paweld/fpc
1
0
Fork 0
mirror of https://gitlab.com/freepascal.org/fpc/source.git synced 2025-09-07 20:50:42 +02:00
Code Issues Packages Projects Releases Wiki Activity
55,320 Commits 101 Branches 76 Tags 308 MiB
56fd8e1208
Commit Graph

55320 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
nickysn
b9a0b03e4b * use stosb instead of mov in DirectPutPixel16 
git-svn-id: trunk@41007 -
 2019-01-22 17:22:06 +00:00
nickysn
4336c4d793 * optimizations in DirectPutPixel16 to avoid the mul instruction 
git-svn-id: trunk@41006 -
 2019-01-22 17:20:38 +00:00
nickysn
a854411e7c * perform the clipping before initializing ES in PutPixel320, so it runs faster 
when the pixel happens to be outside the viewport

git-svn-id: trunk@41005 -
 2019-01-22 16:56:16 +00:00
nickysn
7c4466025a * PutPixel16 converted to pure asm routine 
git-svn-id: trunk@41004 -
 2019-01-22 16:40:51 +00:00
nickysn
1533f1e218 * implement the clipping in assembly instead of pascal in PutPixel16 
git-svn-id: trunk@41003 -
 2019-01-22 16:23:47 +00:00
nickysn
0199177c21 * perform the viewport adjustment in asm instead of pascal in PutPixel16 
git-svn-id: trunk@41002 -
 2019-01-22 16:11:41 +00:00
nickysn
d299ed7756 * moved the loading of the X and Y coordinates to registers in the front of the asm block in PutPixel16 
git-svn-id: trunk@41001 -
 2019-01-22 16:10:37 +00:00
nickysn
16c6090f30 * yet another mov eliminated by using different registers in GetPixel16 
git-svn-id: trunk@41000 -
 2019-01-22 16:07:51 +00:00
nickysn
820ba7dfb5 * use stosb instead of mov for writing a byte (the value doesn't matter) into EGA/VGA memory in PutPixel16 
git-svn-id: trunk@40999 -
 2019-01-22 16:05:29 +00:00
nickysn
83766c7462 * optimizations in PutPixel16 that remove the slow 'mul' instruction 
git-svn-id: trunk@40998 -
 2019-01-22 16:04:08 +00:00
nickysn
36c349696e * some minor optimizations in PutPixel16 (more efficient register usage, resulting in less mov instructions) 
git-svn-id: trunk@40997 -
 2019-01-22 15:50:31 +00:00
nickysn
5ba5b871bf * GetScanLine16 changed so that it reads video memory in 16-bit words, instead 
of 32-bit longints, because 16-bit generates more efficient code on the i8086

git-svn-id: trunk@40996 -
 2019-01-22 15:32:44 +00:00
marco
6cc8ed3256 * don't hangt on corrupt gifs, exit loops at end of streams. 
Fixes mantis #34919

git-svn-id: trunk@40995 -
 2019-01-22 15:11:04 +00:00
nickysn
06dbc518e4 * swap AL and AH in the last 'or' instruction in GetPixel16, so the result is 
combined directly into AL and a 'mov al, ah' instruction is removed

git-svn-id: trunk@40994 -
 2019-01-22 14:49:18 +00:00
nickysn
8cf2792c89 * replaced 'mov al,es:[si]' with 'seges lodsb' in GetPixel16 
git-svn-id: trunk@40993 -
 2019-01-22 14:35:32 +00:00
nickysn
95f2fffaf8 * replaced bl with al in the final bit read in GetPixel16 
git-svn-id: trunk@40992 -
 2019-01-22 14:34:47 +00:00
nickysn
5abb5b0658 * use xchg ax,cx instead of mov in GetPixel16 
git-svn-id: trunk@40991 -
 2019-01-22 14:33:41 +00:00
nickysn
fc61a2e62e * moved the initial GC port output to the beginning of GetPixel16 
git-svn-id: trunk@40990 -
 2019-01-22 14:32:01 +00:00
nickysn
98bbc0ab14 * swapped di and ax in the beginning of GetPixel16 
git-svn-id: trunk@40989 -
 2019-01-22 14:30:18 +00:00
nickysn
f1ad989ba1 * moved the ES register initialization in GetPixel16 to the beginning of the function 
git-svn-id: trunk@40988 -
 2019-01-22 14:28:46 +00:00
nickysn
6753237fe4 * replaced the slow mul instruction in GetPixel16 with faster shifts and adds 
git-svn-id: trunk@40987 -
 2019-01-22 14:27:48 +00:00
nickysn
985ec0688a * some more optimizations in GetPixel16 
git-svn-id: trunk@40986 -
 2019-01-22 12:58:02 +00:00
nickysn
0c440183c6 + enabled the true color support in ptcgraph 
git-svn-id: trunk@40985 -
 2019-01-22 09:07:37 +00:00
nickysn
d653193c36 * some optimizations in GetPixel16 
git-svn-id: trunk@40984 -
 2019-01-22 01:29:02 +00:00
nickysn
3b2fc5d34b * save another 2 bytes from the InitModeX code :) 
git-svn-id: trunk@40983 -
 2019-01-21 23:09:58 +00:00
nickysn
81aea037ec * reenabled the asm version of InitModeX 
git-svn-id: trunk@40982 -
 2019-01-21 23:07:51 +00:00
nickysn
596c1e7efa * huge memory model fixes in InitModeX 
git-svn-id: trunk@40981 -
 2019-01-21 22:54:08 +00:00
nickysn
37c05b3749 + added pascal version of InitModeX 
git-svn-id: trunk@40980 -
 2019-01-21 22:50:50 +00:00
florian
4fb7b5d3ff * fix MovOpMov2MovOp optimization if no -Cpcoreavx2 is used 
git-svn-id: trunk@40979 -
 2019-01-21 21:57:07 +00:00
nickysn
f6aa6e5b58 * another byte saved in the InitModeX routine :) 
git-svn-id: trunk@40978 -
 2019-01-21 18:06:07 +00:00
nickysn
5655b04b1c * use 'cmp mem,const' to check DontClearGraphMemory, instead of first loading it 
into a register in InitModeX. This saves 1 byte of code from an initialization
  routine. Hurrah! :)

git-svn-id: trunk@40977 -
 2019-01-21 17:43:57 +00:00
nickysn
5738ba19bb * converted InitModeX to a pure asm routine 
git-svn-id: trunk@40976 -
 2019-01-21 16:47:30 +00:00
nickysn
c26df36bc5 * size of the DrawnList buffer moved to a local variable, to avoid repeated code and repeated calculations 
git-svn-id: trunk@40975 -
 2019-01-21 16:24:46 +00:00
nickysn
6bb104a16d * unified the out of memory check in FloodFill and perform proper cleanup, so no memory leaks arise 
git-svn-id: trunk@40974 -
 2019-01-21 16:22:11 +00:00
nickysn
b654c34d64 * read the planes in reverse order, so that less shift instructions are required in GetPixel16 
git-svn-id: trunk@40973 -
 2019-01-21 16:09:13 +00:00
nickysn
33c330b3bc * store the size of the s1,s2 and s3 buffers, used inside FloodFill in a 
variable, so memory allocation and disposal becomes easier, with less ifdefs

git-svn-id: trunk@40972 -
 2019-01-21 16:08:38 +00:00
nickysn
f356add031 * moved the s1,s2,s3 scanline memory allocation in FloodFill before changing 
CurrentColor, so it doesn't get changed if the function exits prematurely due
  to out of memory

git-svn-id: trunk@40971 -
 2019-01-21 15:46:34 +00:00
nickysn
003a08f040 * use 'cbw' and 'cwd' instead of 'xor reg, reg' in the asm version of GetPixel16 
git-svn-id: trunk@40970 -
 2019-01-21 15:43:31 +00:00
nickysn
300e584936 * use 'cwd' instead of 'xor dx, dx' in the asm version GetPixel320 
git-svn-id: trunk@40969 -
 2019-01-21 15:30:58 +00:00
nickysn
57c931cb4c * use lodsb with a segment override prefix instead of mov in the asm version of GetPixel320 
git-svn-id: trunk@40968 -
 2019-01-21 15:28:22 +00:00
nickysn
1b4fcabc1d * use the si register instead of di in the asm version of GetPixel320 
git-svn-id: trunk@40967 -
 2019-01-21 15:18:51 +00:00
nickysn
cab9b4cb6b * fixed return of result in the asm version of GetPixel320 
git-svn-id: trunk@40966 -
 2019-01-21 15:16:16 +00:00
nickysn
92304ce573 * moved the check in FloodFill whether x,y are within the current viewport in 
the beginning, before any heap memory allocations, to avoid memory leaks when
  FloodFill is invoked with a point that is outside the current viewport

git-svn-id: trunk@40965 -
 2019-01-21 14:50:26 +00:00
nickysn
e6835e8ff7 * added check to ensure the memory allocation for DrawnList inside FloodFill hasn't returned nil 
git-svn-id: trunk@40964 -
 2019-01-21 14:47:48 +00:00
nickysn
f322b45611 * ignore VideoOfs entirely in mode 13h, since it doesn't support video pages 
git-svn-id: trunk@40962 -
 2019-01-21 14:12:17 +00:00
nickysn
698f8679d4 - removed comment saying that and/or/not put is not implemented, because it is implemented now 
git-svn-id: trunk@40961 -
 2019-01-21 14:10:41 +00:00
nickysn
39e1e4d2e7 + enabled the asm version of DirectPutPixel320 
git-svn-id: trunk@40960 -
 2019-01-21 14:09:25 +00:00
nickysn
907e3afae4 * 8086/8088 support in the asm version of DirectPutPixel320 
git-svn-id: trunk@40959 -
 2019-01-21 13:52:08 +00:00
nickysn
88f062dec8 + implemented all the write modes in the asm version of DirectPutPixel320 
git-svn-id: trunk@40958 -
 2019-01-21 13:51:40 +00:00
nickysn
f1467fdaaf * only load the low byte of CurrentColor in the asm version of DirectPutPixel320 
git-svn-id: trunk@40957 -
 2019-01-21 13:42:59 +00:00