Merge branch 'master' into noc_pack_part

Author: soheilshahrouz

.github/scripts/install_dependencies.sh

@@ -29,6 +29,7 @@ sudo apt install -y \
   libncurses5-dev \
   libx11-dev \
   libxft-dev \
+  libxml2-utils \
   libxml++2.6-dev \
   libreadline-dev \
   tcllib \

.github/workflows/test.yml

@@ -197,6 +197,11 @@ jobs:
             params: '-DVTR_ASSERT_LEVEL=3 -DWITH_BLIFEXPLORER=on -DVPR_USE_EZGL=off',
             suite: 'vtr_reg_basic'
           },
+          {
+            name: 'Basic with CAPNPROTO disabled',
+            params: '-DVTR_ASSERT_LEVEL=3 -DWITH_BLIFEXPLORER=on -DVTR_ENABLE_CAPNPROTO=off',
+            suite: 'vtr_reg_basic'
+          },
           {
             name: 'Basic with VTR_ENABLE_DEBUG_LOGGING',
             params: '-DVTR_ASSERT_LEVEL=3 -DWITH_BLIFEXPLORER=on -DVTR_ENABLE_DEBUG_LOGGING=on',

CMakeLists.txt

@@ -418,11 +418,13 @@ if(${WITH_PARMYS}) # define cmake params to compile Yosys
         set(MAKE_PROGRAM "make")
     endif()
 
-    if(NOT DEFINED "${CMAKE_BUILD_PARALLEL_LEVEL}")
-        set(CUSTOM_BUILD_PARALLEL_LEVEL 16)
-    else()
-        set(CUSTOM_BUILD_PARALLEL_LEVEL "${CMAKE_BUILD_PARALLEL_LEVEL}")
-    endif()
+    # Commented out since a make file should not call another make command with
+    # threads. It should pass this information from the parent automatically.
+    # if(NOT DEFINED "${CMAKE_BUILD_PARALLEL_LEVEL}")
+    #     set(CUSTOM_BUILD_PARALLEL_LEVEL 16)
+    # else()
+    #     set(CUSTOM_BUILD_PARALLEL_LEVEL "${CMAKE_BUILD_PARALLEL_LEVEL}")
+    # endif()
     add_subdirectory(yosys)
 endif()
 

blifexplorer/CMakeLists.txt

@@ -2,7 +2,7 @@ cmake_minimum_required(VERSION 3.16)
 
 project("blifexplorer")
 
-set(CMAKE_CXX_STANDARD 14)
+set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_STANDARD_REQUIRED ON)
 set(CMAKE_CXX_EXTENSIONS OFF)
 

doc/src/vpr/command_line_usage.rst

@@ -1298,6 +1298,13 @@ The following options are only valid when the router is in timing-driven mode (t
 
     **Default:** ``1.2``
 
+.. option:: --router_profiler_astar_fac <float>
+    Controls the directedness of the timing-driven router's exploration when doing router delay profiling of an architecture.
+    The router delay profiling step is currently used to calculate the place delay matrix lookup.
+    Values between 1 and 2 are resonable; higher values trade some quality for reduced run-time.
+
+    **Default:** ``1.2``
+
 .. option:: --max_criticality <float>
 
     Sets the maximum fraction of routing cost that can come from delay (vs. coming from routability) for any net.

doc/src/vtr/benchmarks.rst

@@ -102,7 +102,7 @@ These designs use many precisions including binary, different fixed point types
     proxy               Proxy/synthetic benchmarks
     =================   ======================================
 
-The VTR benchmarks are provided as Verilog (enabling full flexibility to modify and change how the designs are implemented) under: ::
+The Koios benchmarks are provided as Verilog (enabling full flexibility to modify and change how the designs are implemented) under: ::
 
     $VTR_ROOT/vtr_flow/benchmarks/verilog/koios
 
@@ -207,4 +207,4 @@ real application domains. On the other hand, MLP benchmarks include modules that
 and move data. Pre-synthesized netlists for the synthetic benchmarks are added to VTR project, but MLP netlists should
 be downloaded separately.
 
-.. note:: The NoC MLP benchmarks are not included with the VTR release (due to their size). However they can be downloaded and extracted by running ``make get_noc_mlp_benchmarks`` from the root of the VTR tree.  They can also be `downloaded manually <https://www.eecg.utoronto.ca/~vaughn/titan/>`_.
+.. note:: The NoC MLP benchmarks are not included with the VTR release (due to their size). However they can be downloaded and extracted by running ``make get_noc_mlp_benchmarks`` from the root of the VTR tree.  They can also be `downloaded manually <https://www.eecg.utoronto.ca/~vaughn/titan/>`_.

libs/EXTERNAL/libargparse/src/argparse.cpp

@@ -1,10 +1,10 @@
 #include <algorithm>
-#include <array>
 #include <list>
 #include <cassert>
 #include <string>
 #include <set>
 #include <limits>
+#include <utility>
 
 #include "argparse.hpp"
 #include "argparse_util.hpp"
@@ -16,16 +16,16 @@ namespace argparse {
      * ArgumentParser
      */
 
-    ArgumentParser::ArgumentParser(std::string prog_name, std::string description_str, std::ostream& os)
-        : description_(description_str)
+    ArgumentParser::ArgumentParser(const std::string& prog_name, std::string description_str, std::ostream& os)
+        : description_(std::move(description_str))
         , formatter_(new DefaultFormatter())
         , os_(os)
         {
         prog(prog_name);
         argument_groups_.push_back(ArgumentGroup("arguments"));
     }
 
-    ArgumentParser& ArgumentParser::prog(std::string prog_name, bool basename_only) {
+    ArgumentParser& ArgumentParser::prog(const std::string& prog_name, bool basename_only) {
         if (basename_only) {
             prog_ = basename(prog_name);
         } else {
@@ -35,17 +35,17 @@ namespace argparse {
     }
 
     ArgumentParser& ArgumentParser::version(std::string version_str) {
-        version_ = version_str;
+        version_ = std::move(version_str);
         return *this;
     }
 
     ArgumentParser& ArgumentParser::epilog(std::string epilog_str) {
-        epilog_ = epilog_str;
+        epilog_ = std::move(epilog_str);
         return *this;
     }
 
     ArgumentGroup& ArgumentParser::add_argument_group(std::string description_str) {
-        argument_groups_.push_back(ArgumentGroup(description_str));
+        argument_groups_.push_back(ArgumentGroup(std::move(description_str)));
         return argument_groups_[argument_groups_.size() - 1];
     }
 
@@ -72,7 +72,7 @@ namespace argparse {
     void ArgumentParser::parse_args_throw(int argc, const char* const* argv) {
         std::vector<std::string> arg_strs;
         for (int i = 1; i < argc; ++i) {
-            arg_strs.push_back(argv[i]);
+            arg_strs.emplace_back(argv[i]);
         }
 
         parse_args_throw(arg_strs);
@@ -241,7 +241,7 @@ namespace argparse {
                     } else if (arg->nargs() == '+' || arg->nargs() == '*') {
                         if (arg->nargs() == '+') {
                             assert(nargs_read >= 1);
-                            assert(values.size() >= 1);
+                            assert(!values.empty());
                         }
 
                         for (const auto& value : values) {
@@ -410,11 +410,11 @@ namespace argparse {
      * ArgumentGroup
      */
     ArgumentGroup::ArgumentGroup(std::string name_str)
-        : name_(name_str)
+        : name_(std::move(name_str))
         {}
 
     ArgumentGroup& ArgumentGroup::epilog(std::string str) {
-        epilog_ = str;
+        epilog_ = std::move(str);
         return *this;
     }
     std::string ArgumentGroup::name() const { return name_; }
@@ -425,10 +425,10 @@ namespace argparse {
      * Argument
      */
     Argument::Argument(std::string long_opt, std::string short_opt)
-        : long_opt_(long_opt)
-        , short_opt_(short_opt) {
+        : long_opt_(std::move(long_opt))
+        , short_opt_(std::move(short_opt)) {
 
-        if (long_opt_.size() < 1) {
+        if (long_opt_.empty()) {
             throw ArgParseError("Argument must be at least one character long");
         }
 
@@ -445,7 +445,7 @@ namespace argparse {
     }
 
     Argument& Argument::help(std::string help_str) {
-        help_ = help_str;
+        help_ = std::move(help_str);
         return *this;
     }
 
@@ -476,12 +476,12 @@ namespace argparse {
     }
 
     Argument& Argument::metavar(std::string metavar_str) {
-        metavar_ = metavar_str;
+        metavar_ = std::move(metavar_str);
         return *this;
     }
 
     Argument& Argument::choices(std::vector<std::string> choice_values) {
-        choices_ = choice_values;
+        choices_ = std::move(choice_values);
         return *this;
     }
 
@@ -536,7 +536,7 @@ namespace argparse {
     }
 
     Argument& Argument::group_name(std::string grp) {
-        group_name_ = grp;
+        group_name_ = std::move(grp);
         return *this;
     }
 

libs/EXTERNAL/libargparse/src/argparse.hpp

@@ -34,10 +34,10 @@ namespace argparse {
     class ArgumentParser {
         public:
             //Initializes an argument parser
-            ArgumentParser(std::string prog_name, std::string description_str=std::string(), std::ostream& os=std::cout);
+            ArgumentParser(const std::string& prog_name, std::string description_str=std::string(), std::ostream& os=std::cout);
 
             //Overrides the program name
-            ArgumentParser& prog(std::string prog, bool basename_only=true);
+            ArgumentParser& prog(const std::string& prog, bool basename_only=true);
 
             //Sets the program version
             ArgumentParser& version(std::string version);

libs/EXTERNAL/libcatch2

@@ -22,9 +22,13 @@ target_link_libraries(libarchfpga
                         libvtrutil
                         libpugixml
                         libpugiutil
-                        libvtrcapnproto
 )
 
+if(${VTR_ENABLE_CAPNPROTO})
+    target_link_libraries(libarchfpga libvtrcapnproto)
+    target_compile_definitions(libarchfpga PRIVATE VTR_ENABLE_CAPNPROTO)
+endif()
+
 # Using filesystem library requires additional compiler/linker options for GNU implementation prior to 9.1
 # and LLVM implementation prior to LLVM 9.0;
 if(CMAKE_C_COMPILER_ID STREQUAL "GNU")

libs/libarchfpga/CMakeLists.txt

@@ -1,3 +1,10 @@
+
+
+#include "read_fpga_interchange_arch.h"
+#include "vtr_error.h"
+
+#ifdef VTR_ENABLE_CAPNPROTO
+
 #include <algorithm>
 #include <kj/std/iostream.h>
 #include <limits>
@@ -21,7 +28,6 @@
 #include "arch_util.h"
 #include "arch_types.h"
 
-#include "read_fpga_interchange_arch.h"
 
 /*
  * FPGA Interchange Device frontend
@@ -2497,11 +2503,14 @@ struct ArchReader {
     }
 };
 
+#endif  // VTR_ENABLE_CAPNPROTO
+
 void FPGAInterchangeReadArch(const char* FPGAInterchangeDeviceFile,
                              const bool /*timing_enabled*/,
                              t_arch* arch,
                              std::vector<t_physical_tile_type>& PhysicalTileTypes,
                              std::vector<t_logical_block_type>& LogicalBlockTypes) {
+#ifdef VTR_ENABLE_CAPNPROTO
     // Decompress GZipped capnproto device file
     gzFile file = gzopen(FPGAInterchangeDeviceFile, "r");
     VTR_ASSERT(file != Z_NULL);
@@ -2542,4 +2551,12 @@ void FPGAInterchangeReadArch(const char* FPGAInterchangeDeviceFile,
 
     ArchReader reader(arch, device_reader, FPGAInterchangeDeviceFile, PhysicalTileTypes, LogicalBlockTypes);
     reader.read_arch();
+#else   // VTR_ENABLE_CAPNPROTO
+    // If CAPNPROTO is disabled, throw an error.
+    (void)FPGAInterchangeDeviceFile;
+    (void)arch;
+    (void)PhysicalTileTypes;
+    (void)LogicalBlockTypes;
+    throw vtr::VtrError("Unable to read FPGA interchange if CAPNPROTO is not enabled", __FILE__, __LINE__);
+#endif  // VTR_ENABLE_CAPNPROTO
 }

libs/libarchfpga/src/read_fpga_interchange_arch.cpp

@@ -3,13 +3,17 @@
 
 #include "arch_types.h"
 
+#ifdef VTR_ENABLE_CAPNPROTO
+
 #include "DeviceResources.capnp.h"
 #include "LogicalNetlist.capnp.h"
 #include "capnp/serialize.h"
 #include "capnp/serialize-packed.h"
 #include <fcntl.h>
 #include <unistd.h>
 
+#endif  // VTR_ENABLE_CAPNPROTO
+
 #ifdef __cplusplus
 extern "C" {
 #endif

libs/libarchfpga/src/read_fpga_interchange_arch.h

@@ -155,6 +155,13 @@ class edge_sort_iterator {
     using pointer = edge_swapper*;
     using difference_type = ssize_t;
 
+    // In order for this class to be used as an iterator within the std library,
+    // it needs to "act" like a pointer. One thing that it should do is that a
+    // const variable of this type should be de-referenceable. Therefore, this
+    // method should be const method; however, this requires modifying the class
+    // and may yield worst performance. For now the std::stable_sort allows this
+    // but in the future it may not. If this breaks, this is why.
+    // See issue #2517 and PR #2522
     edge_swapper& operator*() {
         return this->swapper_;
     }
@@ -419,7 +426,7 @@ size_t t_rr_graph_storage::count_rr_switches(
     // values.
     //
     // This sort is safe to do because partition_edges() has not been invoked yet.
-    std::sort(
+    std::stable_sort(
         edge_sort_iterator(this, 0),
         edge_sort_iterator(this, edge_dest_node_.size()),
         edge_compare_dest_node());
@@ -527,7 +534,7 @@ void t_rr_graph_storage::partition_edges(const vtr::vector<RRSwitchId, t_rr_swit
     //    by assign_first_edges()
     //  - Edges within a source node have the configurable edges before the
     //    non-configurable edges.
-    std::sort(
+    std::stable_sort(
         edge_sort_iterator(this, 0),
         edge_sort_iterator(this, edge_src_node_.size()),
         edge_compare_src_node_and_configurable_first(rr_switches));

libs/librrgraph/src/base/rr_graph_storage.cpp

@@ -336,6 +336,11 @@ class t_rr_graph_storage {
         return ret;
     }
 
+    /** @brief Get the source node for the specified edge. */
+    RRNodeId edge_src_node(const RREdgeId& edge) const {
+        return edge_src_node_[edge];
+    }
+
     /** @brief Get the destination node for the specified edge. */
     RRNodeId edge_sink_node(const RREdgeId& edge) const {
         return edge_dest_node_[edge];

libs/librrgraph/src/base/rr_graph_storage.h

@@ -321,6 +321,12 @@ class RRGraphView {
     inline short edge_switch(RRNodeId id, t_edge_size iedge) const {
         return node_storage_.edge_switch(id, iedge);
     }
+
+    /** @brief Get the source node for the specified edge. */
+    inline RRNodeId edge_src_node(const RREdgeId edge_id) const {
+        return node_storage_.edge_src_node(edge_id);
+    }
+
     /** @brief Get the destination node for the iedge'th edge from specified RRNodeId.
      *  This method should generally not be used, and instead first_edge and
      *  last_edge should be used.*/

libs/librrgraph/src/base/rr_graph_view.h

@@ -4,10 +4,11 @@
 #Figure out the git revision
 find_package(Git QUIET)
 if(GIT_FOUND)
-    exec_program(${GIT_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}
-                 ARGS describe --always --long --dirty
-                 OUTPUT_VARIABLE VTR_VCS_REVISION
-                 RETURN_VALUE GIT_DESCRIBE_RETURN_VALUE)
+    execute_process(COMMAND ${GIT_EXECUTABLE} describe --always --long --dirty
+                    WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
+                    OUTPUT_VARIABLE VTR_VCS_REVISION
+                    OUTPUT_STRIP_TRAILING_WHITESPACE
+                    RESULT_VARIABLE GIT_DESCRIBE_RETURN_VALUE)
 
     if(NOT GIT_DESCRIBE_RETURN_VALUE EQUAL 0)
         #Git describe failed, usually this means we
@@ -18,10 +19,12 @@ if(GIT_FOUND)
 
     #Call again with exclude to get the revision excluding any tags
     #(i.e. just the commit ID and dirty flag)
-    exec_program(${GIT_EXECUTABLE} ${CMAKE_CURRENT_SOURCE_DIR}
-                 ARGS describe --always --long --dirty --exclude '*'
-                 OUTPUT_VARIABLE VTR_VCS_REVISION_SHORT
-                 RETURN_VALUE GIT_DESCRIBE_RETURN_VALUE)
+    execute_process(COMMAND ${GIT_EXECUTABLE} describe --always --long --dirty --exclude '*'
+                    WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
+                    OUTPUT_VARIABLE VTR_VCS_REVISION_SHORT
+                    OUTPUT_STRIP_TRAILING_WHITESPACE
+                    RESULT_VARIABLE GIT_DESCRIBE_RETURN_VALUE)
+
     if(NOT GIT_DESCRIBE_RETURN_VALUE EQUAL 0)
         #Git describe failed, usually this means we
         #aren't in a git repo -- so don't set a VCS