Presentation is loading. Please wait.

Presentation is loading. Please wait.

CS 594:SCIENTIFIC COMPUTING FOR ENGINEERS PERFORMANCE ANALYSIS TOOLS: PART I Gabriel Marin

Similar presentations


Presentation on theme: "CS 594:SCIENTIFIC COMPUTING FOR ENGINEERS PERFORMANCE ANALYSIS TOOLS: PART I Gabriel Marin"— Presentation transcript:

1 CS 594:SCIENTIFIC COMPUTING FOR ENGINEERS PERFORMANCE ANALYSIS TOOLS: PART I Gabriel Marin

2 Rough Outline 1. Part I Motivation Introduction to Computer Architecture Overview of Performance Analysis techniques 2. Part II Introduction to Hardware Counter Events PAPI: Access to hardware performance counters 3. Part III HPCToolkit: Low overhead, full code profiling using hardware counters sampling MIAMI: Performance diagnosis based on machine-independent application modeling 2

3 Getting results as quickly as possible? Getting correct results as quickly as possible? What about Budget? What about Development Time? What about Hardware Usage? What about Power Consumption? WHAT IS PERFORMANCE ? 3

4 WHY PERFORMANCE ANALYSIS ? Large investments in HPC systems o Procurement costs: ~$40 million / year o Operational costs:~$5 million / year o Electricity costs:1 MW year ~$1 million Efficient usage is important because of expensive and limited resources Scalability is important to achieve next bigger simulation Embedded systems have strict power and memory constraints. 4

5 SIMPLE PERFORMANCE EQUATION N – number of executed instructions C – CPI = cycles per instruction f – processor frequency I – IPC = instructions per cycle Frequency scaling provided “easy” performance gains for many years Power use increases with frequency cubed 5

6 SIMPLE PERFORMANCE EQUATION N – affected by implementation algorithm, compiler, machine instruction set (e.g. SIMD instructions) f – determined by architecture, is not going up anymore I – affected by code optimizations (manual or compiler) and by micro-architecture optimizations Current architectures can issue 6-8 micro-ops per cycle Retire 3-4 instructions per cycle (Itanium can retire 6) IPC > 1.5 is very good, ~1 is OK, many applications get IPC < 1 6

7 FACTORS AFFECTING PERFORMANCE Algorithm – biggest impact O(N*log(N)) performs much better than O(N 2 ) for useful values of N Code implementation Integer factor performance difference between efficient and inefficient implementations of the same algorithm Compiler and compiler flags Architecture 7

8 EXAMPLE: MATRIX MULTIPLY void compute(int reps) { register int i, j, k, r; for (r=0 ; r

9 MATRIX MULTIPLY: DIFFERENT COMPILERS Timing for the full code, not just the compute routine Can we explain the differences in performance? 9

10 MATRIX MULTIPLY: DIFFERENT MACHINES The two machines are part of the same family 10

11 COMPUTER ARCHITECTURE REVIEW Typically we cannot modify our target architecture, but we can tailor our code to the target architecture. In very specialized cases we want to tailor an architecture to a specific application / workload You do not have to be a computer architect to optimize code … but it really helps 11

12 COMPUTER ARCHITECTURE REVIEW CPU front-end CPU back-end 12

13 PROCESSOR FRONT-END MODEL Front-end: operates in-order Instruction fetch Instruction decode Branch predictor - speculative instruction fetch and decode ~ 1 in 5 instructions is a branch Issue buffer – store decoded instructions for the scheduler Reorder buffer – instructions decoded and not retired Branch predictor mispredict rate I-cache miss rate Fetch buffer FD Decode pipeline pipeline depth Issue buffer # entries D Reorder buffer (ROB) # entries DR I Scheduler Back-end 13

14 PROCESSOR FRONT-END STALLS Possible front-end stall events I-Cache or I-TLB miss Branch misprediction Full Reorder Buffer Branch predictor mispredict rate I-cache miss rate Fetch buffer FD Decode pipeline pipeline depth Issue buffer # entries D Reorder buffer (ROB) # entries DR I Scheduler Back-end 14

15 FRONT-END STALLS: I-CACHE or I-TLB MISS Instruction fetch stops Instructions in the front-end buffers continue to be dispatched until front-end drains (hides penalty) Front-end pipeline starts to refill after the miss event is handled; front-end refill time ~ front-end drain time Penalty ~= miss event delay Branch predictor mispredict rate I-cache miss rate Fetch buffer FD Decode pipeline pipeline depth Issue buffer # entries D Reorder buffer (ROB) # entries DR I Scheduler Back-end 15

16 FRONT-END STALLS: I-CACHE or I-TLB MISS Possible causes Execution is spread over large regions of code, with branchy unpredictable control flow Not typical for HPC Large loop footprint + small I-cache Older Itanium2: 16KB I-cache, no hardware prefetcher Space inefficient VLIW instruction set Loop fusion / loop unrolling can create large loop footprints Possible solutions Feedback directed compilation can change code layout Limit loop unrolling or fusion 16

17 FRONT-END STALLS: BRANCH MISPREDICTION Instruction fetch continues along a wrong path Useful instructions before mispredicted branch continue to be dispatched until branch enters window Pipeline is flushed when branch is resolved, mispred detected Instruction fetch starts on the correct path, front-end pipeline starts to refill Penalty ~= branch resolution time + pipeline refill Branch predictor mispredict rate I-cache miss rate Fetch buffer FD Decode pipeline pipeline depth Issue buffer # entries D Reorder buffer (ROB) # entries DR I Scheduler Back-end 17

18 BRANCH MISPREDICTION PENALTY This is the minimum penalty, proportional to the processor pipeline depth. Bulldozer has a deeper pipeline than K10 -> higher penalty Sandy Bridge added a micro-ops cache, which can lower misprediction penalty compared to Nehalem. ArchitectureBranch misprediction penalty AMD K10 (Barcelona, Istanbul, Magny-Cours)12 AMD Bulldozer20 Pentium 420 Core 2 (Conroe, Penryn)15 Nehalem17 Sandy Bridge

19 BRANCH MISPREDICTION PENALTY Branch predictors have improved in time Both Intel and AMD Modern branch predictors have very good accuracy on typical workloads, 95%+ Is there room for improvement? Does it matter if we go from 95% to 96%? 19

20 BRANCH MISPREDICTION PENALTY Branch predictors have improved in time Both Intel and AMD Modern branch predictors have very good accuracy on typical workloads, 95%+ Is there room for improvement? Does it matter if we go from 95% to 96%? Performance loss is proportional to branch misprediction rate 5% to 4% misprediction rate is a 20% improvement ~ 1 in 5 instructions is a branch in typical workloads Losses due to branch misprediction Branch misprediction rate X pipeline depth 20

21 FRONT-END STALLS: ROB FULL ROB maintains in-order state of not yet retired micro-ops μops still in the issue buffer μops in the back-end pipelines (executing) Completed μops, but ROB head micro-op did not On a long data access, other micro-ops continue to issue, but micro-ops dispatched after the stalled load cannot retire Dispatch continues until ROB fills up, then it stalls Branch predictor mispredict rate I-cache miss rate Fetch buffer FD Decode pipeline pipeline depth Issue buffer # entries D Reorder buffer (ROB) # entries DR I Scheduler Back-end 21

22 PROCESSOR BACK-END MODEL Execution units organized in stacks Can issue one μop to each issue port each cycle Deal with different instruction mixes Micro-op scheduler: unified or partitioned Register files (not shown) Bypass network to forward results between stacks Intel Sandy Bridge AMD K10 22

23 PROCESSOR BACK-END MODEL Micro-ops enter the scheduler issue buffer in order Can be dispatched out-of-order, but maintains close to FIFO Picks the oldest ready to execute micro-ops Skip micro-ops that are not ready Increases instruction level parallelism (ILP) Why try to maintain FIFO? Retirement width < Issue width Retirement balances front-end width Larger issue width to account for short variations in ILP 23

24 HOW TO DEFINE PEAK PERFORMANCE? Peak retirement rate (IPC) From the architecture point of view Peak issue of “useful” instructions HPC cares about FLOPS, mainly Adds and Multiplies Peak FLOPS rate, everything else is overhead You need many data movement instructions (Loads, Reg Copy, Data Shuffling, Data Conversion) + address arithmetic and Branches to perform useful work Cannot get close to peak for most workloads, dense linear algebra is an exception What about SIMD instructions? Peak issue of SIMD “useful” instructions 24

25 BACK-END INEFFICIENCIES Mismatch between application instruction mix and available machine resources Contention on a particular execution unit or issue port “useful” instructions are a fraction of all program instructions Instruction dependencies limit available ILP Machine units sit mostly idle Long data access Memory access misses in D-Cache or D-TLB Non-blocking caches, other instructions continue to issue Multiple outstanding accesses to memory possible Retirement stops on a long latency instruction Eventually ROB fills up, dispatch stops 25

26 LONG DATA ACCESSES Typically the main source of performance loss Architecture optimizations Multiple levels of cache – takes advantage of temporal and spatial reuse Hardware prefetchers – bring data closer to the execution units before it is needed works best with streaming memory access patterns Software optimizations High level loop nest optimizations: tiling, fusion, loop interchange, loop splitting, data layout transformations Increase temporal or spatial reuse Software prefetching – uses instruction issue bandwidth 26

27 PERFORMANCE ANALYSIS Performance analysis is part science and part art Many variables affect performance Architecture, algorithm, code implementation, compiler Caches and various micro-architecture optimizations make analysis nondeterministic You must have a feeling of what can go wrong Everyone has a different style Knowledge of computer architecture helps Compilers background helps Math / numerical algorithms background helps 27

28 Measure & Analyze: Have an optimization phase Just like testing & debugging phase It is often skipped Budget or development time constraints PERFORMANCE OPTIMIZATION CYCLE Usage / Production Measure Analyze Modify / Tune functionally complete and correct program complete, correct and well-performing program Instrumentation Code Development 28

29 PERFORMANCE ANALYSIS TECHNIQUES Performance measurement Performance modeling Simulation The line between different techniques can be blurry Modeling can use measurement or simulation results as input 29

30 PERFORMANCE MEASUREMENT Further classification Profiling vs. tracing Instrumentation vs. sampling Advantages Measures performance effects Actual code on a real system Reveals hotspots Disadvantages Instrumentation overhead can perturb measurements Sampling can have attribution errors Measures performance effects Performance insight (diagnosis) not easily apparent 30

31 MEASUREMENT: PROFILING VS TRACING Profiling Aggregate performance metrics No timeline dimension, or ordering of events Number of times a routine was invoked Time spent or cache misses incurred in a loop or a routine Memory, message communication sizes Tracing When and where events took place along a global timeline Time-stamped events (points of interest) Message communication events (sends/receives) are tracked Shows when and from/to where messages were sent Event Trace: collection of all events of a process/program sorted by time 31

32 MEASUREMENT: PROFILING Recording of summary information during execution o inclusive, exclusive time, # calls, hardware counter statistics, … Reflects performance behavior of program entities o functions, loops, basic blocks o user-defined “semantic” entities Very good for low-cost performance assessment Helps to expose hotspots Implemented through either o sampling: periodic OS interrupts or hardware counter traps o instrumentation: direct insertion of measurement code 32

33 MEASUREMENT: TRACING Recording of information about significant points (events) during program execution Save information in event record o Timestamp o CPU identifier, thread identifier o Event type and event-specific information Useful to expose interactions between parallel processes or threads Tracing Disadvantages o Traces can become very large o Instrumentation and tracing add overhead o Event buffering, clock synchronization, … 33

34 PERFORMANCE MODELING Paper-and-pencil or semi-automated Characterize the application and the architecture independently Aims to separately understand the contribution of the application and the architecture to performance Use a convolution process to predict performance Advantages Enables “what if” analysis – explore changes to the application or architecture characteristics Provides bounds on performance Can provide performance insight into bottlenecks Depends on the model’s level of detail Faster than simulation, but less accurate 34

35 PENCIL AND PAPER MODELING Back of the envelope analysis Count the type of operations at a high level Usually assumes ideal conditions, peak machine issue rates Advantages Quick performance estimates Compare algorithms / implementations with similar asymptotic complexities before implementing them Disadvantages Actual performance can be far from estimate Does not account for many low level details Hard to do by hand for large complex applications 35

36 ARCHITECTURE SIMULATION Micro-architecture/device vs. full system simulation Functional vs. timing simulation Functional: emulates target system Timing: simulates timing features of the architecture Trace-driven vs. execution-driven Advantages Obtain detailed performance metrics Account for ordering of dynamic events More accurate than analytical modeling Disadvantages Can be very slow Depends on the level of detail 36

37 Raj Jain (1991) “Contrary to common belief, performance evaluation is an art.... Like artist, each analyst has a unique style. Given the sample problem, two analysts may choose different performance metrics and evaluation methodologies.” … but even they need tools! PERFORMANCE ANALYSIS TOOLS 37


Download ppt "CS 594:SCIENTIFIC COMPUTING FOR ENGINEERS PERFORMANCE ANALYSIS TOOLS: PART I Gabriel Marin"

Similar presentations


Ads by Google