Unformatted text preview:

CMSC 611: AdvancedCMSC 611: AdvancedComputer ArchitectureComputer ArchitectureCost & PerformanceCost & PerformanceSome material adapted from Mohamed Younis, UMBC CMSC 611 Spr 2003 course slidesSome material adapted from Hennessy & Patterson / © 2003 Elsevier ScienceWhat Affects Cost?What Affects Cost?1. Learning curve:– The more experience in manufacturing a component, thebetter the yield– In general, a chip, board or system with twice the yield willhave half the cost.– The learning curve is different for different components,complicating design decisions2. Volume– Larger volume increases rate of learning curve– Doubling the volume typically reduce cost by 10%3. Commodities– Are essentially identical products sold by multiple vendorsin large volumes– Foil the competition and drive the efficiency higher and thusthe cost down$/DRAM chipA dollar in 1977 = $2.95 in 2001Cost/MB = $500 in 1997 = $0.35 in 2000 = $0.08 in 2001Demand exceededsupply ! price slowdropEach generation drops in price by a factor of 10 to 30 over its lifetimeCost Trends for DRAMCost Trends for DRAMIntel List price for 1000 units of the Pentium IIIPrice drop due yield enhancementsCost Trends for ProcessorsCost Trends for ProcessorsComponent CostDirect Cost Gross Margin Average DiscountAvg. Selling PriceList Price15% to 33% 6% to 8%34% to 39%25% to 40%Cost vs. PriceCost vs. Price• Component Costs: raw material cost for the system’s buildingblocks• Direct Costs (add 25% to 40%) recurring costs: labor,purchasing, scrap, warranty• Gross Margin (add 82% to 186%) nonrecurring costs:R&D, marketing, sales, equipment maintenance, rental, financingcost, pretax profits, taxes• Average Discount to get List Price (add 33% to 66%): volumediscounts and/or retailer markupChip Area (mm2) Total Cost Price Comment 386DX 43 $9 $31 486DX2 81 $35 $245 NNoo CCoommppeettiittiioonn PowerPC 601 121 $77 $280 DEC Alpha 234 $202 $1231 Recoup R&D? Pentium 296 $473 $965 Chip Prices (August 1993) for a volume of 10,000 unitsExample: Price vs. CostExample: Price vs. CostThe Role of PerformanceThe Role of Performance• Hardware performance is a key to the effectiveness ofthe entire system• Performance has to be measured and compared toevaluate various design and technological approaches• To optimize the performance, major affecting factorshave to be known• For different types of applications, differentperformance metrics may be appropriate and differentaspects of a computer system may be most significant• Instructions use and implementation, memoryhierarchy and I/O handling are among the factors thataffect the performanceCriteria of performance evaluation differs among users anddesignersDefining PerformanceDefining Performance• Performance means different things to differentpeople, therefore its assessment is subtleAnalogy from the airlines industry:• How to measure performance for an airplane?– Cruising speed (How fast it gets to the destination)– Flight range (How far it can reach)– Passenger capacity (How many passengers it can carry)AirplanePassengercapacityCruising range(miles)Cruising speed(m.p.h)Passenger throughput(Passenger ! m.p.h)Boeing 777 375 4630 610 228,750Boeing 747 470 4150 610 286,700BAC/Sud Concorde 132 4000 1350 178,200Douglas DC-8-50 146 8720 544 79,424Decreasing response time always improves throughputPerformance MetricsPerformance Metrics• Response (execution) time:– The time between the start and the completion of a task– Measures user perception of the system speed– Common in reactive and time critical systems, single-user computer,etc.• Throughput:– The total number of tasks done in a given time– Most relevant to batch processing (billing, credit card processing)– Mainly used for input/output systems (disk access, printer, etc.)• Example– Faster processor! Enhances both response time and throughput– Adding additional processors to parallelize separate tasks! Enhances only throughput! Performance =1Execution timeResponse-time MetricResponse-time Metric• Maximizing performance meansminimizing response (execution) time! Performance =1Execution time ! Performance (P1) > Performance (P2) w.r.t L " Execution time (P1,L) < Execution time (P2,L)Response-time MetricResponse-time Metric• Performance of Processor P1 is betterthan P2 if– For a given work load L– P1 takes less time to execute L than P2! Performance =1Execution time ! CPU Performance (P2)CPU Performance (P1)=Total execution time (P1)Total execution time (P2)Response-time MetricResponse-time Metric• Relative performance captures theperformance ratio– For the same work loadrate Clockprogram a for cycles clock CPU time cycle Clockprogram a for cycles clock CPUprogram a for time execution CPU=!=DesignerDesigner’’s Performances PerformanceMetricsMetrics• Users and designers measure performanceusing different metrics– Users: quotable metrics (GHz)– Designers: program execution• Designer focuses on reducing the clock cycletime and the number of cycles per program• Many techniques to decrease the number ofclock cycles also increase the clock cycle timeor the average number of cycles perinstruction (CPI)A program runs in 10 seconds on a computer “A” with a 400 MHz clock.We desire a faster computer “B” that could run the program in 6 seconds.The designer has determined that a substantial increase in the clock speed ispossible, however it would cause computer “B” to require 1.2 times as many clockcycles as computer “A”. What should be the clock rate of computer “B”?(A) rate Clockcycles clock CPU (A) time CPU = ! 10 seconds =CPU clock cycles of program400 " 106 cycles/second ! CPU clock cycles of program = 10 seconds " 400 " 106 cycles/second = 4000 " 106 cycles ! 6 seconds =1.2 " CPU clock cycles of programclock rate (B)=1.2 " 4000 " 106 cyclesclock rate (B)ondcycles/sec 10 800second 6cycles 10 4000 1.2 (B) rate clock66!=!!=To get the clock rate of the faster computer, we use the same formulaExampleExampleCPU time = Instruction count " CPI " Clock cycle timeOrrate ClockCPIcount nInstructiotime CPU!=cycle ClockSecondsnInstructio cycles ClockProgramnsInstructiotime CPU !!=Component of performance Units of measureCPU execution time for a program Seconds for the programInstruction count Instructions executed for the programClock cycles per instructions (CPI) Average number of clock cycles/instructionClock cycle


View Full Document

UMBC CMSC 611 - Cost & Performance

Download Cost & Performance
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 Cost & Performance 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 Cost & Performance 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?