Techniques for Speding up Pinbased Simulation为加快引脚为基础的模拟技术.ppt

Objective IS : High-level techniques for speeding up Pin-based simulation IS Not : low-level optimizations (in-lining etc.) of Pintools Two usage models Outline Two techniques: Selective simulation Conditional instrumentation PinPoints : Selecting simulation regions with Pin and SimPoint Case Study: Pin ? SimpleScalar.x86 Instruction Counts : Some IPF Applications Problem: Whole-Program Simulation is Slow Solution: Select Simulation Points Select One Point At the beginning (no skip) After 1 billion instructions After skipping a random number of instructions Select Multiple Points Manually by looking at performance data Randomly anywhere Randomly from uniform regions By program phase analysis (SimPoint : UCSD) Fine-grain sampling (SMARTS: CMU) How Pin Supports Selective Simulation? Class CONTROL : in InstLib/control.H(via instlib.H)Pintool includes the class and provides a “Handler” for “start and end of region” Provides a number of switches: For specifying “start of region” -skip instruction count  -start_address Address… For specifying “end of region”-length instruction count-stop_address Address… InstlibExamples/control $ pin –t control –skip 100 –length 500 –- helloip: 0x40000e00 104 Start ip: 0x4000105e 598 Stop Hello world Other example switches:One region: -start_address foo:10 -length 500Multiple regions: -uniform_period 1000 uniform_length 200 -ppfile foo.pp Recap: Instrumentation vs. Analysis Instrumentation routines define where instrumentation is inserted e.g. before instruction C Occurs first time an instruction is executed Analysis routines define what to do when instrumentation is activated e.g. increment counter C Occurs every time an instruction is executed Selective Simulation: Naive Approach: Conditional Analysis LOCALVAR INT32 enabled = 0; VOID Simulation() { if(!enabled) return; // Analysis code for detailed simulation } VOID Handler { switch(ev){ case CONTROL_START: enabled = 1; break; c