1 / 20

Cost and Performance

Cost and Performance. Previous lecture. Goal. Understand Engineering methodology Design techniques Correctness criteria Evaluation methods Technology trends involved in the design of computer systems. Cost Components. Chip Cost. Chip cost is primarily a function of die area

nicholas-park
Download Presentation

Cost and Performance

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Cost and Performance

  2. Previous lecture Goal • Understand • Engineering methodology • Design techniques • Correctness criteria • Evaluation methods • Technology trends involved in the design of computer systems

  3. Cost Components

  4. Chip Cost • Chip cost is primarily a function of die area • increases much faster than linearly due to yield • going larger gives diminishing performance returns

  5. Chip Cost chip cost = Die cost + Testing cost + Packaging cost Final test yield Die cost = Wafer cost Dies per Wafer * Die yield Die Cost goes roughly with die area4

  6. Real World Examples Chip Metal Line Wafer Defect Area Dies/ Yield Die Cost layers width cost /cm2 mm2 wafer 386DX 2 0.90 $900 1.0 43 360 71% $4 486DX2 3 0.80 $1200 1.0 81 181 54% $12 PowerPC 601 4 0.80 $1700 1.3 121 115 28% $53 HP PA 7100 3 0.80 $1300 1.0 196 66 27% $73 DEC Alpha 3 0.70 $1500 1.2 234 53 19% $149 SuperSPARC 3 0.70 $1700 1.6 256 48 13% $272 Pentium 3 0.80 $1500 1.5 296 40 9% $417 • From “Estimating IC Manufacturing Costs”by Linley Gwennap, Microprocessor Report, August 2, 1993, p. 15

  7. What is Relationship of Cost to Price? • Component Costs • Direct Costs(recurring costs): labor, purchasing, scrap, warranty • Gross Margin (nonrecurring costs): R&D, marketing, sales, equipment maintenance, rental, financing cost, pretax profits, taxes • Average Discount: volume discounts and/or retailer markup

  8. Price vs. Cost Figures 1.7 and 1.8

  9. Seoul to Pusan Speed Passengers Throughput 10 hours 100 km/h 5 500 1 hour 1000km/h 100 100,000 Performance Sonata Boeing 727 • Time to run the task • Execution time, response time, latency • Tasks per day, hour, week, sec, ns … • Throughput, bandwidth

  10. Performance and Execution Time Execution time and performance are reciprocals Execution Time(Y) Performance(X) ---------------- = --------------- Execution Time(X) Performance(Y)

  11. Performance Terminology “X is n% faster than Y” means: Execution Time(Y) Performance(X) n ----------------- = -------------- = 1 + ----- Execution Time(X) Performance(Y) 100 n = 100(Performance(X) - Performance(Y)) Performance(Y) n = 100(Execution Time(Y) - Execution Time(X)) Execution Time(X) Example: Y takes 15 seconds to complete a task, X takes 10 seconds. What % faster is X?

  12. Benchmark Programs 1. Real programs - SPEC benchmarks 2. Kernels - Livermore Loops and Linpack 3. Toy benchmarks - Quicksort, etc 4. Synthetic benchmarks - Dhrystone and Whetstone

  13. SPEC: System Performance Evaluation Cooperation http://www.spec.org • First Round 1989 • 10 programs yielding a single number • Second Round 1992 • CINT92 (6 integer programs) and CFP92 (14 floating point programs) • Different compiler flags are allowed for different programs • Third Round 1995 • CINT95 (8 integer programs) and CFP95 (10 floating point programs) • Same compiler flags for all programs of a given language • measures both execution time and throughput • Fourth Round scheduled to be completed by 1999

  14. SPEC Results

  15. SPEC Results

  16. Other SPEC Benchmarks • SFS97 - NFS Performance • Web96 - WWW Server Performance • HPC96 - High-end System Performance • APC, MBC, PLB, OPC, XPC - Graphics System Performance

  17. Summarizing Performance Arithmetic mean Represents total execution time Harmonic mean Consistent independent of reference Geometric mean

  18. Amdahl's Law: assessing enhancement Speedup due to enhancement E: ExTime w/o E Performance w/ E Speedup(E) = ------------- = ------------------- ExTime w/ E Performance w/o E Suppose that enhancement E accelerates a fraction Fractionenhanced of the task by a factor Speedupenhanced, and the remainder of the task is unaffected. What are the new execution time and the overall speedup due to the enhancement?

  19. Amdahl’s Law ExTimenew = ExTimeold x (1 - Fractionenhanced) + Fractionenhanced Speedupenhanced 1 ExTimeold ExTimenew Speedupoverall = = (1 - Fractionenhanced) + Fractionenhanced Speedupenhanced

  20. What’s the implication of Amdahl’s law for computer architects? Integer instructions memory FP instructions others After adding a pipelined integer instruction execution unit and cache memory (with FP emulation) Integer instructions memory FP instructions others

More Related