pmcgee
Members
Content Count
12 
Joined

Last visited
Everything posted by pmcgee

Just wondering if anyone knows a ballpark figure for the "size" of (any of) the Delphi compilers?

Generic circular buffer library released
pmcgee replied to TurboMagic's topic in Algorithms, Data Structures and Class Design
To be fair, there's a Catch22 in that idea. ie don't implement Generics until you have experience implementing Generics. (and the same would apply for trying to use a c++ like move feature) If someone publishes open source ... not part of a (reasonably big) commercial product like RAD ... and it is useful to maybe 80% of people, then sure it's valuable to point out edge cases  but it doesn't mean it should be discouraged. We need *more* Delphi out there. And more understanding of these elements that weren't even part of Delphi in say 2005. 
A bit of a challenge ... to produce Delphi code equaling the performance of C++'s next_permutation. Hopefully more readable in Delphi ... unit D_Next_perm_string; interface procedure swapCh(var a:Ansichar; var b:Ansichar); procedure reverse(var s:AnsiString; const a,x:word); function D_next_permutation(var s:AnsiString; const first, end_:word) : boolean; implementation //uses System.Diagnostics; procedure swapCh( var a : Ansichar; var b : Ansichar ); inline; var c : Ansichar; begin if a<>b then begin c := a; a := b; b := c; end; end; procedure reverse(var s:AnsiString; const a,x:word); inline; var i,j : word; //t : string; begin // x is one past the end of string if a = x1 then exit; j := ( xa ) shr 1; //trunc((xa)/2); for i := 1 to j do swapCh( s[a1+i] , s[xi] ); end; function D_next_permutation(var s : AnsiString; const first, end_ : word) : boolean; inline; var i, ii, j : word; begin if first = end_ then begin result := false; exit; end; i := first + 1; if i = end_ then begin result := false; exit; end; i := end_  1; while true do begin ii := i; dec(i); if s[i] < s[ii] then begin j := end_; repeat dec(j); until s[i] < s[j]; swapCh( s[i] , s[j] ); reverse(s, ii, end_); result := true; exit; end; if i = first then begin reverse(s, first, end_); result := false; exit; end; end; end; end. Ballpark, we're looking at about 6s vs 2s for s = "abcdefghijkl" (12 bytes). Increases by about a factor of 10 per character beyond that. Note the c++ end() iterator is **one past the end of string**, so ... D_next_permutation(s,1,13)

std::next_permutation
pmcgee replied to pmcgee's topic in Algorithms, Data Structures and Class Design
Sure ... byte / ansichar is a simple case ... but you need to start with simple and solve it the best you can before tackling a bigger / harder problem. The testing is made faster by just using a shorter string. The increase in time is about x10 at the 12 mark ... so 11 is quicker, and 10 is almost immediate. 
std::next_permutation
pmcgee replied to pmcgee's topic in Algorithms, Data Structures and Class Design
I'm a little confused. First post *is* the Delphi code. I've been learning about C++ for a couple years, alongside 20 with Delphi, and the ideas from both are interesting to me. Naively, it seems like a Delphi version should be able to produce about the same performance. Teasing that out should be educational. 
std::next_permutation
pmcgee replied to pmcgee's topic in Algorithms, Data Structures and Class Design
The calling code from c++ : int main() { const std::string Test = "abcdefghijkl"; int i = 0; std::string v = Test; Timer timer; timer.start(); do { ++i; } while ( std::next_permutation( v.begin(), v.end() ) ); timer.stop(); std::cout << "C++ " << i << "\n"; std::cout << "Seconds: " << timer.elapsedSeconds() << std::endl; AnsiString s = Test.c_str(); int len = s.Length()+1; i = 0; Timer timer2; timer2.start(); do { ++i; } while ( D_next_permutation(s, 1, len ) ); timer2.stop(); std::cout << "Delphi " << i << "\n"; std::cout << "Seconds: " << timer2.elapsedSeconds() << std::endl; getchar(); return 0; 
std::next_permutation
pmcgee replied to pmcgee's topic in Algorithms, Data Structures and Class Design
@Kas Ob. posted this in the original thread : procedure reverse2(var s: AnsiString; const a, x: word); inline; var i, j: word; tmpChar: Byte; SBytes: pByte absolute s; begin if a = x  1 then exit; j := (x  a) shr 1; // trunc((xa)/2); for i := 1 to j do begin tmpChar := SBytes[a  1 + i]; SBytes[a  1 + i] := SBytes[x  i]; SBytes[x  i] := tmpChar; end; end; var st: AnsiString; i: Integer; D: Uint64; begin st := '1234567890ab'; Writeln(st); d := GetTickCount; for i := 1 to 479001600 do // 12! reverse(st, 1, 12); D := GetTickCount  D; Writeln(d); st := '1234567890ab'; Writeln(st); d := GetTickCount; for i := 1 to 479001600 do reverse2(st, 1, 12); D := GetTickCount  D; Writeln(d); Readln; end. With these results ... (32 bit? 64 bit? compile) 1234567890ab 14125 1234567890ab 4891 
std::next_permutation
pmcgee replied to pmcgee's topic in Algorithms, Data Structures and Class Design
I didn't recall correctly. This was a significant speed up : procedure swapByte( a:Pbyte ; b:Pbyte ); inline; begin if a <> b then begin a^ := a^ + b^; b^ := a^  b^; a^ := a^  b^ ; end; end; swapByte( @s[a1+i] , @s[x_i] ); 
Micro optimization  effect of defined and not used local variables
pmcgee replied to Mike Torrettinni's topic in Algorithms, Data Structures and Class Design
Maybe this is OT for this thread. I'll look up where to start a new one, and add a link. 
Micro optimization  effect of defined and not used local variables
pmcgee replied to Mike Torrettinni's topic in Algorithms, Data Structures and Class Design
I had tried a couple things there ... this was a small improvement. I haven't pulled apart the assembly code yet. It's just an ongoing interest. It'll be fun to try it with char/byte array. procedure swapByte( a:Pbyte ; b:Pbyte ); inline; begin if a <> b then begin a^ := a^ + b^; b^ := a^  b^; a^ := a^  b^ ; end; end; procedure swapChar( var a : Ansichar; var b : Ansichar ); inline; var c : Ansichar; begin if a<>b then begin c := a; a := b; b := c; end; end; 
array [1..4] of TPoint as class property
pmcgee replied to FranzB's topic in Algorithms, Data Structures and Class Design
Just checking ... You can characterise one edge by one TPoint? or one edge by 4 TPoints? or does Fedges describe a parallelogram? Is ok. "GEORect". I get it. (I'm not sure 'edges' is the best name) ... (also ... can't see how to delete posts.) 
Micro optimization  effect of defined and not used local variables
pmcgee replied to Mike Torrettinni's topic in Algorithms, Data Structures and Class Design
I came across this trying to duplicate C++'s std::next_permutation using Delphi. I went through various modifications, and had left a string declaration in where it was no longer needed: procedure reverse(var s:AnsiString; const a,x:word); inline; var i,j : word; //t : string; begin // x is one past the end of string if a = x1 then exit; j := ( xa ) shr 1; // trunc((xa)/2); for i := 1 to j do swapCh( s[a1+i] , s[xi] ); end; All permutations of 12 chars = 479,001,600. C++ = 2s. Commenting out the string reduced the Delphi code from 9s to 6s. (I haven't been back to it since then.)