An ARM Backend for PyPyls Tracing JIT - STUPS Group

More documents

Recommendations

Info

18 5 PYPY’S APPROACH TO TRACING JITSdef interpret(bytecode, a):regs = [0] * 256pc = 0while True:opcode = ord(bytecode[pc])pc += 1if opcode == JUMP_IF_A:target = ord(bytecode[pc])pc += 1if a:pc = targetelif opcode == MOV_A_R:n = ord(bytecode[pc])pc += 1regs[n] = aelif opcode == MOV_R_A:n = ord(bytecode[pc])pc += 1a = regs[n]elif opcode == ADD_R_TO_A:n = ord(bytecode[pc])pc += 1a += regs[n]elif opcode == DECR_A:a -= 1elif opcode == RETURN_A:return aFigure 6: A very simple bytecode interpreter with registers and an accumulator.MOV_A_R 0 # i = aMOV_A_R 1 # copy of ’a’# 4:MOV_R_A 0 # i--DECR_AMOV_A_R 0MOV_R_A 2 # res += aADD_R_TO_A 1MOV_A_R 2MOV_R_A 0 # if i!=0: goto 4JUMP_IF_A 4MOV_R_A 2 # return resRETURN_AFigure 7: Example bytecode: Compute the square of the accumulator
5.1 The Shape of a Trace 19program counter. The call to can_enter_jit on the instance of the JitDriver is usedto mark a backward jump in the language interpreter and thus a possible entry point forthe JIT. Finally the call to jit_merge_point tells the JIT where it can resume normalinterpretation in case it had to leave the optimized code.tlrjitdriver = JitDriver(greens = [’pc’, ’bytecode’],reds = [’a’, ’regs’])def interpret(bytecode, a):regs = [0] * 256pc = 0while True:tlrjitdriver.jit_merge_point()opcode = ord(bytecode[pc])pc += 1if opcode == JUMP_IF_A:target = ord(bytecode[pc])pc += 1if a:if target < pc:tlrjitdriver.can_enter_jit()pc = targetelif opcode == MOV_A_R:... # rest unmodifiedFigure 8: Simple bytecode interpreter with hints appliedWith these hints applied the tracing JIT can perform profiling based on loops found inthe user program while tracing the operations the tracing interpreter performs when executingthe user loop. The downside of this is that the tracing interpreter records alsoinstructions that are related to manipulating the internal state and data structures of thelanguage interpreter. How this problem is handled is explained in Section 5.2. The tracerecorded by the tracing interpreter for a user loop is passed to the optimizer before compilingthe trace.5.1 The Shape of a TraceA trace in PyPy’s JIT is composed of operations and input arguments. The operations area list of instructions provided in SSA form [CFR + 91]. The input arguments are providedas list of variables which are used within the loop. Figure 10 shows the trace for a loopwithin the user program shown in Figure 7. Because the trace corresponds to the executionof a loop, the instructions in the trace build an infinite loop that can only be leftthrough a failing guard.Input Arguments and Types The argument types used in the PyPy JIT are types eitherrepresenting variables or constants. Arguments referring to variables are called boxes andhave the information about the type of the value they box. Constants are also representedby containers, these containers carry the type of the value and the value of the constant.
Page 1: INSTITUT FÜR INFORMATIKSoftwaretec
Page 5: AbstractA large part of the computi
Page 9 and 10: 31 IntroductionARM cores are presen
Page 11 and 12: 52 ARMARM is at the same time the n
Page 13 and 14: 2.2 The ARM Architecture 7The curre
Page 15 and 16: 2.2 The ARM Architecture 9and from
Page 17 and 18: 113 Just-in-Time CompilationJust-in
Page 19 and 20: 3.2 Trace Based Just-In-Time Compil
Page 21 and 22: 154 PyPyThe PyPy project was starte
Page 23: 175 PyPy’s Approach to Tracing JI
Page 27 and 28: 5.2 Optimizations 21loop_start(a0,
Page 29 and 30: 236 A Backend for PyPy’s JITIn th
Page 31 and 32: 6.1 Low level Code Generation Inter
Page 33 and 34: 6.2 Register Allocation 27def gen_l
Page 35 and 36: 6.3 Setup to Execute a Compiled Loo
Page 37 and 38: 6.4 Compiling Trace Operations 31Ac
Page 39 and 40: 6.4 Compiling Trace Operations 33of
Page 41 and 42: 6.5 Guards and Leaving the Loop 35o
Page 43 and 44: 37ARM using a cross compiler, such
Page 45 and 46: 8.2 Benchmarks 398.2.1 Python Bench
Page 47 and 48: 8.2 Benchmarks 41Benchmark cpython
Page 49 and 50: 8.2 Benchmarks 43Ratio6x5x4x3x2x1x0
Page 51 and 52: 8.2 Benchmarks 458.2.2 Prolog Bench
Page 53 and 54: 8.2 Benchmarks 47SWInojit, boehmjit
Page 55 and 56: 499 Related WorkThere are many virt
Page 57 and 58: 5111 AnnexBenchmark cpython [ms] no
Page 59 and 60: REFERENCES 53References[AACM07] ANC
Page 61 and 62: REFERENCES 55annual ACM SIGPLAN con
Page 63 and 64: LIST OF FIGURES 57List of Figures1

An ARM Backend for PyPyls Tracing JIT - STUPS Group

Create successful ePaper yourself

Delete template?

Save as template?