cluster.py revision 13774
1# Copyright (c) 2006-2007 The Regents of The University of Michigan 2# All rights reserved. 3# 4# Redistribution and use in source and binary forms, with or without 5# modification, are permitted provided that the following conditions are 6# met: redistributions of source code must retain the above copyright 7# notice, this list of conditions and the following disclaimer; 8# redistributions in binary form must reproduce the above copyright 9# notice, this list of conditions and the following disclaimer in the 10# documentation and/or other materials provided with the distribution; 11# neither the name of the copyright holders nor the names of its 12# contributors may be used to endorse or promote products derived from 13# this software without specific prior written permission. 14# 15# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 16# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 17# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR 18# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT 19# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, 20# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT 21# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE 25# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26# 27# Authors: Ron Dreslinski 28 29# Simple test script 30# 31# "m5 test.py" 32 33from __future__ import print_function 34from __future__ import absolute_import 35 36import os 37import optparse 38import sys 39 40import m5 41from m5.objects import * 42 43# -------------------- 44# Define Command Line Options 45# ==================== 46 47parser = optparse.OptionParser() 48 49parser.add_option("-d", "--detailed", action="store_true") 50parser.add_option("-t", "--timing", action="store_true") 51parser.add_option("-m", "--maxtick", type="int") 52parser.add_option("-c", "--numclusters", 53 help="Number of clusters", type="int") 54parser.add_option("-n", "--numcpus", 55 help="Number of cpus in total", type="int") 56parser.add_option("-f", "--frequency", 57 default = "1GHz", 58 help="Frequency of each CPU") 59parser.add_option("--l1size", 60 default = "32kB") 61parser.add_option("--l1latency", 62 default = 1) 63parser.add_option("--l2size", 64 default = "256kB") 65parser.add_option("--l2latency", 66 default = 10) 67parser.add_option("--rootdir", 68 help="ROot directory of Splash2", 69 default="/dist/splash2/codes/") 70parser.add_option("-b", "--benchmark", 71 help="Splash 2 benchmark to run") 72 73(options, args) = parser.parse_args() 74 75if args: 76 print("Error: script doesn't take any positional arguments") 77 sys.exit(1) 78 79# -------------------- 80# Define Splash2 Benchmarks 81# ==================== 82class Cholesky(Process): 83 executable = options.rootdir + '/kernels/cholesky/CHOLESKY' 84 cmd = 'CHOLESKY -p' + str(options.numcpus) + ' '\ 85 + options.rootdir + '/kernels/cholesky/inputs/tk23.O' 86 87class FFT(Process): 88 executable = options.rootdir + 'kernels/fft/FFT' 89 cmd = 'FFT -p' + str(options.numcpus) + ' -m18' 90 91class LU_contig(Process): 92 executable = options.rootdir + 'kernels/lu/contiguous_blocks/LU' 93 cmd = 'LU -p' + str(options.numcpus) 94 95class LU_noncontig(Process): 96 executable = options.rootdir + 'kernels/lu/non_contiguous_blocks/LU' 97 cmd = 'LU -p' + str(options.numcpus) 98 99class Radix(Process): 100 executable = options.rootdir + 'kernels/radix/RADIX' 101 cmd = 'RADIX -n524288 -p' + str(options.numcpus) 102 103class Barnes(Process): 104 executable = options.rootdir + 'apps/barnes/BARNES' 105 cmd = 'BARNES' 106 input = options.rootdir + 'apps/barnes/input.p' + str(options.numcpus) 107 108class FMM(Process): 109 executable = options.rootdir + 'apps/fmm/FMM' 110 cmd = 'FMM' 111 input = options.rootdir + 'apps/fmm/inputs/input.2048.p' + str(options.numcpus) 112 113class Ocean_contig(Process): 114 executable = options.rootdir + 'apps/ocean/contiguous_partitions/OCEAN' 115 cmd = 'OCEAN -p' + str(options.numcpus) 116 117class Ocean_noncontig(Process): 118 executable = options.rootdir + 'apps/ocean/non_contiguous_partitions/OCEAN' 119 cmd = 'OCEAN -p' + str(options.numcpus) 120 121class Raytrace(Process): 122 executable = options.rootdir + 'apps/raytrace/RAYTRACE' 123 cmd = 'RAYTRACE -p' + str(options.numcpus) + ' ' \ 124 + options.rootdir + 'apps/raytrace/inputs/teapot.env' 125 126class Water_nsquared(Process): 127 executable = options.rootdir + 'apps/water-nsquared/WATER-NSQUARED' 128 cmd = 'WATER-NSQUARED' 129 input = options.rootdir + 'apps/water-nsquared/input.p' + str(options.numcpus) 130 131class Water_spatial(Process): 132 executable = options.rootdir + 'apps/water-spatial/WATER-SPATIAL' 133 cmd = 'WATER-SPATIAL' 134 input = options.rootdir + 'apps/water-spatial/input.p' + str(options.numcpus) 135 136 137# -------------------- 138# Base L1 Cache Definition 139# ==================== 140 141class L1(Cache): 142 latency = options.l1latency 143 mshrs = 12 144 tgts_per_mshr = 8 145 146# ---------------------- 147# Base L2 Cache Definition 148# ---------------------- 149 150class L2(Cache): 151 latency = options.l2latency 152 mshrs = 92 153 tgts_per_mshr = 16 154 write_buffers = 8 155 156# ---------------------- 157# Define the clusters with their cpus 158# ---------------------- 159class Cluster: 160 pass 161 162cpusPerCluster = options.numcpus/options.numclusters 163 164busFrequency = Frequency(options.frequency) 165busFrequency *= cpusPerCluster 166 167all_cpus = [] 168all_l1s = [] 169all_l1buses = [] 170if options.timing: 171 clusters = [ Cluster() for i in range(options.numclusters)] 172 for j in range(options.numclusters): 173 clusters[j].id = j 174 for cluster in clusters: 175 cluster.clusterbus = L2XBar(clock=busFrequency) 176 all_l1buses += [cluster.clusterbus] 177 cluster.cpus = [TimingSimpleCPU(cpu_id = i + cluster.id, 178 clock=options.frequency) 179 for i in range(cpusPerCluster)] 180 all_cpus += cluster.cpus 181 cluster.l1 = L1(size=options.l1size, assoc = 4) 182 all_l1s += [cluster.l1] 183elif options.detailed: 184 clusters = [ Cluster() for i in range(options.numclusters)] 185 for j in range(options.numclusters): 186 clusters[j].id = j 187 for cluster in clusters: 188 cluster.clusterbus = L2XBar(clock=busFrequency) 189 all_l1buses += [cluster.clusterbus] 190 cluster.cpus = [DerivO3CPU(cpu_id = i + cluster.id, 191 clock=options.frequency) 192 for i in range(cpusPerCluster)] 193 all_cpus += cluster.cpus 194 cluster.l1 = L1(size=options.l1size, assoc = 4) 195 all_l1s += [cluster.l1] 196else: 197 clusters = [ Cluster() for i in range(options.numclusters)] 198 for j in range(options.numclusters): 199 clusters[j].id = j 200 for cluster in clusters: 201 cluster.clusterbus = L2XBar(clock=busFrequency) 202 all_l1buses += [cluster.clusterbus] 203 cluster.cpus = [AtomicSimpleCPU(cpu_id = i + cluster.id, 204 clock=options.frequency) 205 for i in range(cpusPerCluster)] 206 all_cpus += cluster.cpus 207 cluster.l1 = L1(size=options.l1size, assoc = 4) 208 all_l1s += [cluster.l1] 209 210# ---------------------- 211# Create a system, and add system wide objects 212# ---------------------- 213system = System(cpu = all_cpus, l1_ = all_l1s, l1bus_ = all_l1buses, 214 physmem = SimpleMemory(), 215 membus = SystemXBar(clock = busFrequency)) 216system.clock = '1GHz' 217 218system.toL2bus = L2XBar(clock = busFrequency) 219system.l2 = L2(size = options.l2size, assoc = 8) 220 221# ---------------------- 222# Connect the L2 cache and memory together 223# ---------------------- 224 225system.physmem.port = system.membus.master 226system.l2.cpu_side = system.toL2bus.slave 227system.l2.mem_side = system.membus.master 228 229# ---------------------- 230# Connect the L2 cache and clusters together 231# ---------------------- 232for cluster in clusters: 233 cluster.l1.cpu_side = cluster.clusterbus.master 234 cluster.l1.mem_side = system.toL2bus.slave 235 for cpu in cluster.cpus: 236 cpu.icache_port = cluster.clusterbus.slave 237 cpu.dcache_port = cluster.clusterbus.slave 238 239# ---------------------- 240# Define the root 241# ---------------------- 242 243root = Root(full_system = False, system = system) 244 245# -------------------- 246# Pick the correct Splash2 Benchmarks 247# ==================== 248if options.benchmark == 'Cholesky': 249 root.workload = Cholesky() 250elif options.benchmark == 'FFT': 251 root.workload = FFT() 252elif options.benchmark == 'LUContig': 253 root.workload = LU_contig() 254elif options.benchmark == 'LUNoncontig': 255 root.workload = LU_noncontig() 256elif options.benchmark == 'Radix': 257 root.workload = Radix() 258elif options.benchmark == 'Barnes': 259 root.workload = Barnes() 260elif options.benchmark == 'FMM': 261 root.workload = FMM() 262elif options.benchmark == 'OceanContig': 263 root.workload = Ocean_contig() 264elif options.benchmark == 'OceanNoncontig': 265 root.workload = Ocean_noncontig() 266elif options.benchmark == 'Raytrace': 267 root.workload = Raytrace() 268elif options.benchmark == 'WaterNSquared': 269 root.workload = Water_nsquared() 270elif options.benchmark == 'WaterSpatial': 271 root.workload = Water_spatial() 272else: 273 m5.util.panic(""" 274The --benchmark environment variable was set to something improper. 275Use Cholesky, FFT, LUContig, LUNoncontig, Radix, Barnes, FMM, OceanContig, 276OceanNoncontig, Raytrace, WaterNSquared, or WaterSpatial 277""") 278 279# -------------------- 280# Assign the workload to the cpus 281# ==================== 282 283for cluster in clusters: 284 for cpu in cluster.cpus: 285 cpu.workload = root.workload 286 287# ---------------------- 288# Run the simulation 289# ---------------------- 290 291if options.timing or options.detailed: 292 root.system.mem_mode = 'timing' 293 294# instantiate configuration 295m5.instantiate() 296 297# simulate until program terminates 298if options.maxtick: 299 exit_event = m5.simulate(options.maxtick) 300else: 301 exit_event = m5.simulate(m5.MaxTick) 302 303print('Exiting @ tick', m5.curTick(), 'because', exit_event.getCause()) 304 305