Avoid allocation happiness in StringCache
Add nice s-box driven hash function, noticeble improvement (still StringCache alone) Optimize isWhite and drop extra check on first iter of lexWhiteSpace; almost 10% of speed gain Also dialdown -O switch of ldc2, as it makes no difference.
This commit is contained in:
parent
511c785675
commit
f76167f231
2
build.sh
2
build.sh
|
|
@ -1,3 +1,3 @@
|
|||
#dmd *.d std/d/*.d -release -inline -noboundscheck -O -w -wi -m64 -property -ofdscanner -L-lsqlite3 #-inline
|
||||
#dmd *.d std/d/*.d -g -m64 -w -wi -property -ofdscanner -L-lsqlite3 #-unittest
|
||||
ldc2 -O5 *.d std/d/*.d -of=dscanner -release -vectorize -m64
|
||||
ldc2 -O2 -release *.d std/d/*.d -of=dscanner -m64
|
||||
|
|
|
|||
175
std/d/lexer.d
175
std/d/lexer.d
|
|
@ -109,7 +109,7 @@
|
|||
module std.d.lexer;
|
||||
|
||||
import std.algorithm;
|
||||
import std.ascii;
|
||||
import std.ascii : isHexDigit;
|
||||
import std.conv;
|
||||
import std.d.entities;
|
||||
import std.datetime;
|
||||
|
|
@ -120,6 +120,23 @@ import std.traits;
|
|||
import std.regex;
|
||||
import std.container;
|
||||
|
||||
//these should be in std.ascii, using tabulation there is so 90-ies
|
||||
|
||||
bool isAsciiWhite(uint x)
|
||||
{
|
||||
x -= 0x09;
|
||||
enum high = 0x0D-0x09;
|
||||
enum space = 0x20-0x09;
|
||||
return x <= high || x == space;
|
||||
}
|
||||
|
||||
bool isAsciiDigit(uint x)
|
||||
{
|
||||
x -= '0';
|
||||
enum lastNum = '9' - '0';
|
||||
return x <= lastNum;
|
||||
}
|
||||
|
||||
public:
|
||||
|
||||
/**
|
||||
|
|
@ -432,13 +449,10 @@ private:
|
|||
current.column = column;
|
||||
current.value = null;
|
||||
|
||||
if (isWhite(currentElement()))
|
||||
{
|
||||
lexWhitespace();
|
||||
return;
|
||||
}
|
||||
|
||||
outer: switch (currentElement())
|
||||
auto e = currentElement();
|
||||
if (isAsciiWhite(e))
|
||||
return lexWhitespace();
|
||||
outer: switch (e)
|
||||
{
|
||||
// pragma(msg, generateCaseTrie(
|
||||
mixin(generateCaseTrie(
|
||||
|
|
@ -694,12 +708,17 @@ private:
|
|||
}
|
||||
|
||||
void lexWhitespace()
|
||||
in
|
||||
{
|
||||
current.type = TokenType.whitespace;
|
||||
while (!isEoF() && isWhite(currentElement()))
|
||||
assert (isAsciiWhite(currentElement()));
|
||||
}
|
||||
body
|
||||
{
|
||||
current.type = TokenType.whitespace;
|
||||
do
|
||||
{
|
||||
keepChar();
|
||||
}
|
||||
}while (!isEoF() && isAsciiWhite(currentElement()));
|
||||
if (config.iterStyle & IterationStyle.includeWhitespace)
|
||||
setTokenValue();
|
||||
}
|
||||
|
|
@ -791,7 +810,7 @@ private:
|
|||
{
|
||||
keepChar();
|
||||
}
|
||||
else if (isWhite(currentElement()) && (config.tokenStyle & TokenStyle.notEscaped))
|
||||
else if (isAsciiWhite(currentElement()) && (config.tokenStyle & TokenStyle.notEscaped))
|
||||
{
|
||||
keepChar();
|
||||
}
|
||||
|
|
@ -830,7 +849,7 @@ private:
|
|||
void lexNumber()
|
||||
in
|
||||
{
|
||||
assert(isDigit(cast(char) currentElement()) || currentElement() == '.');
|
||||
assert(isAsciiDigit(cast(char) currentElement()) || currentElement() == '.');
|
||||
}
|
||||
body
|
||||
{
|
||||
|
|
@ -2395,8 +2414,6 @@ struct StringCache
|
|||
{
|
||||
string get(const ubyte[] bytes)
|
||||
{
|
||||
|
||||
import std.stdio;
|
||||
size_t bucket;
|
||||
hash_t h;
|
||||
string* val = find(bytes, bucket, h);
|
||||
|
|
@ -2406,41 +2423,133 @@ struct StringCache
|
|||
}
|
||||
else
|
||||
{
|
||||
auto s = (cast(char[]) bytes).idup;
|
||||
index[bucket] ~= s;
|
||||
return s;
|
||||
auto s = putIntoCache(bytes);
|
||||
index[bucket] ~= s;
|
||||
return s;
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
private:
|
||||
|
||||
|
||||
import std.stdio;
|
||||
string* find(const ubyte[] data, out size_t bucket, out hash_t h)
|
||||
{
|
||||
h = hash(data);
|
||||
bucket = h % mapSize;
|
||||
bucket = h % mapSize;
|
||||
foreach (i; 0 .. index[bucket].length)
|
||||
{
|
||||
if (index[bucket][i] == data)
|
||||
{
|
||||
return &index[bucket][i];
|
||||
}
|
||||
}
|
||||
return null;
|
||||
}
|
||||
|
||||
static hash_t hash(const(ubyte)[] data)
|
||||
{
|
||||
hash_t h = 5381;
|
||||
int c;
|
||||
size_t i;
|
||||
while (i < data.length)
|
||||
{
|
||||
c = data[i++];
|
||||
h = ((h << 5) + h) ^ c;
|
||||
}
|
||||
return h;
|
||||
uint hash = 0;
|
||||
foreach ( b; data)
|
||||
{
|
||||
hash ^= sbox[b];
|
||||
hash *= 3;
|
||||
}
|
||||
return hash;
|
||||
}
|
||||
|
||||
immutable mapSize = 997;
|
||||
string[][mapSize] index;
|
||||
|
||||
enum mapSize = 2048;
|
||||
string[][mapSize] index;
|
||||
// leave some slack for alloctors/GC meta-data
|
||||
enum chunkSize = 16*1024 - size_t.sizeof*8;
|
||||
ubyte*[] chunkS;
|
||||
size_t next = chunkSize;
|
||||
|
||||
string putIntoCache(const ubyte[] data)
|
||||
{
|
||||
import core.memory;
|
||||
|
||||
if(next + data.length > chunkSize)
|
||||
{
|
||||
// avoid huge strings
|
||||
if(data.length > chunkSize/4)
|
||||
return (cast(char[])data).idup;
|
||||
chunkS ~= cast(ubyte*)GC.malloc(chunkSize,
|
||||
GC.BlkAttr.NO_SCAN | GC.BlkAttr.NO_INTERIOR);
|
||||
next = 0;
|
||||
}
|
||||
auto slice = chunkS[$-1][next..next+data.length];
|
||||
slice[] = data[];
|
||||
next += data.length;
|
||||
return cast(string)slice;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
immutable uint[] sbox = [
|
||||
0xF53E1837, 0x5F14C86B, 0x9EE3964C, 0xFA796D53,
|
||||
0x32223FC3, 0x4D82BC98, 0xA0C7FA62, 0x63E2C982,
|
||||
0x24994A5B, 0x1ECE7BEE, 0x292B38EF, 0xD5CD4E56,
|
||||
0x514F4303, 0x7BE12B83, 0x7192F195, 0x82DC7300,
|
||||
0x084380B4, 0x480B55D3, 0x5F430471, 0x13F75991,
|
||||
0x3F9CF22C, 0x2FE0907A, 0xFD8E1E69, 0x7B1D5DE8,
|
||||
0xD575A85C, 0xAD01C50A, 0x7EE00737, 0x3CE981E8,
|
||||
0x0E447EFA, 0x23089DD6, 0xB59F149F, 0x13600EC7,
|
||||
0xE802C8E6, 0x670921E4, 0x7207EFF0, 0xE74761B0,
|
||||
0x69035234, 0xBFA40F19, 0xF63651A0, 0x29E64C26,
|
||||
0x1F98CCA7, 0xD957007E, 0xE71DDC75, 0x3E729595,
|
||||
0x7580B7CC, 0xD7FAF60B, 0x92484323, 0xA44113EB,
|
||||
0xE4CBDE08, 0x346827C9, 0x3CF32AFA, 0x0B29BCF1,
|
||||
0x6E29F7DF, 0xB01E71CB, 0x3BFBC0D1, 0x62EDC5B8,
|
||||
0xB7DE789A, 0xA4748EC9, 0xE17A4C4F, 0x67E5BD03,
|
||||
0xF3B33D1A, 0x97D8D3E9, 0x09121BC0, 0x347B2D2C,
|
||||
0x79A1913C, 0x504172DE, 0x7F1F8483, 0x13AC3CF6,
|
||||
0x7A2094DB, 0xC778FA12, 0xADF7469F, 0x21786B7B,
|
||||
0x71A445D0, 0xA8896C1B, 0x656F62FB, 0x83A059B3,
|
||||
0x972DFE6E, 0x4122000C, 0x97D9DA19, 0x17D5947B,
|
||||
0xB1AFFD0C, 0x6EF83B97, 0xAF7F780B, 0x4613138A,
|
||||
0x7C3E73A6, 0xCF15E03D, 0x41576322, 0x672DF292,
|
||||
0xB658588D, 0x33EBEFA9, 0x938CBF06, 0x06B67381,
|
||||
0x07F192C6, 0x2BDA5855, 0x348EE0E8, 0x19DBB6E3,
|
||||
0x3222184B, 0xB69D5DBA, 0x7E760B88, 0xAF4D8154,
|
||||
0x007A51AD, 0x35112500, 0xC9CD2D7D, 0x4F4FB761,
|
||||
0x694772E3, 0x694C8351, 0x4A7E3AF5, 0x67D65CE1,
|
||||
0x9287DE92, 0x2518DB3C, 0x8CB4EC06, 0xD154D38F,
|
||||
0xE19A26BB, 0x295EE439, 0xC50A1104, 0x2153C6A7,
|
||||
0x82366656, 0x0713BC2F, 0x6462215A, 0x21D9BFCE,
|
||||
0xBA8EACE6, 0xAE2DF4C1, 0x2A8D5E80, 0x3F7E52D1,
|
||||
0x29359399, 0xFEA1D19C, 0x18879313, 0x455AFA81,
|
||||
0xFADFE838, 0x62609838, 0xD1028839, 0x0736E92F,
|
||||
0x3BCA22A3, 0x1485B08A, 0x2DA7900B, 0x852C156D,
|
||||
0xE8F24803, 0x00078472, 0x13F0D332, 0x2ACFD0CF,
|
||||
0x5F747F5C, 0x87BB1E2F, 0xA7EFCB63, 0x23F432F0,
|
||||
0xE6CE7C5C, 0x1F954EF6, 0xB609C91B, 0x3B4571BF,
|
||||
0xEED17DC0, 0xE556CDA0, 0xA7846A8D, 0xFF105F94,
|
||||
0x52B7CCDE, 0x0E33E801, 0x664455EA, 0xF2C70414,
|
||||
0x73E7B486, 0x8F830661, 0x8B59E826, 0xBB8AEDCA,
|
||||
0xF3D70AB9, 0xD739F2B9, 0x4A04C34A, 0x88D0F089,
|
||||
0xE02191A2, 0xD89D9C78, 0x192C2749, 0xFC43A78F,
|
||||
0x0AAC88CB, 0x9438D42D, 0x9E280F7A, 0x36063802,
|
||||
0x38E8D018, 0x1C42A9CB, 0x92AAFF6C, 0xA24820C5,
|
||||
0x007F077F, 0xCE5BC543, 0x69668D58, 0x10D6FF74,
|
||||
0xBE00F621, 0x21300BBE, 0x2E9E8F46, 0x5ACEA629,
|
||||
0xFA1F86C7, 0x52F206B8, 0x3EDF1A75, 0x6DA8D843,
|
||||
0xCF719928, 0x73E3891F, 0xB4B95DD6, 0xB2A42D27,
|
||||
0xEDA20BBF, 0x1A58DBDF, 0xA449AD03, 0x6DDEF22B,
|
||||
0x900531E6, 0x3D3BFF35, 0x5B24ABA2, 0x472B3E4C,
|
||||
0x387F2D75, 0x4D8DBA36, 0x71CB5641, 0xE3473F3F,
|
||||
0xF6CD4B7F, 0xBF7D1428, 0x344B64D0, 0xC5CDFCB6,
|
||||
0xFE2E0182, 0x2C37A673, 0xDE4EB7A3, 0x63FDC933,
|
||||
0x01DC4063, 0x611F3571, 0xD167BFAF, 0x4496596F,
|
||||
0x3DEE0689, 0xD8704910, 0x7052A114, 0x068C9EC5,
|
||||
0x75D0E766, 0x4D54CC20, 0xB44ECDE2, 0x4ABC653E,
|
||||
0x2C550A21, 0x1A52C0DB, 0xCFED03D0, 0x119BAFE2,
|
||||
0x876A6133, 0xBC232088, 0x435BA1B2, 0xAE99BBFA,
|
||||
0xBB4F08E4, 0xA62B5F49, 0x1DA4B695, 0x336B84DE,
|
||||
0xDC813D31, 0x00C134FB, 0x397A98E6, 0x151F0E64,
|
||||
0xD9EB3E69, 0xD3C7DF60, 0xD2F2C336, 0x2DDD067B,
|
||||
0xBD122835, 0xB0B3BD3A, 0xB0D54E46, 0x8641F1E4,
|
||||
0xA0B38F96, 0x51D39199, 0x37A6AD75, 0xDF84EE41,
|
||||
0x3C034CBA, 0xACDA62FC, 0x11923B8B, 0x45EF170A,
|
||||
];
|
||||
|
||||
//void main(string[] args) {}
|
||||
|
|
|
|||
Loading…
Reference in New Issue