From: Steinar H. Gunderson <sgunderson@bigfoot.com>
Date: Mon, 15 Dec 2014 01:08:17 +0000 (+0100)
Subject: Drop the arena; BPFENs are bounded by 32 bytes anyway, and the typical is something... 
X-Git-Url: https://git.sesse.net/?p=remoteglot-book;a=commitdiff_plain;h=33d7fcf900b9c456aea2fc88900fe3c7dd4344e8

Drop the arena; BPFENs are bounded by 32 bytes anyway, and the typical is something like 20, so using those eight bytes for a pointer is a rather slight memory gain. This speeds up the sorting by a few percent due to less memory chasing.
---

diff --git a/Makefile b/Makefile
index 0e95cd4..1dc36ce 100644
--- a/Makefile
+++ b/Makefile
@@ -4,11 +4,11 @@ PROTOC=protoc
 
 all: binloader binlookup binmerger
 
-binloader: binloader.o hash.o arena.o count.pb.o
+binloader: binloader.o hash.o count.pb.o
 binmerger: binmerger.o merge_count.o count.pb.o
 binlookup: binlookup.o hash.o merge_count.o count.pb.o
 
-binloader.o: binloader.cpp arena.cpp count.pb.h
+binloader.o: binloader.cpp count.pb.h
 binmerger.o: binmerger.cpp merge_count.h count.pb.h
 binlookup.o: binlookup.cpp merge_count.h count.pb.h
 
@@ -17,4 +17,4 @@ binlookup.o: binlookup.cpp merge_count.h count.pb.h
 
 .PHONY: clean
 clean:
-	$(RM) binloader binlookup binmerger binloader.o binmerger.o binlookup.o hash.o arena.o count.pb.o count.pb.h count.pb.cc
+	$(RM) binloader binlookup binmerger binloader.o binmerger.o binlookup.o hash.o count.pb.o count.pb.h count.pb.cc
diff --git a/arena.cpp b/arena.cpp
deleted file mode 100644
index 048e6db..0000000
--- a/arena.cpp
+++ /dev/null
@@ -1,33 +0,0 @@
-#include <stdlib.h>
-#include <assert.h>
-#include "arena.h"
-
-Arena::Arena() : first(NULL) {}
-
-Arena::~Arena()
-{
-	Block *next;
-	for (Block *b = first; b != NULL; b = next) {
-		delete[] b->memory;
-
-		next = b->next;
-		delete b;
-	}
-}
-
-char *Arena::alloc(size_t bytes)
-{
-	assert(bytes < BLOCK_SIZE);  // Can fix, but we don't need to.
-
-	if (first == NULL || first->used + bytes > BLOCK_SIZE) {
-		Block *b = new Block;
-		b->memory = new char[BLOCK_SIZE];
-		b->used = 0;
-		b->next = first;
-		first = b;
-	}
-
-	char *ret = first->memory + first->used;
-	first->used += bytes;
-	return ret;
-}
diff --git a/arena.h b/arena.h
deleted file mode 100644
index 5d254dd..0000000
--- a/arena.h
+++ /dev/null
@@ -1,24 +0,0 @@
-#ifndef _ARENA_H
-#define _ARENA_H
-
-// A simple arena for allocating lots of short strings.
-
-class Arena {
-public:
-	Arena();
-	~Arena();
-
-	char *alloc(size_t bytes);
-
-private:
-	static constexpr size_t BLOCK_SIZE = 1048576;
-
-	struct Block {
-		char *memory;
-		size_t used;
-		Block *next;
-	};
-	Block *first;
-};
-
-#endif  // _ARENA_H
diff --git a/binloader.cpp b/binloader.cpp
index 3877ffc..acf14d7 100644
--- a/binloader.cpp
+++ b/binloader.cpp
@@ -12,7 +12,6 @@
 #include <unordered_set>
 #include <string.h>
 #include "count.pb.h"
-#include "arena.h"
 #include "hash.h"
 
 #define DUMMY_TIMESTAMP 32503680000
@@ -42,7 +41,7 @@ static inline int memcmp_different_len(const void *s1, size_t n1, const void *s2
 
 enum Result { WHITE = 0, DRAW, BLACK };
 struct Element {
-	char *bpfen;  // includes prev_board_hash
+	char bpfen[32];  // includes prev_board_hash
 	int bpfen_len;
 	char move[8];   // Na1xc3+
 	Result result;
@@ -59,7 +58,6 @@ struct Element {
 
 struct ShardData {
 	vector<Element> elems;
-	unique_ptr<Arena> arena;  // Used to allocate bpfen.
 	int num_written_subshards = 0;
 };
 
@@ -119,7 +117,6 @@ void write_subshard(const char *basename, ShardData* shard, int bucket)
 	mtbl_writer_destroy(&mtbl);
 
 	shard->elems.clear();
-	shard->arena.reset(new Arena);
 }
 
 int main(int argc, char **argv)
@@ -132,7 +129,6 @@ int main(int argc, char **argv)
 
 	for (int i = 0; i < num_buckets; ++i) {
 		shards[i].elems.reserve(num_pos_per_subshard);
-		shards[i].arena.reset(new Arena);
 	}
 
 	size_t num_elems = 0;
@@ -154,6 +150,7 @@ int main(int argc, char **argv)
 			if (bpfen_len == -1) {
 				break;
 			}
+			assert(bpfen_len <= 32);
 			if (bpfen_len >= int(sizeof(bpfen))) {
 				fprintf(stderr, "Overlong BPFEN (%d bytes)\n", bpfen_len);
 		//		exit(1);
@@ -226,7 +223,6 @@ int main(int argc, char **argv)
 
 			int bucket = hash_key_to_bucket(bpfen, bpfen_len, num_buckets);
 			Element e;
-			e.bpfen = shards[bucket].arena->alloc(bpfen_len);
 			memcpy(e.bpfen, bpfen, bpfen_len);
 			e.bpfen_len = bpfen_len;
 			strcpy(e.move, move);