1

• Measure, Report, and Summarize

• Make intelligent choices

• See through the marketing hype

• Key to understanding underlying organizational motivation

Why is some hardware better than others for different programs?

What factors of system performance are hardware related?(e.g., Do we need a new machine, or a new operating system?)

How does the machine's instruction set affect performance?

Chapter 2: Performance

2

Which of these airplanes has the best performance?

Airplane Passengers Range (mi) Speed (mph)

Boeing 737-100 101 630 598Boeing 747 470 4150 610BAC/Sud Concorde 132 4000 1350Douglas DC-8-50 146 8720 544

•How much faster is the Concorde compared to the 747?

•How much bigger is the 747 than the Douglas DC-8?

3

Definitions

• Performance is in units of things-per-second

– bigger is better

• If we are primarily concerned with response time

– performance(x) = 1 execution_time(x)

" X is n times faster than Y" means

Performance(X)

n = ----------------------

Performance(Y)

4

Example

• Time of Concorde vs. Boeing 747?

• Concord is 1350 mph / 610 mph = 2.2 times faster

= 6.5 hours / 3 hours

• Throughput of Concorde vs. Boeing 747 ?

• Concord is 178,200 pmph / 286,700 pmph = 0.62 “times faster”

• Boeing is 286,700 pmph / 178,200 pmph = 1.6 “times faster”

• Boeing is 1.6 times (“60%”)faster in terms of throughput

• Concord is 2.2 times (“120%”) faster in terms of flying time

We will focus primarily on execution time for a single job

5

• Response Time (latency)

— How long does it take for my job to run?

— How long does it take to execute a job?

— How long must I wait for the database query?

• Throughput

— How many jobs can the machine run at once?

— What is the average execution rate?

— How much work is getting done?

If we upgrade a machine with a new processor what do we increase?

If we add a new machine to the lab what do we increase?

Computer Performance: TIME, TIME, TIME

6

• Elapsed Time

– counts everything (disk and memory accesses, I/O , etc.)

– a useful number, but often not good for comparison purposes

• CPU time

– doesn't count I/O or time spent running other programs

– can be broken up into system time, and user time

• Our focus: user CPU time

– time spent executing the lines of code that are "in" our program

Execution Time

7

• For some program running on machine X,

PerformanceX = 1 / Execution timeX

• "X is n times faster than Y"

PerformanceX / PerformanceY = n

• Problem:

– machine A runs a program in 20 seconds

– machine B runs the same program in 25 seconds

Book's Definition of Performance

8

Clock Cycles

• Instead of reporting execution time in seconds, we often use cycles

• Clock “ticks” indicate when to start activities (one abstraction):

• cycle time = time between ticks = seconds per cycle

• clock rate (frequency) = cycles per second (1 Hz. = 1 cycle/sec)

A 200 Mhz. clock has a cycle time

time

seconds

program

cycles

program

seconds

cycle

1

200 106 109 5 nanoseconds

9

So, to improve performance (everything else being equal) you can either

________ the # of required cycles for a program, or

________ the clock cycle time or, said another way,

________ the clock rate.

How to Improve Performance

seconds

program

cycles

program

seconds

cycle

10

• Could assume that # of cycles = # of instructions

This assumption is incorrect,

different instructions take different amounts of time on different machines.

Why? hint: remember that these are machine instructions, not lines of C code

time

1st

inst

ruct

ion

2nd

inst

ruct

ion

3rd

inst

ruct

ion

4th

5th

6th ...

How many cycles are required for a program?

11

• Multiplication takes more time than addition

• Floating point operations take longer than integer ones

• Accessing memory takes more time than accessing registers

• Important point: changing the cycle time often changes the number of cycles required for various instructions (more later)

time

Different numbers of cycles for different instructions

12

• Our favorite program runs in 10 seconds on computer A, which has a 400 Mhz. clock. We are trying to help a computer designer build a new machine B, that will run this program in 6 seconds. The designer can use new (or perhaps more expensive) technology to substantially increase the clock rate, but has informed us that this increase will affect the rest of the CPU design, causing machine B to require 1.2 times as many clock cycles as machine A for the same program. What clock rate should we tell the designer to target?"

• Don't Panic, can easily work this out from basic principles

Example

13

• A given program will require

– some number of instructions (machine instructions)

– some number of cycles

– some number of seconds

• We have a vocubulary that relates these quantities:

– cycle time (seconds per cycle)

– clock rate (cycles per second)

– CPI (cycles per instruction)

a floating point intensive application might have a higher CPI

– MIPS (millions of instructions per second)

this would be higher for a program using simple instructions

Now that we understand cycles

14

Performance

• Performance is determined by execution time

• Do any of the other variables equal performance?

– # of cycles to execute program?

– # of instructions in program?

– # of cycles per second?

– average # of cycles per instruction?

– average # of instructions per second?

• Common pitfall: thinking one of the variables is indicative of performance when it really isn’t.

15

• Suppose we have two implementations of the same instruction set architecture (ISA).

For some program,

Machine A has a clock cycle time of 10 ns. and a CPI of 2.0 Machine B has a clock cycle time of 20 ns. and a CPI of 1.2

What machine is faster for this program, and by how much?

• If two machines have the same ISA which of our quantities (e.g., clock rate, CPI, execution time, # of instructions, MIPS) will always be identical?

CPI Example

16

• A compiler designer is trying to decide between two code sequences for a particular machine. Based on the hardware implementation, there are three different classes of instructions: Class A, Class B, and Class C, and they require one, two, and three cycles (respectively).

The first code sequence has 5 instructions: 2 of A, 1 of B, and 2 of CThe second sequence has 6 instructions: 4 of A, 1 of B, and 1 of C.

Which sequence will be faster? How much?What is the CPI for each sequence?

# of Instructions Example

17

• Two different compilers are being tested for a 100 MHz. machine with three different classes of instructions: Class A, Class B, and Class C, which require one, two, and three cycles (respectively). Both compilers are used to produce code for a large piece of software.

The first compiler's code uses 5 million Class A instructions, 1 million Class B instructions, and 1 million Class C instructions.

The second compiler's code uses 10 million Class A instructions, 1 million Class B instructions, and 1 million Class C instructions.

• Which sequence will be faster according to MIPS?• Which sequence will be faster according to execution time?

MIPS example

18

• Performance best determined by running a real application

– Use programs typical of expected workload

– Or, typical of expected class of applicationse.g., compilers/editors, scientific applications, graphics, etc.

• Small benchmarks

– nice for architects and designers

– easy to standardize

– can be abused

• SPEC (System Performance Evaluation Cooperative)

– companies have agreed on a set of real program and inputs

– can still be abused (Intel’s “other” bug)

– valuable indicator of performance (and compiler technology)

Benchmarks

19

Basis of Evaluation

Actual Target Workload

Full Application Benchmarks

Small “Kernel” Benchmarks

Microbenchmarks

Pros Cons

• representative• very specific• non-portable• difficult to run, or measure• hard to identify cause

• portable• widely used• improvements useful in reality

• easy to run, early in design cycle

• identify peak capability and potential bottlenecks

•less representative

• easy to “fool”

• “peak” may be a long way from application performance

20

SPEC ‘89

• Compiler “enhancements” and performance

21

SPEC95

• Eighteen application benchmarks (with inputs) reflecting a technical computing workload

• Eight integer– go, m88ksim, gcc, compress, li, ijpeg, perl, vortex

• Ten floating-point intensive– tomcatv, swim, su2cor, hydro2d, mgrid, applu, turb3d,

apsi, fppp, wave5• Must run with standard compiler flags

– eliminate special undocumented incantations that may not even generate working code for real programs

22

SPEC ‘95

go Artificial intelligence; plays the game of Gom88ksim Motorola 88k chip simulator; runs test programgcc The Gnu C compiler generating SPARC codecompress Compresses and decompresses file in memoryli Lisp interpreterijpeg Graphic compression and decompressionperl Manipulates strings and prime numbers in the special-purpose programming language Perlvortex A database programtomcatv A mesh generation programswim Shallow water model with 513 x 513 gridsu2cor quantum physics; Monte Carlo simulationhydro2d Astrophysics; Hydrodynamic Naiver Stokes equationsmgrid Multigrid solver in 3-D potential fieldapplu Parabolic/elliptic partial differential equationstrub3d Simulates isotropic, homogeneous turbulence in a cubeapsi Solves problems regarding temperature, wind velocity, and distribution of pollutantfpppp Quantum chemistrywave5 Plasma physics; electromagnetic particle simulation

23

SPEC CPU2000

164.gzip C Compression175.vpr C FPGA Circuit Placement and Routing176.gcc C C Programming Language Compiler181.mcf C Combinatorial Optimization186.crafty C Game Playing: Chess197.parser C Word Processing252.eon C++ Computer Visualization253.perlbmk C PERL Programming Language254.gap C Group Theory, Interpreter255.vortex C Object-oriented Database256.bzip2 C Compression300.twolf C Place and Route Simulator168.wupwise Fortran 77Physics / Quantum Chromodynamics171.swim Fortran 77Shallow Water Modeling172.mgrid Fortran 77Multi-grid Solver: 3D Potential Field173.applu Fortran 77Parabolic / Elliptic Partial Differential 177.mesa C 3-D Graphics Library178.galgel Fortran 90Computational Fluid Dynamics179.art C Image Recognition / Neural Networks183.equake C Seismic Wave Propagation Simulation187.facerec Fortran 90Image Processing: Face Recognition188.ammp C Computational Chemistry189.lucas Fortran 90Number Theory / Primality Testing191.fma3d Fortran 90Finite-element Crash Simulation200.sixtrack Fortran 77High Energy Nuclear Physics Accelerator 301.apsi Fortran 77Meteorology: Pollutant Distribution

24

SPEC ‘95

Does doubling the clock rate double the performance?

Can a machine with a slower clock rate have better performance?

25

SpecCPU2000 Benchmark results

• Advanced Micro Devices AMD Athlon (TM) XP 2700+ 878 913• Advanced Micro Devices, AMD Athlon (TM) XP 2800+ 898 933• Advanced Micro Devices, AMD Athlon (TM) XP 3000+ 960 995• IBM Corporation pSeries 630 (1450 MHz, 1 CPU) 884 910• Intel Corporation (2.67 GHz, Pentium 4 processor) 998 1005• Intel Corporation (2.8 GHz, Pentium 4 processor) 1032 1040• Intel Corporation (3.06 GHz, Pentium 4 processor 1099 1107• Sun Microsystems Sun Blade 2000 (1.015GHz) 516 576 • Sun Microsystems Sun Blade Model 2050 537 610

26

More SPEC CPU2000 Results• This data is current as of Wed Mar 19 20:01:27 PST 2003 • SPECint ® 2000• 1099.0 3067.0 MHz Pentium 4 HT Intel D850EMVR motherboard (3.06 GHz, Pentium 4 processor wi • 1085.0 3066.0 MHz Pentium 4 Dell Precision WorkStation 350 (3.06 GHz P4) • 967.0 2800.0 MHz Xeon Fujitsu Siemens Computers CELSIUS R610 • 960.0 2167.0 MHz Athlon ASUS A7N8X Deluxe Motherboard, AMD Athlon (TM) XP 3000+ • 909.0 1450.0 MHz POWER4+ IBM Corporation IBM eServer pSeries 650 Model 6M2 (1450 MHz, 1 CPU) • 898.0 2250.0 MHz Athlon XP ASUS A7N8X (REV 1.02) Motherboard, AMD Athlon (TM) XP 2800+ • 845.0 1250.0 MHz Alpha Hewlett-Packard Company hp AlphaServer ES45 68/1250 • 822.0 1300.0 MHz POWER4 IBM Corporation IBM eServer pSeries 655 Model 651 (1300 MHz, 1 CPU) • 810.0 1000.0 MHz Itanium 2 Hewlett-Packard Company hp server rx2600 (1000 MHz, Itanium 2) • 751.0 2133.0 MHz Athlon MSI K7D Master Motherboard, AMD Athlon (TM) MP 2600+ • 747.0 1350.0 MHz SPARC64 V Fujitsu Limited PRIMEPOWER900 (1350MHz) • 737.0 2000.0 MHz Athlon MP Asus A7M266-D Motherboard, AMD Athlon (TM) MP 2400+ • 648.0 1400.0 MHz Pentium III Dell PowerEdge 1500SC (1.4 GHz PIII) • 642.0 875.0 MHz PA-RISC Hewlett-Packard Company hp workstation c3750 • 569.0 750.0 MHz PA-RISC 8700 Hewlett-Packard Company hp workstation j6700 • 537.0 1050.0 MHz UltraSPARC III Cu Sun Microsystems Sun Blade Model 2050 • 512.0 810.0 MHz SPARC64 GP Fujitsu Limited PRIMEPOWER650 (810MHz) • 483.0 600.0 MHz R14000 SGI SGI Origin 3200 1X 600MHz R14k • 471.0 600.0 MHz R14000A SGI SGI Origin 300 1X 600MHz R14000A • 439.0 900.0 MHz UltraSPARC III Sun Microsystems Sun Blade 1000 Model 1900 • 437.0 1000.0 MHz Pentium III Xeon Dell PowerEdge 4400 (1.0 GHz PIII Xeon) • 431.0 750.0 MHz RS64 IV IBM Corporation IBM eServer pSeries 620 Model 6F0 (750 MHz) • 417.0 552.0 MHz PA-RISC 8600 Hewlett-Packard Company hp visualize j6000 UNIX workstation • 379.0 800.0 MHz Itanium Hewlett-Packard Company hp server rx4610 • 333.0 450.0 MHz POWER3 II IBM Corporation RS/6000 44P-170 (450 MHz) • 328.0 400.0 MHz R12000 SGI SGI Origin 3200 400MHz R12k • 313.0 400.0 MHz PA-RISC 8500 Hewlett-Packard Company hp visualize b2000 UNIX workstation • 264.0 500.0 MHz PowerPC RS64 III IBM Corporation RS/6000 Model 7026-M80 (1 CPU) • 225.0 480.0 MHz UltraSPARC II Sun Microsystems Sun Enterprise 450 • 168.0 340.0 MHz RS64 II IBM Corporation RISC System/6000 H70 (1 CPU) • 165.0 500.0 MHz UltraSPARC IIe Sun Microsystems Sun Blade 100 • 99.4 250.0 MHzP owerPC 604e IBM Corporation RS/6000 43P-150 (250MHz)

27

Metrics of performance

Compiler

Programming Language

Application

DatapathControl

Transistors Wires Pins

ISA

Function Units

(millions) of Instructions per second – MIPS(millions) of (F.P.) operations per second – MFLOP/s

Cycles per second (clock rate)

Megabytes per second

Answers per month

Useful Operations per second

Each metric has a place and a purpose, and each can be misused

28

Aspects of CPU Performance

CPU time = Seconds = Instructions x Cycles x Seconds

Program Program Instruction Cycle

CPU time = Seconds = Instructions x Cycles x Seconds

Program Program Instruction Cycle

instr. count CPI clock rateProgram

Compiler

Instr. Set Arch.

Organization

Technology

30

CPI

CPU time = ClockCycleTime * CPI * Ii = 1

n

i i

CPI = CPI * F where F = I i = 1

n

i i i i

Instruction Count

"instruction frequency"

Invest Resources where time is Spent!

CPI = (CPU Time * Clock Rate) / Instruction Count = Clock Cycles / Instruction Count

“Average cycles per instruction”

31

Example (RISC processor)

Typical Mix

Base Machine (Reg / Reg)

Op Freq Cycles CPI(i) % Time

ALU 50% 1 .5 23%

Load 20% 5 1.0 45%

Store 10% 3 .3 14%

Branch 20% 2 .4 18%

2.2

How much faster would the machine be if a better data cachereduced the average load time to 2 cycles?

How does this compare with using branch prediction to shave a cycle off the branch time?

What if two ALU instructions could be executed at once?

32

Execution Time After Improvement =

Execution Time Unaffected +( Execution Time Affected / Amount of Improvement )

• Example:

"Suppose a program runs in 100 seconds on a machine, with multiply responsible for 80 seconds of this time. How much do we have to improve the speed of multiplication if we want the program to run 4 times faster?"

How about making it 5 times faster?

• Principle: Make the common case fast

Amdahl's Law

33

Amdahl's Law

Speedup due to enhancement E: ExTime w/o E Performance w/ ESpeedup(E) = -------------------- = --------------------- ExTime w/ E Performance w/o E

Suppose that enhancement E accelerates a fraction F of the taskby a factor S and the remainder of the task is unaffected then,

ExTime(with E) Š ((1-F) + F/S) X ExTime(without E)

Speedup(with E) Š 1 (1-F) + F/S

34

• Suppose we enhance a machine making all floating-point instructions run five times faster. If the execution time of some benchmark before the floating-point enhancement is 10 seconds, what will the speedup be if half of the 10 seconds is spent executing floating-point instructions?

• We are looking for a benchmark to show off the new floating-point unit described above, and want the overall benchmark to show a speedup of 3. One benchmark we are considering runs for 100 seconds with the old floating-point hardware. How much of the execution time would floating-point instructions have to account for in this program in order to yield our desired speedup on this benchmark?

Example

35

• Performance is specific to a particular program/s

– Total execution time is a consistent summary of performance

• For a given architecture performance increases come from:

– increases in clock rate (without adverse CPI affects)

– improvements in processor organization that lower CPI

– compiler enhancements that lower CPI and/or instruction count

• Pitfall: expecting improvement in one aspect of a machine’s

performance to affect the total performance

• You should not always believe everything you read! Read carefully!

(see newspaper articles, e.g., Exercise 2.37)

Remember