Unformatted text preview:

COMP 206: Computer Architecture and ImplementationOutlineQuantitative Principles of Computer DesignComparisonAmdahl’s Law (1967)Amdahl’s LawExample of Amdahl’s LawAmdahl’s Law (HP3 book, pp. 40-41)Implications of Amdahl’s Law“Make The Common Case Fast”“Make The Common Case Fast” (2)“Make The Common Case Fast” (3a)“Make The Common Case Fast” (3b)Basics of PerformanceDetails of CPIMIPSMFLOP/sAspects of CPU Performance1COMP 206:COMP 206:Computer Architecture and Computer Architecture and ImplementationImplementationMontek SinghMontek SinghMon., Aug 30, 2004Mon., Aug 30, 2004Lecture 2Lecture 22OutlineOutlineQuantitative Principles of Computer DesignQuantitative Principles of Computer DesignAmdahl’s law (make the common case fast)Amdahl’s law (make the common case fast)Performance MetricsPerformance MetricsMIPS, FLOPS, and all that…MIPS, FLOPS, and all that…ExamplesExamples3Quantitative Principles of Computer Quantitative Principles of Computer DesignDesignT1P Execution timeResponse timeLatencyExecution timeResponse timeLatencyPerformanceRate of producing resultsThroughputBandwidthPerformanceRate of producing resultsThroughputBandwidthbitn / instructio / programresult / work / timetimebits / nsinstructio / program / resultswork /4ComparisonComparisonYX 0 nnXY 1  nnXY 0 1001“Y is n times larger than X”“Y is n% larger than X”5“Validity of the single processor approach to achieving large scale computing capabilities”, G. M. Amdahl, AFIPS Conference Proceedings, pp. 483-485, April 1967Amdahl’s Law (1967)Amdahl’s Law (1967)Historical contextHistorical contextAmdahl was demonstrating “the continued validity of the Amdahl was demonstrating “the continued validity of the single processor approach and of the weaknesses of the single processor approach and of the weaknesses of the multiple processor approach”multiple processor approach”Paper contains no mathematical formulation, just Paper contains no mathematical formulation, just arguments and simulationarguments and simulation““The nature of this overhead appears to be sequential so that it The nature of this overhead appears to be sequential so that it is unlikely to be amenable to parallel processing techniques.”is unlikely to be amenable to parallel processing techniques.”““A fairly obvious conclusion which can be drawn at this point is A fairly obvious conclusion which can be drawn at this point is that the effort expended on achieving high parallel performance that the effort expended on achieving high parallel performance rates is wasted unless it is accompanied by achievements in rates is wasted unless it is accompanied by achievements in sequential processing rates of very nearly the same sequential processing rates of very nearly the same magnitude.”magnitude.”Nevertheless, it is of widespread applicability in all Nevertheless, it is of widespread applicability in all kinds of situationskinds of situations6Amdahl’s LawAmdahl’s Law11iiiiiavgFRFR  secondresults11secondresultsFraction of resultsgenerated at this rateAverage execution rate(performance)Weighted harmonic meanNote: Not “fractionof time spent workingat this rate”Note: Not “fractionof time spent workingat this rate”“Bottleneckology: Evaluating Supercomputers”, Jack Worlton, COMPCOM 85, pp. 405-4067Example of Amdahl’s LawExample of Amdahl’s Law30% of results are generated at the rate of 1 MFLOPS,20% at 10 MFLOPS,50% at 100 MFLOPS.What is the average performance?What is the bottleneck?30% of results are generated at the rate of 1 MFLOPS,20% at 10 MFLOPS,50% at 100 MFLOPS.What is the average performance?What is the bottleneck?MFLOPS 08.35.321005.02301001005.0102.013.01Ravg%5.15.325.0%,2.65.322%,3.925.3230Bottleneck: the rate that consumes most of the time0 0.2 0.4 0.6 0.8 18Amdahl’s Law (HP3 book, pp. 40-Amdahl’s Law (HP3 book, pp. 40-41)41) obtainedactually speedup speedup attainable maximum rate slowat produced results offraction ratefast at produced results offraction rate old) d,(unenhance slow rate new) (enhanced,fast RRRRFFRRsavgsfsfsf RRFFRRRRFFRRFRFRsfffsavgfsfssffssavg111FractionenhancedSpeedupenhancedSpeedupoverallSpeedupoverallSpeedupenhancedFractionenhancedFractionenhanced9Implications of Amdahl’s LawImplications of Amdahl’s LawThe performance improvements provided by a feature The performance improvements provided by a feature are limited by how often that feature is usedare limited by how often that feature is usedAs stated, Amdahl’s Law is valid only if the system As stated, Amdahl’s Law is valid only if the system always works with exactly one of the ratesalways works with exactly one of the ratesIf a non-blocking cache is used, or there is overlap between If a non-blocking cache is used, or there is overlap between CPU and I/O operations, Amdahl’s Law as given here is not CPU and I/O operations, Amdahl’s Law as given here is not applicableapplicableBottleneck is the most promising target for Bottleneck is the most promising target for improvementsimprovements““Make the common case fast”Make the common case fast”Infrequent events, even if they consume a lot of time, will Infrequent events, even if they consume a lot of time, will make little difference to performancemake little difference to performanceTypical use: Change only one parameter of system, and Typical use: Change only one parameter of system, and compute effect of this changecompute effect of this changeThe same program, with the same input data, should run on The same program, with the same input data, should run on the machine in both casesthe machine in both cases10““Make The Common Case Fast”Make The Common Case Fast”All instructions require an instruction fetch, All instructions require an instruction fetch, only a fraction require a data fetch/storeonly a fraction require a data fetch/storeOptimize instruction access over data accessOptimize instruction access over data accessPrograms exhibit localityPrograms exhibit localitySpatial LocalitySpatial Localityitems with addresses near one another tend to be items with addresses near one another tend to be referenced close


View Full Document

UNC-Chapel Hill COMP 206 - Lecture 2 Notes

Download Lecture 2 Notes
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Lecture 2 Notes and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Lecture 2 Notes 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?