2010-01-30 05:29:27 +01:00
|
|
|
# Copyright (c) 2006-2007 The Regents of The University of Michigan
|
|
|
|
# Copyright (c) 2009 Advanced Micro Devices, Inc.
|
|
|
|
# All rights reserved.
|
|
|
|
#
|
|
|
|
# Redistribution and use in source and binary forms, with or without
|
|
|
|
# modification, are permitted provided that the following conditions are
|
|
|
|
# met: redistributions of source code must retain the above copyright
|
|
|
|
# notice, this list of conditions and the following disclaimer;
|
|
|
|
# redistributions in binary form must reproduce the above copyright
|
|
|
|
# notice, this list of conditions and the following disclaimer in the
|
|
|
|
# documentation and/or other materials provided with the distribution;
|
|
|
|
# neither the name of the copyright holders nor the names of its
|
|
|
|
# contributors may be used to endorse or promote products derived from
|
|
|
|
# this software without specific prior written permission.
|
|
|
|
#
|
|
|
|
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
|
|
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
|
|
|
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
|
|
|
# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
|
|
|
|
# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
|
|
|
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
|
|
|
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|
|
|
# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|
|
|
# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
|
|
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
|
|
|
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
|
|
#
|
|
|
|
# Authors: Brad Beckmann
|
|
|
|
|
|
|
|
import math
|
|
|
|
import m5
|
|
|
|
from m5.objects import *
|
|
|
|
from m5.defines import buildEnv
|
2012-07-11 07:51:53 +02:00
|
|
|
from Ruby import create_topology
|
2010-01-30 05:29:27 +01:00
|
|
|
|
|
|
|
#
|
|
|
|
# Note: the L1 Cache latency is only used by the sequencer on fast path hits
|
|
|
|
#
|
|
|
|
class L1Cache(RubyCache):
|
|
|
|
latency = 3
|
|
|
|
|
|
|
|
#
|
|
|
|
# Note: the L2 Cache latency is not currently used
|
|
|
|
#
|
|
|
|
class L2Cache(RubyCache):
|
|
|
|
latency = 15
|
|
|
|
|
2010-08-20 20:44:09 +02:00
|
|
|
def define_options(parser):
|
|
|
|
return
|
|
|
|
|
2014-03-17 23:40:15 +01:00
|
|
|
def create_system(options, system, dma_ports, ruby_system):
|
2014-01-04 07:03:33 +01:00
|
|
|
|
|
|
|
if buildEnv['PROTOCOL'] != 'MESI_Two_Level':
|
|
|
|
fatal("This script requires the MESI_Two_Level protocol to be built.")
|
2010-01-30 05:29:27 +01:00
|
|
|
|
|
|
|
cpu_sequencers = []
|
2014-01-04 07:03:33 +01:00
|
|
|
|
2010-01-30 05:29:27 +01:00
|
|
|
#
|
|
|
|
# The ruby network creation expects the list of nodes in the system to be
|
|
|
|
# consistent with the NetDest list. Therefore the l1 controller nodes must be
|
|
|
|
# listed before the directory nodes and directory nodes before dma nodes, etc.
|
|
|
|
#
|
|
|
|
l1_cntrl_nodes = []
|
|
|
|
l2_cntrl_nodes = []
|
|
|
|
dir_cntrl_nodes = []
|
|
|
|
dma_cntrl_nodes = []
|
|
|
|
|
|
|
|
#
|
|
|
|
# Must create the individual controllers before the network to ensure the
|
|
|
|
# controller constructors are called before the network constructor
|
|
|
|
#
|
2011-03-25 18:13:50 +01:00
|
|
|
l2_bits = int(math.log(options.num_l2caches, 2))
|
|
|
|
block_size_bits = int(math.log(options.cacheline_size, 2))
|
2014-01-04 07:03:33 +01:00
|
|
|
|
2010-01-30 05:29:27 +01:00
|
|
|
for i in xrange(options.num_cpus):
|
|
|
|
#
|
|
|
|
# First create the Ruby objects associated with this cpu
|
|
|
|
#
|
|
|
|
l1i_cache = L1Cache(size = options.l1i_size,
|
2011-03-25 18:13:50 +01:00
|
|
|
assoc = options.l1i_assoc,
|
2012-10-27 23:04:30 +02:00
|
|
|
start_index_bit = block_size_bits,
|
|
|
|
is_icache = True)
|
2010-01-30 05:29:27 +01:00
|
|
|
l1d_cache = L1Cache(size = options.l1d_size,
|
2011-03-25 18:13:50 +01:00
|
|
|
assoc = options.l1d_assoc,
|
2012-10-27 23:04:30 +02:00
|
|
|
start_index_bit = block_size_bits,
|
|
|
|
is_icache = False)
|
2010-01-30 05:29:27 +01:00
|
|
|
|
2012-12-11 17:05:56 +01:00
|
|
|
prefetcher = RubyPrefetcher.Prefetcher()
|
|
|
|
|
2011-05-23 23:29:23 +02:00
|
|
|
l1_cntrl = L1Cache_Controller(version = i,
|
2013-05-21 18:32:38 +02:00
|
|
|
L1Icache = l1i_cache,
|
|
|
|
L1Dcache = l1d_cache,
|
2011-07-01 02:49:26 +02:00
|
|
|
l2_select_num_bits = l2_bits,
|
2012-01-23 18:07:14 +01:00
|
|
|
send_evictions = (
|
|
|
|
options.cpu_type == "detailed"),
|
2012-12-11 17:05:56 +01:00
|
|
|
prefetcher = prefetcher,
|
|
|
|
ruby_system = ruby_system,
|
2014-09-01 23:55:30 +02:00
|
|
|
clk_domain=system.cpu[i].clk_domain,
|
2013-08-20 18:32:31 +02:00
|
|
|
transitions_per_cycle=options.ports,
|
2012-12-11 17:05:56 +01:00
|
|
|
enable_prefetch = False)
|
2011-05-23 23:29:23 +02:00
|
|
|
|
2010-03-22 05:22:20 +01:00
|
|
|
cpu_seq = RubySequencer(version = i,
|
|
|
|
icache = l1i_cache,
|
2010-01-30 05:29:27 +01:00
|
|
|
dcache = l1d_cache,
|
2014-09-01 23:55:30 +02:00
|
|
|
clk_domain=system.cpu[i].clk_domain,
|
2011-07-01 02:49:26 +02:00
|
|
|
ruby_system = ruby_system)
|
2010-01-30 05:29:27 +01:00
|
|
|
|
2011-05-23 23:29:23 +02:00
|
|
|
l1_cntrl.sequencer = cpu_seq
|
2013-01-14 17:05:14 +01:00
|
|
|
exec("ruby_system.l1_cntrl%d = l1_cntrl" % i)
|
2014-01-04 07:03:33 +01:00
|
|
|
|
2010-01-30 05:29:27 +01:00
|
|
|
# Add controllers and sequencers to the appropriate lists
|
|
|
|
cpu_sequencers.append(cpu_seq)
|
|
|
|
l1_cntrl_nodes.append(l1_cntrl)
|
2014-01-04 07:03:33 +01:00
|
|
|
|
2014-09-01 23:55:47 +02:00
|
|
|
# Connect the L1 controllers and the network
|
|
|
|
l1_cntrl.requestFromL1Cache = ruby_system.network.slave
|
|
|
|
l1_cntrl.responseFromL1Cache = ruby_system.network.slave
|
|
|
|
l1_cntrl.unblockFromL1Cache = ruby_system.network.slave
|
|
|
|
|
|
|
|
l1_cntrl.requestToL1Cache = ruby_system.network.master
|
|
|
|
l1_cntrl.responseToL1Cache = ruby_system.network.master
|
|
|
|
|
|
|
|
|
2011-03-25 18:13:50 +01:00
|
|
|
l2_index_start = block_size_bits + l2_bits
|
|
|
|
|
2010-01-30 05:29:27 +01:00
|
|
|
for i in xrange(options.num_l2caches):
|
|
|
|
#
|
|
|
|
# First create the Ruby objects associated with this cpu
|
|
|
|
#
|
|
|
|
l2_cache = L2Cache(size = options.l2_size,
|
2011-03-25 18:13:50 +01:00
|
|
|
assoc = options.l2_assoc,
|
|
|
|
start_index_bit = l2_index_start)
|
2010-01-30 05:29:27 +01:00
|
|
|
|
|
|
|
l2_cntrl = L2Cache_Controller(version = i,
|
2013-05-21 18:32:38 +02:00
|
|
|
L2cache = l2_cache,
|
2013-08-20 18:32:31 +02:00
|
|
|
transitions_per_cycle=options.ports,
|
2011-07-01 02:49:26 +02:00
|
|
|
ruby_system = ruby_system)
|
2014-01-04 07:03:33 +01:00
|
|
|
|
2013-01-14 17:05:14 +01:00
|
|
|
exec("ruby_system.l2_cntrl%d = l2_cntrl" % i)
|
2010-01-30 05:29:27 +01:00
|
|
|
l2_cntrl_nodes.append(l2_cntrl)
|
2014-01-04 07:03:33 +01:00
|
|
|
|
2014-09-01 23:55:47 +02:00
|
|
|
# Connect the L2 controllers and the network
|
|
|
|
l2_cntrl.DirRequestFromL2Cache = ruby_system.network.slave
|
|
|
|
l2_cntrl.L1RequestFromL2Cache = ruby_system.network.slave
|
|
|
|
l2_cntrl.responseFromL2Cache = ruby_system.network.slave
|
|
|
|
|
|
|
|
l2_cntrl.unblockToL2Cache = ruby_system.network.master
|
|
|
|
l2_cntrl.L1RequestToL2Cache = ruby_system.network.master
|
|
|
|
l2_cntrl.responseToL2Cache = ruby_system.network.master
|
|
|
|
|
|
|
|
|
2013-08-19 09:52:27 +02:00
|
|
|
phys_mem_size = sum(map(lambda r: r.size(), system.mem_ranges))
|
2013-06-29 04:36:11 +02:00
|
|
|
assert(phys_mem_size % options.num_dirs == 0)
|
2010-01-30 05:29:27 +01:00
|
|
|
mem_module_size = phys_mem_size / options.num_dirs
|
|
|
|
|
2014-09-01 23:55:47 +02:00
|
|
|
|
sim: Add the notion of clock domains to all ClockedObjects
This patch adds the notion of source- and derived-clock domains to the
ClockedObjects. As such, all clock information is moved to the clock
domain, and the ClockedObjects are grouped into domains.
The clock domains are either source domains, with a specific clock
period, or derived domains that have a parent domain and a divider
(potentially chained). For piece of logic that runs at a derived clock
(a ratio of the clock its parent is running at) the necessary derived
clock domain is created from its corresponding parent clock
domain. For now, the derived clock domain only supports a divider,
thus ensuring a lower speed compared to its parent. Multiplier
functionality implies a PLL logic that has not been modelled yet
(create a separate clock instead).
The clock domains should be used as a mechanism to provide a
controllable clock source that affects clock for every clocked object
lying beneath it. The clock of the domain can (in a future patch) be
controlled by a handler responsible for dynamic frequency scaling of
the respective clock domains.
All the config scripts have been retro-fitted with clock domains. For
the System a default SrcClockDomain is created. For CPUs that run at a
different speed than the system, there is a seperate clock domain
created. This domain incorporates the CPU and the associated
caches. As before, Ruby runs under its own clock domain.
The clock period of all domains are pre-computed, such that no virtual
functions or multiplications are needed when calling
clockPeriod. Instead, the clock period is pre-computed when any
changes occur. For this to be possible, each clock domain tracks its
children.
2013-06-27 11:49:49 +02:00
|
|
|
# Run each of the ruby memory controllers at a ratio of the frequency of
|
|
|
|
# the ruby system
|
|
|
|
# clk_divider value is a fix to pass regression.
|
|
|
|
ruby_system.memctrl_clk_domain = DerivedClockDomain(
|
|
|
|
clk_domain=ruby_system.clk_domain,
|
|
|
|
clk_divider=3)
|
|
|
|
|
2010-01-30 05:29:27 +01:00
|
|
|
for i in xrange(options.num_dirs):
|
|
|
|
#
|
|
|
|
# Create the Ruby objects associated with the directory controller
|
|
|
|
#
|
|
|
|
|
sim: Add the notion of clock domains to all ClockedObjects
This patch adds the notion of source- and derived-clock domains to the
ClockedObjects. As such, all clock information is moved to the clock
domain, and the ClockedObjects are grouped into domains.
The clock domains are either source domains, with a specific clock
period, or derived domains that have a parent domain and a divider
(potentially chained). For piece of logic that runs at a derived clock
(a ratio of the clock its parent is running at) the necessary derived
clock domain is created from its corresponding parent clock
domain. For now, the derived clock domain only supports a divider,
thus ensuring a lower speed compared to its parent. Multiplier
functionality implies a PLL logic that has not been modelled yet
(create a separate clock instead).
The clock domains should be used as a mechanism to provide a
controllable clock source that affects clock for every clocked object
lying beneath it. The clock of the domain can (in a future patch) be
controlled by a handler responsible for dynamic frequency scaling of
the respective clock domains.
All the config scripts have been retro-fitted with clock domains. For
the System a default SrcClockDomain is created. For CPUs that run at a
different speed than the system, there is a seperate clock domain
created. This domain incorporates the CPU and the associated
caches. As before, Ruby runs under its own clock domain.
The clock period of all domains are pre-computed, such that no virtual
functions or multiplications are needed when calling
clockPeriod. Instead, the clock period is pre-computed when any
changes occur. For this to be possible, each clock domain tracks its
children.
2013-06-27 11:49:49 +02:00
|
|
|
mem_cntrl = RubyMemoryControl(
|
|
|
|
clk_domain = ruby_system.memctrl_clk_domain,
|
|
|
|
version = i,
|
|
|
|
ruby_system = ruby_system)
|
2010-01-30 05:29:27 +01:00
|
|
|
|
|
|
|
dir_size = MemorySize('0B')
|
|
|
|
dir_size.value = mem_module_size
|
|
|
|
|
|
|
|
dir_cntrl = Directory_Controller(version = i,
|
|
|
|
directory = \
|
|
|
|
RubyDirectoryMemory(version = i,
|
2012-01-23 18:33:52 +01:00
|
|
|
size = dir_size,
|
|
|
|
use_map =
|
|
|
|
options.use_map),
|
2011-07-01 02:49:26 +02:00
|
|
|
memBuffer = mem_cntrl,
|
2013-08-20 18:32:31 +02:00
|
|
|
transitions_per_cycle = options.ports,
|
2011-07-01 02:49:26 +02:00
|
|
|
ruby_system = ruby_system)
|
2010-01-30 05:29:27 +01:00
|
|
|
|
2013-01-14 17:05:14 +01:00
|
|
|
exec("ruby_system.dir_cntrl%d = dir_cntrl" % i)
|
2010-01-30 05:29:27 +01:00
|
|
|
dir_cntrl_nodes.append(dir_cntrl)
|
|
|
|
|
2014-09-01 23:55:47 +02:00
|
|
|
# Connect the directory controllers and the network
|
|
|
|
dir_cntrl.requestToDir = ruby_system.network.master
|
|
|
|
dir_cntrl.responseToDir = ruby_system.network.master
|
|
|
|
dir_cntrl.responseFromDir = ruby_system.network.slave
|
|
|
|
|
|
|
|
|
2012-04-05 18:09:19 +02:00
|
|
|
for i, dma_port in enumerate(dma_ports):
|
2010-01-30 05:29:27 +01:00
|
|
|
# Create the Ruby objects associated with the dma controller
|
|
|
|
dma_seq = DMASequencer(version = i,
|
2011-07-26 19:20:22 +02:00
|
|
|
ruby_system = ruby_system)
|
2014-01-04 07:03:33 +01:00
|
|
|
|
2010-01-30 05:29:27 +01:00
|
|
|
dma_cntrl = DMA_Controller(version = i,
|
2011-07-26 19:20:22 +02:00
|
|
|
dma_sequencer = dma_seq,
|
2013-08-20 18:32:31 +02:00
|
|
|
transitions_per_cycle = options.ports,
|
2011-07-26 19:20:22 +02:00
|
|
|
ruby_system = ruby_system)
|
2010-01-30 05:29:27 +01:00
|
|
|
|
2013-01-14 17:05:14 +01:00
|
|
|
exec("ruby_system.dma_cntrl%d = dma_cntrl" % i)
|
|
|
|
exec("ruby_system.dma_cntrl%d.dma_sequencer.slave = dma_port" % i)
|
2010-01-30 05:29:27 +01:00
|
|
|
dma_cntrl_nodes.append(dma_cntrl)
|
2011-04-29 02:18:14 +02:00
|
|
|
|
2014-09-01 23:55:47 +02:00
|
|
|
# Connect the dma controller to the network
|
|
|
|
dma_cntrl.responseFromDir = ruby_system.network.master
|
|
|
|
dma_cntrl.requestToDir = ruby_system.network.slave
|
|
|
|
|
|
|
|
|
2010-01-30 05:29:27 +01:00
|
|
|
all_cntrls = l1_cntrl_nodes + \
|
|
|
|
l2_cntrl_nodes + \
|
|
|
|
dir_cntrl_nodes + \
|
|
|
|
dma_cntrl_nodes
|
|
|
|
|
2012-07-11 07:51:53 +02:00
|
|
|
topology = create_topology(all_cntrls, options)
|
|
|
|
|
|
|
|
return (cpu_sequencers, dir_cntrl_nodes, topology)
|