Merge pull request #329 from haberman/benchmark-ads-descriptor

Added benchmark for loading ads descriptor.

WORKSPACE

@@ -1,6 +1,7 @@
 workspace(name = "upb")
 
 load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
+load("@bazel_tools//tools/build_defs/repo:git.bzl", "new_git_repository")
 load("//bazel:workspace_deps.bzl", "upb_deps")
 
 upb_deps()
@@ -37,3 +38,11 @@ http_archive(
     strip_prefix = "benchmark-16703ff83c1ae6d53e5155df3bb3ab0bc96083be",
     sha256 = "59f918c8ccd4d74b6ac43484467b500f1d64b40cc1010daa055375b322a43ba3",
 )
+
+new_git_repository(
+    name = "com_google_googleapis",
+    remote = "https://github.com/googleapis/googleapis.git",
+    branch = "master",
+    build_file = "//benchmarks:BUILD.googleapis",
+    patch_cmds = ["find google -type f -name BUILD.bazel -delete"],
+)

benchmarks/BUILD

@@ -21,6 +21,11 @@ upb_proto_reflection_library(
     deps = [":benchmark_descriptor_proto"],
 )
 
+upb_proto_reflection_library(
+    name = "ads_upb_proto_reflection",
+    deps = ["@com_google_googleapis//:ads_proto"],
+)
+
 cc_proto_library(
     name = "benchmark_descriptor_cc_proto",
     deps = [":benchmark_descriptor_proto"],
@@ -45,6 +50,7 @@ cc_binary(
         ":benchmark_descriptor_sv_cc_proto",
         ":benchmark_descriptor_upb_proto",
         ":benchmark_descriptor_upb_proto_reflection",
+        ":ads_upb_proto_reflection",
         "//:descriptor_upb_proto",
         "//:reflection",
         "@com_github_google_benchmark//:benchmark_main",

benchmarks/BUILD.googleapis

@@ -0,0 +1,29 @@
+load(
+    "@rules_proto//proto:defs.bzl",
+    "proto_library",
+)
+
+proto_library(
+    name = "ads_proto",
+    srcs = glob([
+        "google/ads/googleads/v5/**/*.proto",
+        "google/api/**/*.proto",
+        "google/rpc/**/*.proto",
+        "google/longrunning/**/*.proto",
+        "google/logging/**/*.proto",
+    ]),
+    #srcs = ["google/ads/googleads/v5/services/google_ads_service.proto"],
+    visibility = ["//visibility:public"],
+    deps = [
+       "@com_google_protobuf//:any_proto",
+       "@com_google_protobuf//:empty_proto",
+       "@com_google_protobuf//:descriptor_proto",
+       "@com_google_protobuf//:field_mask_proto",
+       "@com_google_protobuf//:duration_proto",
+       "@com_google_protobuf//:timestamp_proto",
+       "@com_google_protobuf//:struct_proto",
+       "@com_google_protobuf//:api_proto",
+       "@com_google_protobuf//:type_proto",
+       "@com_google_protobuf//:wrappers_proto",
+    ],
+)

benchmarks/benchmark.cc

@@ -9,9 +9,10 @@
 #include "benchmarks/descriptor_sv.pb.h"
 
 // For for benchmarks of building descriptors.
-#include "google/protobuf/descriptor.upb.h"
+#include "google/ads/googleads/v5/services/google_ads_service.upbdefs.h"
 #include "google/protobuf/descriptor.pb.h"
-
+#include "google/protobuf/descriptor.upb.h"
+#include "google/protobuf/descriptor.upbdefs.h"
 #include "upb/def.hpp"
 
 upb_strview descriptor = benchmarks_descriptor_proto_upbdefinit.descriptor;
@@ -20,6 +21,16 @@ namespace protobuf = ::google::protobuf;
 /* A buffer big enough to parse descriptor.proto without going to heap. */
 char buf[65535];
 
+void CollectFileDescriptors(const upb_def_init* file,
+                            std::vector<upb_strview>& serialized_files,
+                            std::unordered_set<const upb_def_init*>& seen) {
+  if (!seen.insert(file).second) return;
+  for (upb_def_init **deps = file->deps; *deps; deps++) {
+    CollectFileDescriptors(*deps, serialized_files, seen);
+  }
+  serialized_files.push_back(file->descriptor);
+}
+
 static void BM_ArenaOneAlloc(benchmark::State& state) {
   for (auto _ : state) {
     upb_arena* arena = upb_arena_new();
@@ -39,22 +50,28 @@ static void BM_ArenaInitialBlockOneAlloc(benchmark::State& state) {
 BENCHMARK(BM_ArenaInitialBlockOneAlloc);
 
 static void BM_LoadDescriptor_Upb(benchmark::State& state) {
+  size_t bytes_per_iter;
   for (auto _ : state) {
     upb::SymbolTable symtab;
-    upb::Arena arena;
-    google_protobuf_FileDescriptorProto* file_proto =
-        google_protobuf_FileDescriptorProto_parse(descriptor.data,
-                                                  descriptor.size, arena.ptr());
-    upb::FileDefPtr file_def = symtab.AddFile(file_proto, NULL);
-    if (!file_def) {
-      printf("Failed to add file.\n");
-      exit(1);
-    }
+    google_protobuf_DescriptorProto_getmsgdef(symtab.ptr());
+    bytes_per_iter = _upb_symtab_bytesloaded(symtab.ptr());
   }
   state.SetBytesProcessed(state.iterations() * descriptor.size);
 }
 BENCHMARK(BM_LoadDescriptor_Upb);
 
+static void BM_LoadAdsDescriptor_Upb(benchmark::State& state) {
+  size_t bytes_per_iter;
+  for (auto _ : state) {
+    upb::SymbolTable symtab;
+    google_ads_googleads_v5_services_SearchGoogleAdsRequest_getmsgdef(
+        symtab.ptr());
+    bytes_per_iter = _upb_symtab_bytesloaded(symtab.ptr());
+  }
+  state.SetBytesProcessed(state.iterations() * bytes_per_iter);
+}
+BENCHMARK(BM_LoadAdsDescriptor_Upb);
+
 static void BM_LoadDescriptor_Proto2(benchmark::State& state) {
   for (auto _ : state) {
     protobuf::Arena arena;
@@ -73,6 +90,35 @@ static void BM_LoadDescriptor_Proto2(benchmark::State& state) {
 }
 BENCHMARK(BM_LoadDescriptor_Proto2);
 
+static void BM_LoadAdsDescriptor_Proto2(benchmark::State& state) {
+  extern upb_def_init google_ads_googleads_v5_services_google_ads_service_proto_upbdefinit;
+  std::vector<upb_strview> serialized_files;
+  std::unordered_set<const upb_def_init*> seen_files;
+  CollectFileDescriptors(
+      &google_ads_googleads_v5_services_google_ads_service_proto_upbdefinit,
+      serialized_files, seen_files);
+  size_t bytes_per_iter;
+  for (auto _ : state) {
+    bytes_per_iter = 0;
+    protobuf::Arena arena;
+    protobuf::DescriptorPool pool;
+    for (auto file : serialized_files) {
+      protobuf::StringPiece input(file.data, file.size);
+      auto proto = protobuf::Arena::CreateMessage<protobuf::FileDescriptorProto>(
+          &arena);
+      bool ok = proto->ParseFrom<protobuf::MessageLite::kMergePartial>(input) &&
+                pool.BuildFile(*proto) != nullptr;
+      if (!ok) {
+        printf("Failed to add file.\n");
+        exit(1);
+      }
+      bytes_per_iter += input.size();
+    }
+  }
+  state.SetBytesProcessed(state.iterations() * bytes_per_iter);
+}
+BENCHMARK(BM_LoadAdsDescriptor_Proto2);
+
 static void BM_Parse_Upb_FileDesc_WithArena(benchmark::State& state) {
   size_t bytes = 0;
   for (auto _ : state) {

cmake/make_cmakelists.py

@@ -177,6 +177,9 @@ class WorkspaceFileFunctions(object):
   def git_repository(self, **kwargs):
     pass
 
+  def new_git_repository(self, **kwargs):
+    pass
+
   def bazel_version_repository(self, **kwargs):
     pass
 

upb/def.c

@@ -118,6 +118,7 @@ struct upb_symtab {
   upb_arena *arena;
   upb_strtable syms;  /* full_name -> packed def ptr */
   upb_strtable files;  /* file_name -> upb_filedef* */
+  size_t bytes_loaded;
 };
 
 /* Inside a symtab we store tagged pointers to specific def types. */
@@ -2084,6 +2085,7 @@ upb_symtab *upb_symtab_new(void) {
   }
 
   s->arena = upb_arena_new();
+  s->bytes_loaded = 0;
   alloc = upb_arena_alloc(s->arena);
 
   if (!upb_strtable_init2(&s->syms, UPB_CTYPE_CONSTPTR, 32, alloc) ||
@@ -2187,6 +2189,7 @@ bool _upb_symtab_loaddefinit(upb_symtab *s, const upb_def_init *init) {
 
   file = google_protobuf_FileDescriptorProto_parse(
       init->descriptor.data, init->descriptor.size, arena);
+  s->bytes_loaded += init->descriptor.size;
 
   if (!file) {
     upb_status_seterrf(
@@ -2209,5 +2212,9 @@ err:
   return false;
 }
 
+size_t _upb_symtab_bytesloaded(const upb_symtab *s) {
+  return s->bytes_loaded;
+}
+
 #undef CHK
 #undef CHK_OOM

upb/def.h

@@ -293,6 +293,7 @@ int upb_symtab_filecount(const upb_symtab *s);
 const upb_filedef *upb_symtab_addfile(
     upb_symtab *s, const google_protobuf_FileDescriptorProto *file,
     upb_status *status);
+size_t _upb_symtab_bytesloaded(const upb_symtab *s);
 
 /* For generated code only: loads a generated descriptor. */
 typedef struct upb_def_init {