUSOIDAL OSCILLATOR
 HOPE, There is !
 Such characteristic is found in organic things e.g.
 There are reports of fabricated Si-devices for lab use.
 An Advantage from this knowledge
 More precise control on every conventional circuits
 Faster oscillator
 Better PID controller
 Any rate of attenuation electronic filter
 Greener electronic devices

9.
FASTER MICROPROCESSORS
How to become FASTER
YIN / YANG
An Era of Parallel Computing
Combination Dedicated Functionalities
Dark Silicon Gap

10.
FASTER MICROPROCESSORS
How to become FASTER
2 choices
 Work HARDER – Overclocking, Brute-force
 Work SMARTER – Better algorithms and management

11.
FASTER MICROPROCESSORS
 YIN / YANG
 Everything has both advantages and disadvantages
 Analog systems
 No loss of data
 Very sensitive to interference
 Digital systems
 Reconfigurable / Distortion Immunity
 Limited Range of Data (freq range)
 Smaller MOSFET
 Faster / Lower power
 Higher power density / Undeterministic Quantum Mechanic Behavior
 Single Electron Transistor
 Even lower power consumption
 Blurred digital state
 It is we, the engineers, whose task is to push through the limitation and shift to
new paradigm via BREAKTHROUGH

12.
FASTER MICROPROCESSORS
 An Era of Parallel Computing
 We cannot keep clock frequency rising
 Power consumption / Heat
 Move to the new paradigm
 Share works with friends
 Teamwork is the key
 Everybody may not be perfect
 But, everybody can take part in a work
to get it done
 But, as we know in every group work we
have faced as students, researchers,
employees, and etc.
 Unfair work distribution – Better Arbiter
 Waterfall workflow – Better Dataflow
 Communication problem - NoC
The ANALOGY of modern
microprocessors is now same as
URBAN PLANNING
Transportation – Communication between modules
Company – Functionality
People - Data

13.
FASTER MICROPROCESSORS
Combination Dedicated
Functionalities
 One does not fit all
 Give a right job to a right person
 AMD APU – A combination of CPU
and GPU on a single chip
 CPU – less core / more memory
 Control intensive
 GPU – more core / less memory
 Computation intensive
 CPU + FPGA – dynamic
functionalities

14.
FASTER MICROPROCESSORS
Dark Silicon Gap
 A term by H. Esmaeilzadeh etal. – Dark Silicon and the End of
Multicore Scaling ’12
 Underutiliztion of transistor integration capacity
 As a number of cores keep rising, the efficiency of utilization from
parallelization becomes WORST