15 213 The Class That Gives CMU Its Zip Introduction to Computer Systems Randal E Bryant August 27 2002 Topics class01a ppt n Theme n Five great realities of computer systems n How this fits within CS curriculum CS 213 F 02 Course Theme n Abstraction is good but don t forget reality Courses to date emphasize abstraction n Abstract data types n Asymptotic analysis These abstractions have limits n Especially in the presence of bugs n Need to understand underlying implementations Useful outcomes n Become more effective programmers l Able to find and eliminate bugs efficiently l Able to tune program performance n Prepare for later systems classes in CS ECE l Compilers Operating Systems Networks Computer 2 Architecture Embedded Systems 15 213 F 02 Great Reality 1 Int s are not Integers Float s are not Reals Examples n Is x2 0 l Float s Yes l Int s 40000 40000 1600000000 50000 50000 n Is x y z x y z l Unsigned Signed Int s Yes l Float s 1e20 1e20 3 14 3 14 1e20 1e20 3 14 3 15 213 F 02 Computer Arithmetic Does not generate random values n Arithmetic operations have important mathematical properties Cannot assume usual properties n Due to finiteness of representations n Integer operations satisfy ring properties l Commutativity associativity distributivity n Floating point operations satisfy ordering properties l Monotonicity values of signs Observation 4 n Need to understand which abstractions apply in which contexts n Important issues for compiler writers and serious application programmers 15 213 F 02 Great Reality 2 You ve got to know assembly Chances are you ll never write program in assembly n Compilers are much better more patient than you are Understanding assembly key to machine level execution model n Behavior of programs in presence of bugs l High level language model breaks down n Tuning program performance l Understanding sources of program inefficiency n Implementing system software l Compiler has machine code as target l Operating systems must manage process state 5 15 213 F 02 Assembly Code Example Time Stamp Counter n Special 64 bit register in Intel compatible machines n Incremented every clock cycle n Read with rdtsc instruction Application n Measure time required by procedure l In units of clock cycles double t start counter P t get counter printf P required f clock cycles n t 6 15 213 F 02 Code to Read Read Counter Counter n Write small amount of assembly code using GCC s asm facility n Inserts assembly code into machine code generated by compiler static unsigned cyc hi 0 static unsigned cyc lo 0 Set hi and lo to the high and low order bits of the cycle counter void access counter unsigned hi unsigned lo asm rdtsc movl edx 0 movl eax 1 r hi r lo edx eax 7 15 213 F 02 Code to Read Read Counter Counter Record the current value of the cycle counter void start counter access counter cyc hi cyc lo Number of cycles since the last call to start counter double get counter unsigned ncyc hi ncyc lo unsigned hi lo borrow Get cycle counter access counter ncyc hi ncyc lo Do double precision subtraction lo ncyc lo cyc lo borrow lo ncyc lo hi ncyc hi cyc hi borrow return double hi 1 30 4 lo 8 15 213 F 02 Measuring Time Trickier than it Might Look n Many sources of variation Example n Sum integers from 1 to n n 100 1 000 1 000 10 000 10 000 1 000 000 1 000 000 1 000 000 000 9 Cycles 961 8 407 8 426 82 861 82 876 8 419 907 8 425 181 8 371 2305 591 Cycles n 9 61 8 41 8 43 8 29 8 29 8 42 8 43 8 37 15 213 F 02 Great Reality 3 Memory Matters Memory is not unbounded n It must be allocated and managed n Many applications are memory dominated Memory referencing bugs especially pernicious n Effects are distant in both time and space Memory performance is not uniform 10 n Cache and virtual memory effects can greatly affect program performance n Adapting program to characteristics of memory system can lead to major speed improvements 15 213 F 02 Memory Referencing Bug Example Example main main long long int int a 2 a 2 double double dd 3 14 3 14 a 2 a 2 1073741824 1073741824 Out Out of of bounds bounds reference reference printf d 15g n d printf d 15g n d exit 0 exit 0 Alpha MIPS Linux g 5 30498947741318e 315 3 1399998664856 3 14 O 3 14 3 14 3 14 Linux version gives correct result but implementing as separate function gives segmentation fault 11 15 213 F 02 Memory Referencing Errors C and C do not provide any memory protection n Out of bounds array references n Invalid pointer values n Abuses of malloc free Can lead to nasty bugs n Whether or not bug has any effect depends on system and compiler n Action at a distance l Corrupted object logically unrelated to one being accessed l Effect of bug may be first observed long after it is generated How can I deal with this 12 n Program in Java Lisp or ML n Understand what possible interactions may occur n Use or develop tools to detect referencing errors 15 213 F 02 Memory Performance Example Implementations of Matrix Multiplication n Multiple ways to nest loops ijk ijk for for i 0 i 0 i n i n i i for for j 0 j 0 j n j n j j sum sum 0 0 0 0 for for k 0 k 0 k n k n k k sum sum a i k a i k b k j b k j c i j c i j sum sum 13 jik jik for for j 0 j 0 j n j n j j for for i 0 i 0 i n i n i i sum sum 0 0 0 0 for for k 0 k 0 k n k n k k sum sum a i k a i k b k j b k j c i j c i j sum sum 15 213 F 02 Matmult Performance Alpha 21164 Too big for L1 Cache Too big for L2 Cache 160 140 120 ijk ikj jik jki kij kji 100 80 60 40 20 0 matrix size n 14 15 213 F 02 Blocked matmult perf perf Alpha Alpha 21164 21164 160 140 120 100 bijk bikj ijk ikj 80 60 40 20 0 50 7 5 100 125 150 175 200 225 250 275 300 325 350 375 400 425 450 475 500 matrix size n 15 15 213 F 02 Great Reality 4 There s more to performance than asymptotic complexity Constant factors matter too n Easily see 10 1 performance range depending on how code written n Must optimize at multiple levels algorithm data representations procedures and loops Must understand system to optimize performance 16 n How programs compiled and …
View Full Document
Unlocking...