diff options
Diffstat (limited to 'configs/common')
| -rw-r--r-- | configs/common/CacheConfig.py | 7 | ||||
| -rw-r--r-- | configs/common/CpuConfig.py | 25 | ||||
| -rw-r--r-- | configs/common/MemConfig.py | 10 | ||||
| -rw-r--r-- | configs/common/Options.py | 13 | ||||
| -rw-r--r-- | configs/common/Simulation.py | 10 |
5 files changed, 63 insertions, 2 deletions
diff --git a/configs/common/CacheConfig.py b/configs/common/CacheConfig.py index d54df7490..b96a7011c 100644 --- a/configs/common/CacheConfig.py +++ b/configs/common/CacheConfig.py @@ -69,6 +69,13 @@ def config_cache(options, system): # Set the cache line size of the system system.cache_line_size = options.cacheline_size + # If elastic trace generation is enabled, make sure the memory system is + # minimal so that compute delays do not include memory access latencies. + # Configure the compulsory L1 caches for the O3CPU, do not configure + # any more caches. + if options.l2cache and options.elastic_trace_en: + fatal("When elastic trace is enabled, do not configure L2 caches.") + if options.l2cache: # Provide a clock for the L2 and the L1-to-L2 bus here as they # are not connected using addTwoLevelCacheHierarchy. Use the diff --git a/configs/common/CpuConfig.py b/configs/common/CpuConfig.py index 6b530623d..3220acecd 100644 --- a/configs/common/CpuConfig.py +++ b/configs/common/CpuConfig.py @@ -53,6 +53,7 @@ _cpu_aliases_all = [ ("minor", "MinorCPU"), ("detailed", "DerivO3CPU"), ("kvm", ("ArmKvmCPU", "ArmV8KvmCPU", "X86KvmCPU")), + ("trace", "TraceCPU"), ] # Filtered list of aliases. Only aliases for existing CPUs exist in @@ -108,6 +109,30 @@ def cpu_names(): """Return a list of valid CPU names.""" return _cpu_classes.keys() + _cpu_aliases.keys() +def config_etrace(cpu_cls, cpu_list, options): + if issubclass(cpu_cls, m5.objects.DerivO3CPU): + # Assign the same file name to all cpus for now. This must be + # revisited when creating elastic traces for multi processor systems. + for cpu in cpu_list: + # Attach the elastic trace probe listener. Set the protobuf trace + # file names. Set the dependency window size equal to the cpu it + # is attached to. + cpu.traceListener = m5.objects.ElasticTrace( + instFetchTraceFile = options.inst_trace_file, + dataDepTraceFile = options.data_trace_file, + depWindowSize = 3 * cpu.numROBEntries) + # Make the number of entries in the ROB, LQ and SQ very + # large so that there are no stalls due to resource + # limitation as such stalls will get captured in the trace + # as compute delay. For replay, ROB, LQ and SQ sizes are + # modelled in the Trace CPU. + cpu.numROBEntries = 512; + cpu.LQEntries = 128; + cpu.SQEntries = 128; + else: + fatal("%s does not support data dependency tracing. Use a CPU model of" + " type or inherited from DerivO3CPU.", cpu_cls) + # The ARM detailed CPU is special in the sense that it doesn't exist # in the normal object hierarchy, so we have to add it manually. try: diff --git a/configs/common/MemConfig.py b/configs/common/MemConfig.py index 286898798..4685cd5d1 100644 --- a/configs/common/MemConfig.py +++ b/configs/common/MemConfig.py @@ -187,6 +187,11 @@ def config_mem(options, system): cls = get(options.mem_type) mem_ctrls = [] + if options.elastic_trace_en and not issubclass(cls, \ + m5.objects.SimpleMemory): + fatal("When elastic trace is enabled, configure mem-type as " + "simple-mem.") + # The default behaviour is to interleave memory channels on 128 # byte granularity, or cache line granularity if larger than 128 # byte. This value is based on the locality seen across a large @@ -206,6 +211,11 @@ def config_mem(options, system): options.mem_ranks: mem_ctrl.ranks_per_channel = options.mem_ranks + if options.elastic_trace_en: + mem_ctrl.latency = '1ns' + print "For elastic trace, over-riding Simple Memory " \ + "latency to 1ns." + mem_ctrls.append(mem_ctrl) subsystem.mem_ctrls = mem_ctrls diff --git a/configs/common/Options.py b/configs/common/Options.py index f4cf9fbd6..45be8e2f8 100644 --- a/configs/common/Options.py +++ b/configs/common/Options.py @@ -87,6 +87,19 @@ def addCommonOptions(parser): Only used if multiple programs are specified. If true, then the number of threads per cpu is same as the number of programs.""") + parser.add_option("--elastic-trace-en", action="store_true", + help="""Enable capture of data dependency and instruction + fetch traces using elastic trace probe.""") + # Trace file paths input to trace probe in a capture simulation and input + # to Trace CPU in a replay simulation + parser.add_option("--inst-trace-file", action="store", type="string", + help="""Instruction fetch trace file input to + Elastic Trace probe in a capture simulation and + Trace CPU in a replay simulation""", default="") + parser.add_option("--data-trace-file", action="store", type="string", + help="""Data dependency trace file input to + Elastic Trace probe in a capture simulation and + Trace CPU in a replay simulation""", default="") # Memory Options parser.add_option("--list-mem-types", diff --git a/configs/common/Simulation.py b/configs/common/Simulation.py index 73874674c..13edf601b 100644 --- a/configs/common/Simulation.py +++ b/configs/common/Simulation.py @@ -459,10 +459,11 @@ def run(options, root, testsys, cpu_class): for i in xrange(np): if options.fast_forward: testsys.cpu[i].max_insts_any_thread = int(options.fast_forward) - switch_cpus[i].system = testsys + switch_cpus[i].system = testsys switch_cpus[i].workload = testsys.cpu[i].workload switch_cpus[i].clk_domain = testsys.cpu[i].clk_domain - switch_cpus[i].progress_interval = testsys.cpu[i].progress_interval + switch_cpus[i].progress_interval = \ + testsys.cpu[i].progress_interval # simulation period if options.maxinsts: switch_cpus[i].max_insts_any_thread = options.maxinsts @@ -470,6 +471,11 @@ def run(options, root, testsys, cpu_class): if options.checker: switch_cpus[i].addCheckerCpu() + # If elastic tracing is enabled attach the elastic trace probe + # to the switch CPUs + if options.elastic_trace_en: + CpuConfig.config_etrace(cpu_class, switch_cpus, options) + testsys.switch_cpus = switch_cpus switch_cpu_list = [(testsys.cpu[i], switch_cpus[i]) for i in xrange(np)] |