ACCURATE FLOATING-POINT SUMMATION IN CUB

Size: px
Start display at page:

Download "ACCURATE FLOATING-POINT SUMMATION IN CUB"

Transcription

1 ACCURATE FLOATING-POINT SUMMATION IN CUB URI VERNER Summe inten

2 OUTLINE Who need accuate floating-point ummation?! Round-off eo: ouce and ecovey A new method fo accuate FP ummation on a GPU Added a a function to the open-ouce CUB libay How fat i it? Download link

3 NORMAL FLOATING-POINT SUMMATION INPUT RESULT INACCURATE RESULTS NON-DETERMINISTIC RESULTS!

4 ACCURATE FP SUMMATION INPUT EXACT SUM ACCURATE SUM AS FLOATING-POINT Ou method compute both!

5 EXISTING WORK: EXBLAS (OPENCL) By Iakymchuk, Collange, et al. Ue Kulich accumulato (vey wide fied-peciion vaiable) Ou method ue a diffeent appoach

6 WHERE IS THIS USEFUL? High Pefomance Computing application An eample coming net Co-platfom application Debugging: bit-eact eult fo floating point! if (d_eult!= d_efeence) eo( wong anwe! );

7 EXAMPLE: LATTICE QCD COMPUTATIONS QCD - Quantum Chomodynamic Decibe the tong foce that bind quak and gluon GPU acceleated QUDA libay lattice.github.io/quda Accuate ummation can potentially impove convegence and educe computation time

8 CONVERGENCE OF ITERATIVE ALGORITHM BiCGtab Algoithm: Diac equation olve convegence tagnation

9 ROUND-OFF ERROR: SOURCE AND RECOVERY

10 IEEE-754 FLOATING-POINT STANDARD 32/64 bit S EXP SIGNIFICANT DIGITS = ( 1) 2 EXP 1. D b ingle-peciion double-peciion Fomat width 32 bit 64 bit Eponent ange (8 bit) (11 bit) Significant digit 23 (+1 implicit) 52 (+1 implicit) Special cae: +/-NaN, +/-Inf, +/-0, ubnomal

11 SOURCE OF NON-REPRODUCIBILITY Non-aociative opeation Ode of opeation matte: 1,000,000 + ( ) -> 1,000,001 (1,000, ) > 1,000,000 diffeent implementation etun diffeent um value

12 SOURCE OF ACCURACY LOSS Round-off eo in compute opeation Compute um: accuate actual bigge diffeence in magnitude => moe digit lot

13 TWO-SUM ALGORITHM (KNUTH) TwoSum(a,b) Round(a+b) ound-off eo 6 FP opeation [,] = TwoSum(a,b) <- a+b z <- -b <- (b-(-z)) + (a-z)

14 FAST TWO-SUM (DEKKER) FatTwoSum(a,b) Round(a+b) ound-off eo 3 FP opeation Requie EXP(a) EXP(b) [,] = FatTwoSum(a,b) <- a+b z <- -a <- b-z

15 ERROR-FREE PARALLEL SUMMATION

16 INTEGRATION INTO CUB LIBRARY CUB: Paallel pimitive in CUDA Include paallel pimitive like Sum, Scan, Sot, etc. Pefomance tuned fo evey NVIDIA GPU achitectue Reduction Aim: ue Reduction with TwoSum() fo an eo-fee um

17 REDUCTION+TwoSum: PROBLEM #1 The output of TwoSum i two FP, intead of one! + = + = Paallel eduction TwoSum (1,2)=2Sum(1,2) (y1,y2)=2sum(1,2) 2=2+y1 (1,2)=fat(1,2) 2=2+y2 (3,3)=fat(1,2) 1. Convet (, 0.0) 2. Define (1,1) + (2,2) (3,3)

18 REDUCTION+TwoSum: PROBLEM #2 Limited accuacy E.g.: Multiple value with imila eponent can be added without oveflow

19 DIVIDE THE EXPONENT RANGE INTO BINS Numbe of EXP value: 2048 (double) add to bin #3 floo 1023 / = 3 3 (bin id)

20 HOW MANY EXPONENT VALUES PER BIN? binay digit Suppoe we add n numbe to a bin: a i = 2 e i m i, whee e l e i < e h. Ou budget i 106 digit log 2 n + e h e l a l = a h = Fo n = 2 20, e h e l = 32 diffeent eponent! e h e l

21 ALGORITHM: ERROR-FREE SUMMATION ON GPUS

22 EACH THREAD DOES THE FOLLOWING EXPONENT=11bit binid=6bit bin=5bit ead input adi-ot by binid educe-by-key binid update mem bin SHARED MEMORY

23 FINAL SUMMATION PHASE Bin Bin 0 Block 0 Block Reult SUM X (eial phae)

24 ALGORITHM SUMMARY 1. Fo each thead block: Repeat: Read input tile Radi-ot item by bin ID Compute um fo each bin with Reduce-by-key in egite in egite (+ temp buffe in haed) in egite (+ temp buffe in haed) Update bin in haed memoy in haed memoy Save bin to global memoy in global memoy 2. Mege bin with the ame bin ID 3. Nomalize bin by adding them fom low to high 4. Rounded eult i in the highet wod

25 GItem/ec PERFORMANCE (K40) Billion Item pe econd Cont Random 1 Nomal ummation i ~6 time fate 0 Numbe of item

26 DOWNLOAD AND CONTRIBUTE Get it at: Uage intuction in README.ACCUSUM It open ouce. Ue it, impove it!

27 THANK YOU

28 BACKUP

Chapter 19 Webassign Help Problems

Chapter 19 Webassign Help Problems Chapte 9 Webaign Help Poblem 4 5 6 7 8 9 0 Poblem 4: The pictue fo thi poblem i a bit mileading. They eally jut give you the pictue fo Pat b. So let fix that. Hee i the pictue fo Pat (a): Pat (a) imply

More information

A Deep Convolutional Neural Network Based on Nested Residue Number System

A Deep Convolutional Neural Network Based on Nested Residue Number System A Deep Convolutional Neual Netwok Based on Nested Residue Numbe System Hioki Nakahaa Ehime Univesity, Japan Tsutomu Sasao Meiji Univesity, Japan Abstact A pe-tained deep convolutional neual netwok (DCNN)

More information

Estimation and Confidence Intervals: Additional Topics

Estimation and Confidence Intervals: Additional Topics Chapte 8 Etimation and Confidence Inteval: Additional Topic Thi chapte imply follow the method in Chapte 7 fo foming confidence inteval The text i a bit dioganized hee o hopefully we can implify Etimation:

More information

COMP Parallel Computing SMM (3) OpenMP Case Study: The Barnes-Hut N-body Algorithm

COMP Parallel Computing SMM (3) OpenMP Case Study: The Barnes-Hut N-body Algorithm COMP 633 - Paallel Computing Lectue 8 Septembe 14, 2017 SMM (3) OpenMP Case Study: The Banes-Hut N-body Algoithm Topics Case study: the Banes-Hut algoithm Study an impotant algoithm in scientific computing»

More information

Multifrontal sparse QR factorization on the GPU

Multifrontal sparse QR factorization on the GPU Multifontal spase QR factoization on the GPU Tim Davis, Sanjay Ranka, Shaanyan Chetlu, Nui Yealan Univesity of Floida Feb 2012 GPU-based Multifontal QR factoization why spase QR? multifontal spase QR in

More information

Histogram Processing

Histogram Processing Hitogam Poceing Lectue 4 (Chapte 3) Hitogam Poceing The hitogam of a digital image with gay level fom to L- i a dicete function h( )=n, whee: i the th gay level n i the numbe of pixel in the image with

More information

Rotational Kinetic Energy

Rotational Kinetic Energy Add Impotant Rotational Kinetic Enegy Page: 353 NGSS Standad: N/A Rotational Kinetic Enegy MA Cuiculum Famewok (006):.1,.,.3 AP Phyic 1 Leaning Objective: N/A, but olling poblem have appeaed on peviou

More information

Algebra-based Physics II

Algebra-based Physics II lgebabased Physics II Chapte 19 Electic potential enegy & The Electic potential Why enegy is stoed in an electic field? How to descibe an field fom enegetic point of view? Class Website: Natual way of

More information

Topic 4a Introduction to Root Finding & Bracketing Methods

Topic 4a Introduction to Root Finding & Bracketing Methods /8/18 Couse Instucto D. Raymond C. Rumpf Office: A 337 Phone: (915) 747 6958 E Mail: cumpf@utep.edu Topic 4a Intoduction to Root Finding & Backeting Methods EE 4386/531 Computational Methods in EE Outline

More information

Truncated Squarers with Constant and Variable Correction

Truncated Squarers with Constant and Variable Correction Please veify that ) all pages ae pesent, 2) all figues ae acceptable, 3) all fonts and special chaactes ae coect, and ) all text and figues fit within the Tuncated Squaes with Constant and Vaiable Coection

More information

Chapter Eight Notes N P U1C8S4-6

Chapter Eight Notes N P U1C8S4-6 Chapte Eight Notes N P UC8S-6 Name Peiod Section 8.: Tigonometic Identities An identit is, b definition, an equation that is alwas tue thoughout its domain. B tue thoughout its domain, that is to sa that

More information

Section 25 Describing Rotational Motion

Section 25 Describing Rotational Motion Section 25 Decibing Rotational Motion What do object do and wh do the do it? We have a ve thoough eplanation in tem of kinematic, foce, eneg and momentum. Thi include Newton thee law of motion and two

More information

How can you find the dimensions of a square or a circle when you are given its area? When you multiply a number by itself, you square the number.

How can you find the dimensions of a square or a circle when you are given its area? When you multiply a number by itself, you square the number. 7. Finding Squae Root How can you find the dimenion of a quae o a cicle when you ae given it aea? When you multiply a numbe by itelf, you quae the numbe. Symbol fo quaing i the exponent. = = 6 quaed i

More information

Force & Motion: Newton s Laws

Force & Motion: Newton s Laws oce & otion: Newton Law ( t Law) If no net foce act on a body then the body velocity cannot change. Zeo net foce implie zeo acceleation. The ma of an object detemine how difficult it i to change the object

More information

Solutions Practice Test PHYS 211 Exam 2

Solutions Practice Test PHYS 211 Exam 2 Solution Pactice Tet PHYS 11 Exam 1A We can plit thi poblem up into two pat, each one dealing with a epaate axi. Fo both the x- and y- axe, we have two foce (one given, one unknown) and we get the following

More information

Particle Systems. University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell

Particle Systems. University of Texas at Austin CS384G - Computer Graphics Fall 2010 Don Fussell Paticle Systems Univesity of Texas at Austin CS384G - Compute Gaphics Fall 2010 Don Fussell Reading Requied: Witkin, Paticle System Dynamics, SIGGRAPH 97 couse notes on Physically Based Modeling. Witkin

More information

Determining solar characteristics using planetary data

Determining solar characteristics using planetary data Detemining sola chaacteistics using planetay data Intoduction The Sun is a G-type main sequence sta at the cente of the Sola System aound which the planets, including ou Eath, obit. In this investigation

More information

Detailed solution of IES 2014 (ECE) Conventional Paper II. solve I 0 and use same formula again. Saturation region

Detailed solution of IES 2014 (ECE) Conventional Paper II. solve I 0 and use same formula again. Saturation region etailed olution of IS 4 (C) Conventional Pape II qv qv Sol. (a) IC I e Ie K K 4 I =.7 Fo I C = m olve I and ue ame fomula again K IC V ln 5ln 4 q I.7 =.8576 Volt Sol. (b) VGS VS Vupply 5V N MOS channel,

More information

Basic propositional and. The fundamentals of deduction

Basic propositional and. The fundamentals of deduction Baic ooitional and edicate logic The fundamental of deduction 1 Logic and it alication Logic i the tudy of the atten of deduction Logic lay two main ole in comutation: Modeling : logical entence ae the

More information

V V The circumflex (^) tells us this is a unit vector

V V The circumflex (^) tells us this is a unit vector Vecto Vecto have Diection and Magnitude Mike ailey mjb@c.oegontate.edu Magnitude: V V V V x y z vecto.pptx Vecto Can lo e Defined a the oitional Diffeence etween Two oint 3 Unit Vecto have a Magnitude

More information

Supplementary Figure 1. Circular parallel lamellae grain size as a function of annealing time at 250 C. Error bars represent the 2σ uncertainty in

Supplementary Figure 1. Circular parallel lamellae grain size as a function of annealing time at 250 C. Error bars represent the 2σ uncertainty in Supplementay Figue 1. Cicula paallel lamellae gain size as a function of annealing time at 50 C. Eo bas epesent the σ uncetainty in the measued adii based on image pixilation and analysis uncetainty contibutions

More information

Development of Model Reduction using Stability Equation and Cauer Continued Fraction Method

Development of Model Reduction using Stability Equation and Cauer Continued Fraction Method Intenational Jounal of Electical and Compute Engineeing. ISSN 0974-90 Volume 5, Numbe (03), pp. -7 Intenational Reeach Publication Houe http://www.iphoue.com Development of Model Reduction uing Stability

More information

Rigid Body Dynamics 2. CSE169: Computer Animation Instructor: Steve Rotenberg UCSD, Winter 2018

Rigid Body Dynamics 2. CSE169: Computer Animation Instructor: Steve Rotenberg UCSD, Winter 2018 Rigid Body Dynamics 2 CSE169: Compute Animation nstucto: Steve Rotenbeg UCSD, Winte 2018 Coss Poduct & Hat Opeato Deivative of a Rotating Vecto Let s say that vecto is otating aound the oigin, maintaining

More information

Homework 1 Solutions CSE 101 Summer 2017

Homework 1 Solutions CSE 101 Summer 2017 Homewok 1 Soutions CSE 101 Summe 2017 1 Waming Up 1.1 Pobem and Pobem Instance Find the smaest numbe in an aay of n integes a 1, a 2,..., a n. What is the input? What is the output? Is this a pobem o a

More information

Teoría del Funcional de la Densidad (Density Functional Theory)

Teoría del Funcional de la Densidad (Density Functional Theory) Teoía del Funcional de la Densidad (Density Functional Theoy) Motivation: limitations of the standad appoach based on the wave function. The electonic density n() as the key vaiable: Functionals & Thomas-Femi

More information

763620SS STATISTICAL PHYSICS Solutions 2 Autumn 2012

763620SS STATISTICAL PHYSICS Solutions 2 Autumn 2012 763620SS STATISTICAL PHYSICS Solutions 2 Autumn 2012 1. Continuous Random Walk Conside a continuous one-dimensional andom walk. Let w(s i ds i be the pobability that the length of the i th displacement

More information

Merging to ordered sequences. Efficient (Parallel) Sorting. Merging (cont.)

Merging to ordered sequences. Efficient (Parallel) Sorting. Merging (cont.) Efficient (Paae) Soting One of the most fequent opeations pefomed by computes is oganising (soting) data The access to soted data is moe convenient/faste Thee is a constant need fo good soting agoithms

More information

CSCE 478/878 Lecture 4: Experimental Design and Analysis. Stephen Scott. 3 Building a tree on the training set Introduction. Outline.

CSCE 478/878 Lecture 4: Experimental Design and Analysis. Stephen Scott. 3 Building a tree on the training set Introduction. Outline. In Homewok, you ae (supposedly) Choosing a data set 2 Extacting a test set of size > 3 3 Building a tee on the taining set 4 Testing on the test set 5 Repoting the accuacy (Adapted fom Ethem Alpaydin and

More information

HOW TO TEACH THE FUNDAMENTALS OF INFORMATION SCIENCE, CODING, DECODING AND NUMBER SYSTEMS?

HOW TO TEACH THE FUNDAMENTALS OF INFORMATION SCIENCE, CODING, DECODING AND NUMBER SYSTEMS? 6th INTERNATIONAL MULTIDISCIPLINARY CONFERENCE HOW TO TEACH THE FUNDAMENTALS OF INFORMATION SCIENCE, CODING, DECODING AND NUMBER SYSTEMS? Cecília Sitkuné Göömbei College of Nyíegyháza Hungay Abstact: The

More information

Analysis of Arithmetic. Analysis of Arithmetic. Analysis of Arithmetic Round-Off Errors. Analysis of Arithmetic. Analysis of Arithmetic

Analysis of Arithmetic. Analysis of Arithmetic. Analysis of Arithmetic Round-Off Errors. Analysis of Arithmetic. Analysis of Arithmetic In the fixed-oint imlementation of a digital filte only the esult of the multilication oeation is quantied The eesentation of a actical multilie with the quantie at its outut is shown below u v Q ^v The

More information

Chapter 5 Force and Motion

Chapter 5 Force and Motion Chapte 5 Foce and Motion In Chaptes 2 and 4 we have studied kinematics, i.e., we descibed the motion of objects using paametes such as the position vecto, velocity, and acceleation without any insights

More information

Chapter 5 Force and Motion

Chapter 5 Force and Motion Chapte 5 Foce and Motion In chaptes 2 and 4 we have studied kinematics i.e. descibed the motion of objects using paametes such as the position vecto, velocity and acceleation without any insights as to

More information

ITI Introduction to Computing II

ITI Introduction to Computing II ITI 1121. Intoduction to Computing II Macel Tucotte School of Electical Engineeing and Compute Science Abstact data type: Stack Stack-based algoithms Vesion of Febuay 2, 2013 Abstact These lectue notes

More information

where a = x 10-3 for units of kcal/mol

where a = x 10-3 for units of kcal/mol Detemining the Enegy of Activation Paametes fom Dynamic MR Expeiments: -D. Rich Shoemake (Souce: Dynamic MR Spectoscopy by J. Sandstöm, and me) he esults contained in this document have been published:

More information

Linear Algebra Math 221

Linear Algebra Math 221 Linea Algeba Math Open Book Eam Open Notes Sept Calculatos Pemitted Sho all ok (ecept #). ( pts) Gien the sstem of equations a) ( pts) Epess this sstem as an augmented mati. b) ( pts) Bing this mati to

More information

Gravity. David Barwacz 7778 Thornapple Bayou SE, Grand Rapids, MI David Barwacz 12/03/2003

Gravity. David Barwacz 7778 Thornapple Bayou SE, Grand Rapids, MI David Barwacz 12/03/2003 avity David Bawacz 7778 Thonapple Bayou, and Rapid, MI 495 David Bawacz /3/3 http://membe.titon.net/daveb Uing the concept dicued in the peceding pape ( http://membe.titon.net/daveb ), I will now deive

More information

Stanford University CS259Q: Quantum Computing Handout 8 Luca Trevisan October 18, 2012

Stanford University CS259Q: Quantum Computing Handout 8 Luca Trevisan October 18, 2012 Stanfod Univesity CS59Q: Quantum Computing Handout 8 Luca Tevisan Octobe 8, 0 Lectue 8 In which we use the quantum Fouie tansfom to solve the peiod-finding poblem. The Peiod Finding Poblem Let f : {0,...,

More information

Physics 2001 Problem Set 5 Solutions

Physics 2001 Problem Set 5 Solutions Physics 2001 Poblem Set 5 Solutions Jeff Kissel Octobe 16, 2006 1. A puck attached to a sting undegoes cicula motion on an ai table. If the sting beaks at the point indicated in the figue, which path (A,

More information

SPH3UW/SPH4U Unit 3.2 Forces in Cetripetal Motion Page 1 of 6. Notes Physics Tool Box

SPH3UW/SPH4U Unit 3.2 Forces in Cetripetal Motion Page 1 of 6. Notes Physics Tool Box SPH3UW/SPH4U Unit 3. Foce in Cetipetal Motion Page 1 o 6 Note Phyic Tool Box Net Foce: acting on an object in uniom cicula motion act towad the cente o the cicle. Magnitude o Net Foce: combine Newton Second

More information

Theory. Single Soil Layer. ProShake User s Manual

Theory. Single Soil Layer. ProShake User s Manual PoShake Ue Manual Theoy PoShake ue a fequency domain appoach to olve the gound epone poblem. In imple tem, the input motion i epeented a the um of a eie of ine wave of diffeent amplitude, fequencie, and

More information

1) Consider an object of a parabolic shape with rotational symmetry z

1) Consider an object of a parabolic shape with rotational symmetry z Umeå Univesitet, Fysik 1 Vitaly Bychkov Pov i teknisk fysik, Fluid Mechanics (Stömningsläa), 01-06-01, kl 9.00-15.00 jälpmedel: Students may use any book including the tetbook Lectues on Fluid Dynamics.

More information

A STUDY OF HAMMING CODES AS ERROR CORRECTING CODES

A STUDY OF HAMMING CODES AS ERROR CORRECTING CODES AGU Intenational Jounal of Science and Technology A STUDY OF HAMMING CODES AS ERROR CORRECTING CODES Ritu Ahuja Depatment of Mathematics Khalsa College fo Women, Civil Lines, Ludhiana-141001, Punjab, (India)

More information

FI 2201 Electromagnetism

FI 2201 Electromagnetism FI Electomagnetim Aleande A. Ikanda, Ph.D. Phyic of Magnetim and Photonic Reeach Goup ecto Analyi CURILINEAR COORDINAES, DIRAC DELA FUNCION AND HEORY OF ECOR FIELDS Cuvilinea Coodinate Sytem Cateian coodinate:

More information

Precision Spectrophotometry

Precision Spectrophotometry Peciion Spectophotomety Pupoe The pinciple of peciion pectophotomety ae illutated in thi expeiment by the detemination of chomium (III). ppaatu Spectophotomete (B&L Spec 20 D) Cuvette (minimum 2) Pipet:

More information

PHYS Summer Professor Caillault Homework Solutions. Chapter 5

PHYS Summer Professor Caillault Homework Solutions. Chapter 5 PHYS 1111 - Summe 2007 - Pofesso Caillault Homewok Solutions Chapte 5 7. Pictue the Poblem: The ball is acceleated hoizontally fom est to 98 mi/h ove a distance of 1.7 m. Stategy: Use equation 2-12 to

More information

MULTILAYER PERCEPTRONS

MULTILAYER PERCEPTRONS Last updated: Nov 26, 2012 MULTILAYER PERCEPTRONS Outline 2 Combining Linea Classifies Leaning Paametes Outline 3 Combining Linea Classifies Leaning Paametes Implementing Logical Relations 4 AND and OR

More information

Elementary Statistics and Inference. Elementary Statistics and Inference. 11. Regression (cont.) 22S:025 or 7P:025. Lecture 14.

Elementary Statistics and Inference. Elementary Statistics and Inference. 11. Regression (cont.) 22S:025 or 7P:025. Lecture 14. Elementay tatistics and Infeence :05 o 7P:05 Lectue 14 1 Elementay tatistics and Infeence :05 o 7P:05 Chapte 10 (cont.) D. Two Regession Lines uppose two vaiables, and ae obtained on 100 students, with

More information

15.081J/6.251J Introduction to Mathematical Programming. Lecture 6: The Simplex Method II

15.081J/6.251J Introduction to Mathematical Programming. Lecture 6: The Simplex Method II 15081J/6251J Intoduction to Mathematical Pogamming ectue 6: The Simplex Method II 1 Outline Revised Simplex method Slide 1 The full tableau implementation Anticycling 2 Revised Simplex Initial data: A,

More information

1 Explicit Explore or Exploit (E 3 ) Algorithm

1 Explicit Explore or Exploit (E 3 ) Algorithm 2.997 Decision-Making in Lage-Scale Systems Mach 3 MIT, Sping 2004 Handout #2 Lectue Note 9 Explicit Exploe o Exploit (E 3 ) Algoithm Last lectue, we studied the Q-leaning algoithm: [ ] Q t+ (x t, a t

More information

The Millikan Experiment: Determining the Elementary Charge

The Millikan Experiment: Determining the Elementary Charge LAB EXERCISE 7.5.1 7.5 The Elementay Chage (p. 374) Can you think of a method that could be used to suggest that an elementay chage exists? Figue 1 Robet Millikan (1868 1953) m + q V b The Millikan Expeiment:

More information

Objects usually are charged up through the transfer of electrons from one object to the other.

Objects usually are charged up through the transfer of electrons from one object to the other. 1 Pat 1: Electic Foce 1.1: Review of Vectos Review you vectos! You should know how to convet fom pola fom to component fom and vice vesa add and subtact vectos multiply vectos by scalas Find the esultant

More information

Considerations Regarding the Flux Estimation in Induction Generator with Application at the Control of Unconventional Energetic Conversion Systems

Considerations Regarding the Flux Estimation in Induction Generator with Application at the Control of Unconventional Energetic Conversion Systems Conideation Regading the Flux Etimation in Induction Geneato with Application at the Contol of Unconventional Enegetic Conveion Sytem Ioif Szeidet, Octavian Potean, Ioan Filip, Vaa Citian Depatment of

More information

COMP303 Computer Architecture Lecture 11. An Overview of Pipelining

COMP303 Computer Architecture Lecture 11. An Overview of Pipelining COMP303 Compute Achitectue Lectue 11 An Oveview of Pipelining Pipelining Pipelining povides a method fo executing multiple instuctions at the same time. Laundy Example: Ann, Bian, Cathy, Dave each have

More information

THE STUDY AND IMPROVEMENT OF X-BAND ATTENUATION ESTIMATION USING DUAL POLARIMETRIC RADAR TECHNIQUES

THE STUDY AND IMPROVEMENT OF X-BAND ATTENUATION ESTIMATION USING DUAL POLARIMETRIC RADAR TECHNIQUES THE STUDY AND IMPROVEMENT OF X-BAND ATTENUATION ESTIMATION USING DUAL POLARIMETRIC RADAR TECHNIQUES Yuxiang Liu Advisos: V.N. Bingi and V. Chandaseka Decembe 4, 006 T ct } ) ( ) ( 4 ln exp{ ), ( 1 0 1

More information

Goodness-of-fit for composite hypotheses.

Goodness-of-fit for composite hypotheses. Section 11 Goodness-of-fit fo composite hypotheses. Example. Let us conside a Matlab example. Let us geneate 50 obsevations fom N(1, 2): X=nomnd(1,2,50,1); Then, unning a chi-squaed goodness-of-fit test

More information

A Bijective Approach to the Permutational Power of a Priority Queue

A Bijective Approach to the Permutational Power of a Priority Queue A Bijective Appoach to the Pemutational Powe of a Pioity Queue Ia M. Gessel Kuang-Yeh Wang Depatment of Mathematics Bandeis Univesity Waltham, MA 02254-9110 Abstact A pioity queue tansfoms an input pemutation

More information

Σr2=0. Σ Br. Σ br. Σ r=0. br = Σ. Σa r-s b s (1.2) s=0. Σa r-s b s-t c t (1.3) t=0. cr = Σ. dr = Σ. Σa r-s b s-t c t-u d u (1.4) u =0.

Σr2=0. Σ Br. Σ br. Σ r=0. br = Σ. Σa r-s b s (1.2) s=0. Σa r-s b s-t c t (1.3) t=0. cr = Σ. dr = Σ. Σa r-s b s-t c t-u d u (1.4) u =0. 0 Powe of Infinite Seie. Multiple Cauchy Poduct The multinomial theoem i uele fo the powe calculation of infinite eie. Thi i becaue the polynomial theoem depend on the numbe of tem, o it can not be applied

More information

Force and Work: Reminder

Force and Work: Reminder Electic Potential Foce and Wok: Reminde Displacement d a: initial point b: final point Reminde fom Mechanics: Foce F if thee is a foce acting on an object (e.g. electic foce), this foce may do some wok

More information

Light Time Delay and Apparent Position

Light Time Delay and Apparent Position Light Time Delay and ppaent Position nalytical Gaphics, Inc. www.agi.com info@agi.com 610.981.8000 800.220.4785 Contents Intoduction... 3 Computing Light Time Delay... 3 Tansmission fom to... 4 Reception

More information

Centripetal Force OBJECTIVE INTRODUCTION APPARATUS THEORY

Centripetal Force OBJECTIVE INTRODUCTION APPARATUS THEORY Centipetal Foce OBJECTIVE To veify that a mass moving in cicula motion expeiences a foce diected towad the cente of its cicula path. To detemine how the mass, velocity, and adius affect a paticle's centipetal

More information

Multiple Experts with Binary Features

Multiple Experts with Binary Features Multiple Expets with Binay Featues Ye Jin & Lingen Zhang Decembe 9, 2010 1 Intoduction Ou intuition fo the poect comes fom the pape Supevised Leaning fom Multiple Expets: Whom to tust when eveyone lies

More information

History of Astronomy - Part II. Tycho Brahe - An Observer. Johannes Kepler - A Theorist

History of Astronomy - Part II. Tycho Brahe - An Observer. Johannes Kepler - A Theorist Histoy of Astonomy - Pat II Afte the Copenican Revolution, astonomes stived fo moe obsevations to help bette explain the univese aound them Duing this time (600-750) many majo advances in science and astonomy

More information

Towards Fast, Accurate and Reproducible LU Factorization

Towards Fast, Accurate and Reproducible LU Factorization Towards Fast, Accurate and Reproducible LU Factorization Roman Iakymchuk 1, David Defour 2, and Stef Graillat 3 1 KTH Royal Institute of Technology, CSC, CST/PDC 2 Université de Perpignan, DALI LIRMM 3

More information

Conjugate Gradient Methods. Michael Bader. Summer term 2012

Conjugate Gradient Methods. Michael Bader. Summer term 2012 Gadient Methods Outlines Pat I: Quadatic Foms and Steepest Descent Pat II: Gadients Pat III: Summe tem 2012 Pat I: Quadatic Foms and Steepest Descent Outlines Pat I: Quadatic Foms and Steepest Descent

More information

Tuning And Understanding MILC Performance In Cray XK6 GPU Clusters. Mike Showerman, Guochun Shi Steven Gottlieb

Tuning And Understanding MILC Performance In Cray XK6 GPU Clusters. Mike Showerman, Guochun Shi Steven Gottlieb Tuning And Understanding MILC Performance In Cray XK6 GPU Clusters Mike Showerman, Guochun Shi Steven Gottlieb Outline Background Lattice QCD and MILC GPU and Cray XK6 node architecture Implementation

More information

A Simple Model of Communication APIs Application to Dynamic Partial-order Reduction

A Simple Model of Communication APIs Application to Dynamic Partial-order Reduction Simple Model of Communication PIs pplication to Dynamic Patial-ode Reduction Cistian Rosa Stephan Mez Matin Quinson VOCS 2010 22/09/2010 1 / 18 Motivation Distibuted lgoithms ae had to get ight: lack of

More information

Multi-Objective Optimization Algorithms for Finite Element Model Updating

Multi-Objective Optimization Algorithms for Finite Element Model Updating Multi-Objective Optimization Algoithms fo Finite Element Model Updating E. Ntotsios, C. Papadimitiou Univesity of Thessaly Geece Outline STRUCTURAL IDENTIFICATION USING MEASURED MODAL DATA Weighted Modal

More information

Theorem 2: Proof: Note 1: Proof: Note 2:

Theorem 2: Proof: Note 1: Proof: Note 2: A New 3-Dimenional Polynomial Intepolation Method: An Algoithmic Appoach Amitava Chattejee* and Rupak Bhattachayya** A new 3-dimenional intepolation method i intoduced in thi pape. Coeponding to the method

More information

Transverse Wakefield in a Dielectric Tube with Frequency Dependent Dielectric Constant

Transverse Wakefield in a Dielectric Tube with Frequency Dependent Dielectric Constant ARDB-378 Bob Siemann & Alex Chao /4/5 Page of 8 Tansvese Wakefield in a Dielectic Tube with Fequency Dependent Dielectic Constant This note is a continuation of ARDB-368 that is now extended to the tansvese

More information

Online-routing on the butterfly network: probabilistic analysis

Online-routing on the butterfly network: probabilistic analysis Online-outing on the buttefly netwok: obabilistic analysis Andey Gubichev 19.09.008 Contents 1 Intoduction: definitions 1 Aveage case behavio of the geedy algoithm 3.1 Bounds on congestion................................

More information

Improvement in Accuracy for Design of Multidielectric Layers Microstrip Patch Antenna

Improvement in Accuracy for Design of Multidielectric Layers Microstrip Patch Antenna 498 Impovement in Accuacy fo Design of Multidielectic Layes Micostip Patch Antenna Sami Dev Gupta*, Anvesh Gag and Anuag P. Saan Jaypee Institute of Infomation Technology Univesity Noida, Utta Padesh,

More information

Probablistically Checkable Proofs

Probablistically Checkable Proofs Lectue 12 Pobablistically Checkable Poofs May 13, 2004 Lectue: Paul Beame Notes: Chis Re 12.1 Pobablisitically Checkable Poofs Oveview We know that IP = PSPACE. This means thee is an inteactive potocol

More information

Welcome to Physics 272

Welcome to Physics 272 Welcome to Physics 7 Bob Mose mose@phys.hawaii.edu http://www.phys.hawaii.edu/~mose/physics7.html To do: Sign into Masteing Physics phys-7 webpage Registe i-clickes (you i-clicke ID to you name on class-list)

More information

Math 1525 Excel Lab 3 Exponential and Logarithmic Functions Spring, 2001

Math 1525 Excel Lab 3 Exponential and Logarithmic Functions Spring, 2001 Math 1525 Excel Lab 3 Exponential and Logaithmic Functions 1 Math 1525 Excel Lab 3 Exponential and Logaithmic Functions Sping, 21 Goal: The goals of Lab 3 ae to illustate exponential, logaithmic, and logistic

More information

Equilibrium Reconstruction with Grad-Shafranov Equation

Equilibrium Reconstruction with Grad-Shafranov Equation Equilibium Reconstuction with Gad-Shafanov Equation Qian Peng (qp215) Apil 14, 211 1 Gad-Shafanov equation [1]Fo axial symmetic case, the tooidal B eld is B to = F φ. Whee 2π F (, θ) is the total poloidal

More information

, and the curve BC is symmetrical. Find also the horizontal force in x-direction on one side of the body. h C

, and the curve BC is symmetrical. Find also the horizontal force in x-direction on one side of the body. h C Umeå Univesitet, Fysik 1 Vitaly Bychkov Pov i teknisk fysik, Fluid Dynamics (Stömningsläa), 2013-05-31, kl 9.00-15.00 jälpmedel: Students may use any book including the textbook Lectues on Fluid Dynamics.

More information

Pearson s Chi-Square Test Modifications for Comparison of Unweighted and Weighted Histograms and Two Weighted Histograms

Pearson s Chi-Square Test Modifications for Comparison of Unweighted and Weighted Histograms and Two Weighted Histograms Peason s Chi-Squae Test Modifications fo Compaison of Unweighted and Weighted Histogams and Two Weighted Histogams Univesity of Akueyi, Bogi, v/noduslód, IS-6 Akueyi, Iceland E-mail: nikolai@unak.is Two

More information

Universal Gravitation

Universal Gravitation Chapte 1 Univesal Gavitation Pactice Poblem Solutions Student Textbook page 580 1. Conceptualize the Poblem - The law of univesal gavitation applies to this poblem. The gavitational foce, F g, between

More information

ASTR415: Problem Set #6

ASTR415: Problem Set #6 ASTR45: Poblem Set #6 Cuan D. Muhlbege Univesity of Mayland (Dated: May 7, 27) Using existing implementations of the leapfog and Runge-Kutta methods fo solving coupled odinay diffeential equations, seveal

More information

High precision computer simulation of cyclotrons KARAMYSHEVA T., AMIRKHANOV I. MALININ V., POPOV D.

High precision computer simulation of cyclotrons KARAMYSHEVA T., AMIRKHANOV I. MALININ V., POPOV D. High pecision compute simulation of cyclotons KARAMYSHEVA T., AMIRKHANOV I. MALININ V., POPOV D. Abstact Effective and accuate compute simulations ae highly impotant in acceleatos design and poduction.

More information

Semi-Custom VLSI Design and Implementation of a New Efficient RNS Division Algorithm

Semi-Custom VLSI Design and Implementation of a New Efficient RNS Division Algorithm Semi-Custom VLSI Design and Implementation of a New Efficient RNS Division Algoithm AHMAD A. HIASAT AND HODA ABDEL-AT-ZOHD 2 Elect. Eng. Dept., Pincess Sumaya Univesity, PO Box 438, Amman 94, Jodan 2 Elect.

More information

LC transfer of energy between the driving source and the circuit will be a maximum.

LC transfer of energy between the driving source and the circuit will be a maximum. The Q of oscillatos efeences: L.. Fotney Pinciples of Electonics: Analog and Digital, Hacout Bace Jovanovich 987, Chapte (AC Cicuits) H. J. Pain The Physics of Vibations and Waves, 5 th edition, Wiley

More information

Phases of Matter. Since liquids and gases are able to flow, they are called fluids. Compressible? Able to Flow? shape?

Phases of Matter. Since liquids and gases are able to flow, they are called fluids. Compressible? Able to Flow? shape? Fluids Chapte 3 Lectue Sequence. Pessue (Sections -3). Mechanical Popeties (Sections 5, and 7) 3. Gauge Pessue (Sections 4, and 6) 4. Moving Fluids (Sections 8-0) Pessue Phases of Matte Phase Retains its

More information

C/CS/Phys C191 Shor s order (period) finding algorithm and factoring 11/12/14 Fall 2014 Lecture 22

C/CS/Phys C191 Shor s order (period) finding algorithm and factoring 11/12/14 Fall 2014 Lecture 22 C/CS/Phys C9 Sho s ode (peiod) finding algoithm and factoing /2/4 Fall 204 Lectue 22 With a fast algoithm fo the uantum Fouie Tansfom in hand, it is clea that many useful applications should be possible.

More information

Image Enhancement: Histogram-based methods

Image Enhancement: Histogram-based methods Image Enhancement: Hitogam-baed method The hitogam of a digital image with gayvalue, i the dicete function,, L n n # ixel with value Total # ixel image The function eeent the faction of the total numbe

More information

Markscheme May 2017 Calculus Higher level Paper 3

Markscheme May 2017 Calculus Higher level Paper 3 M7/5/MATHL/HP3/ENG/TZ0/SE/M Makscheme May 07 Calculus Highe level Pape 3 pages M7/5/MATHL/HP3/ENG/TZ0/SE/M This makscheme is the popety of the Intenational Baccalaueate and must not be epoduced o distibuted

More information

270 Int'l Conf. Par. and Dist. Proc. Tech. and Appl. PDPTA'17

270 Int'l Conf. Par. and Dist. Proc. Tech. and Appl. PDPTA'17 270 Int'l Conf. Pa. and Dist. Poc. Tech. and Appl. PDPTA'17 Pefomance Evaluation of a Combination of the Paallel Bisection Method and the Block Invese Iteation Method with Reothogonalization fo Eigenvalue

More information

Physics 4A Chapter 8: Dynamics II Motion in a Plane

Physics 4A Chapter 8: Dynamics II Motion in a Plane Physics 4A Chapte 8: Dynamics II Motion in a Plane Conceptual Questions and Example Poblems fom Chapte 8 Conceptual Question 8.5 The figue below shows two balls of equal mass moving in vetical cicles.

More information

Analysis of Finite Word-Length Effects

Analysis of Finite Word-Length Effects T-6.46 Digital Signal Pocessing and Filteing 8.9.4 Intoduction Analysis of Finite Wod-Length Effects Finite wodlength effects ae caused by: Quantization of the filte coefficients ounding / tuncation of

More information

Boise State University Department of Electrical and Computer Engineering ECE470 Electric Machines

Boise State University Department of Electrical and Computer Engineering ECE470 Electric Machines Boie State Univeity Depatment of Electical and Compute Engineeing ECE470 Electic Machine Deivation of the Pe-Phae Steady-State Equivalent Cicuit of a hee-phae Induction Machine Nomenclatue θ: oto haft

More information

Current Balance Warm Up

Current Balance Warm Up PHYSICS EXPERIMENTS 133 Cuent Balance-1 Cuent Balance Wam Up 1. Foce between cuent-caying wies Wie 1 has a length L (whee L is "long") and caies a cuent I 0. What is the magnitude of the magnetic field

More information

Lecture 8 - Gauss s Law

Lecture 8 - Gauss s Law Lectue 8 - Gauss s Law A Puzzle... Example Calculate the potential enegy, pe ion, fo an infinite 1D ionic cystal with sepaation a; that is, a ow of equally spaced chages of magnitude e and altenating sign.

More information

Coarse Mesh Radiation Transport Code COMET Radiation Therapy Application*

Coarse Mesh Radiation Transport Code COMET Radiation Therapy Application* Coase Mesh Radiation Tanspot Code COMT Radiation Theapy Application* Fazad Rahnema Nuclea & Radiological ngineeing and Medical Physics Pogams Geogia Institute of Technology Computational Medical Physics

More information

Exploration of the three-person duel

Exploration of the three-person duel Exploation of the thee-peson duel Andy Paish 15 August 2006 1 The duel Pictue a duel: two shootes facing one anothe, taking tuns fiing at one anothe, each with a fixed pobability of hitting his opponent.

More information

Analysis of the Mean Absolute Error (MAE) and the Root Mean Square Error (RMSE) in Assessing Rounding Model

Analysis of the Mean Absolute Error (MAE) and the Root Mean Square Error (RMSE) in Assessing Rounding Model IOP Confeence Seies: Mateials Science and Engineeing PAPER OPEN ACCESS Analysis of the Mean Absolute Eo () and the Root Mean Squae Eo (RMSE) in Assessing Rounding Model To cite this aticle: Weijie Wang

More information

The Substring Search Problem

The Substring Search Problem The Substing Seach Poblem One algoithm which is used in a vaiety of applications is the family of substing seach algoithms. These algoithms allow a use to detemine if, given two chaacte stings, one is

More information

SEARCH-BASED DYNAMIC IDENTIFICATION OF INDUCTION MOTORS

SEARCH-BASED DYNAMIC IDENTIFICATION OF INDUCTION MOTORS SEARCH-BASED DYNAMIC IDENTIFICATION OF INDUCTION MOTORS Alexande Vladimiovich Nesteovskiy Veniamin Geogievich ashiskikh Valey Mihailovich Zavyalov and Iina Yuyevna Semykina uzbass State Technical Univesity

More information

LINEAR AND NONLINEAR ANALYSES OF A WIND-TUNNEL BALANCE

LINEAR AND NONLINEAR ANALYSES OF A WIND-TUNNEL BALANCE LINEAR AND NONLINEAR ANALYSES O A WIND-TUNNEL INTRODUCTION BALANCE R. Kakehabadi and R. D. Rhew NASA LaRC, Hampton, VA The NASA Langley Reseach Cente (LaRC) has been designing stain-gauge balances fo utilization

More information

Physics 201 Lecture 18

Physics 201 Lecture 18 Phsics 0 ectue 8 ectue 8 Goals: Define and anale toque ntoduce the coss poduct Relate otational dnamics to toque Discuss wok and wok eneg theoem with espect to otational motion Specif olling motion (cente

More information

The Laws of Motion ( ) N SOLUTIONS TO PROBLEMS ! F = ( 6.00) 2 + ( 15.0) 2 N = 16.2 N. Section 4.4. Newton s Second Law The Particle Under a Net Force

The Laws of Motion ( ) N SOLUTIONS TO PROBLEMS ! F = ( 6.00) 2 + ( 15.0) 2 N = 16.2 N. Section 4.4. Newton s Second Law The Particle Under a Net Force SOLUTIONS TO PROBLEMS The Laws of Motion Section 4.3 Mass P4. Since the ca is moving with constant speed and in a staight line, the esultant foce on it must be zeo egadless of whethe it is moving (a) towad

More information