sad_mmx.s 19.1 KB
Newer Older
F
Fabrice Bellard 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798
;  MMX/SSE optimized routines for SAD of 16*16 macroblocks
;	Copyright (C) Juan J. Sierralta P. <juanjo@atmlab.utfsm.cl>
;
;  dist1_* Original Copyright (C) 2000 Chris Atenasio <chris@crud.net>
;  Enhancements and rest Copyright (C) 2000 Andrew Stevens <as@comlab.ox.ac.uk>

;
;  This program is free software; you can redistribute it and/or
;  modify it under the terms of the GNU General Public License
;  as published by the Free Software Foundation; either version 2
;  of the License, or (at your option) any later version.
;
;  This program is distributed in the hope that it will be useful,
;  but WITHOUT ANY WARRANTY; without even the implied warranty of
;  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
;  GNU General Public License for more details.
;
;  You should have received a copy of the GNU General Public License
;  along with this program; if not, write to the Free Software
;  Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.
;

global pix_abs16x16_mmx

; int  pix_abs16x16_mmx(unsigned char *pix1,unsigned char *pix2, int lx, int h);
; esi = p1 (init:		blk1)
; edi = p2 (init:		blk2)
; ecx = rowsleft (init:	 h)
; edx = lx;

; mm0 = distance accumulators (4 words)
; mm1 = distance accumulators (4 words) 
; mm2 = temp 
; mm3 = temp
; mm4 = temp
; mm5 = temp 
; mm6 = 0
; mm7 = temp


align 32
pix_abs16x16_mmx:
	push				ebp					; save frame pointer
	mov				ebp, esp

	push				ebx					; Saves registers (called saves convention in
	push				ecx					; x86 GCC it seems)
	push				edx					; 
	push				esi
	push				edi
		
	pxor				mm0, mm0				; zero acculumators
	pxor				mm1, mm1
	pxor				mm6, mm6
	mov				esi, [ebp+8]		; get pix1
	mov				edi, [ebp+12]		; get pix2
	mov				edx, [ebp+16]		; get lx
	mov				ecx, [ebp+20]		; get rowsleft
	jmp				.nextrow
align 32

.nextrow:
	; First 8 bytes of the row
	
	movq				mm4, [edi]	; load first 8 bytes of pix2 row 
	movq				mm5, [esi]	; load first 8 bytes of pix1 row
	movq				mm3, mm4		; mm4 := abs(mm4-mm5)
	movq				mm2,[esi+8]	; load last 8 bytes of pix1 row	
	psubusb			mm4, mm5
	movq				mm7,[edi+8]	; load last 8 bytes of pix2 row 		
	psubusb			mm5, mm3
	por				mm4, mm5
	
	; Last 8 bytes of the row	

	movq				mm3, mm7		; mm7 := abs(mm7-mm2)
	psubusb			mm7, mm2
	psubusb			mm2, mm3
	por				mm7, mm2
	
	; Now mm4 and mm7 have 16 absdiffs to add
	
	; First 8 bytes of the row2
	
	
	add				edi, edx
	movq				mm2, [edi]	; load first 8 bytes of pix2 row 
	add				esi, edx
	movq				mm5, [esi]	; load first 8 bytes of pix1 row
	
	
	
	movq				mm3, mm2		; mm2 := abs(mm2-mm5)
	psubusb			mm2, mm5
	movq				mm6,[esi+8]	; load last 8 bytes of pix1 row
	psubusb			mm5, mm3
	por				mm2, mm5
		
	; Last 8 bytes of the row2
	
	movq				mm5,[edi+8]	; load last 8 bytes of pix2 row 
	
	
	movq				mm3, mm5		; mm5 := abs(mm5-mm6)
	psubusb			mm5, mm6
	psubusb			mm6, mm3
	por				mm5, mm6

	; Now mm2, mm4, mm5, mm7 have 32 absdiffs
	
	movq				mm3, mm7
	
	pxor				mm6, mm6		; Zero mm6
		
	punpcklbw		mm3, mm6		; Unpack to words and add
	punpckhbw		mm7, mm6
	paddusw			mm7, mm3
	
	movq				mm3, mm5
	
	punpcklbw		mm3, mm6		; Unpack to words and add
	punpckhbw		mm5, mm6
	paddusw			mm5, mm3
		
	paddusw			mm0, mm7		; Add to the acumulator (mm0)
	paddusw			mm1, mm5		; Add to the acumulator (mm1)
		
	movq				mm3, mm4
	
	punpcklbw		mm3, mm6		; Unpack to words and add
	punpckhbw		mm4, mm6
	movq				mm5, mm2
	paddusw			mm4, mm3

	
	
	punpcklbw		mm5, mm6		; Unpack to words and add
	punpckhbw		mm2, mm6
	paddusw			mm2, mm5

	; Loop termination
	
	add				esi, edx		; update pointers to next row
	paddusw			mm0, mm4		; Add to the acumulator (mm0)
	add				edi, edx		
	sub				ecx,2
	paddusw			mm1, mm2		; Add to the acumulator (mm1)
	test				ecx, ecx		; check rowsleft
	jnz				near .nextrow
	
	paddusw			mm0, mm1
	movq				mm2, mm0		; Copy mm0 to mm2
	psrlq				mm2, 32
	paddusw			mm0, mm2		; Add 
	movq				mm3, mm0
	psrlq				mm3, 16
	paddusw			mm0, mm3
	movd				eax, mm0		; Store return value
	and				eax, 0xffff

	pop edi
	pop esi	
	pop edx			
	pop ecx			
	pop ebx			

	pop ebp							; restore stack pointer

	;emms								; clear mmx registers
	ret								; return

global pix_abs16x16_sse

; int  pix_abs16x16_mmx(unsigned char *pix1,unsigned char *pix2, int lx, int h);
; esi = p1 (init:		blk1)
; edi = p2 (init:		blk2)
; ecx = rowsleft (init:	 h)
; edx = lx;

; mm0 = distance accumulators (4 words)
; mm1 = distance accumulators (4 words) 
; mm2 = temp 
; mm3 = temp
; mm4 = temp
; mm5 = temp 
; mm6 = temp
; mm7 = temp


align 32
pix_abs16x16_sse:
	push				ebp					; save frame pointer
	mov				ebp, esp

	push				ebx					; Saves registers (called saves convention in
	push				ecx					; x86 GCC it seems)
	push				edx					; 
	push				esi
	push				edi
		
	pxor				mm0, mm0				; zero acculumators
	pxor				mm1, mm1
	mov				esi, [ebp+8]		; get pix1
	mov				edi, [ebp+12]		; get pix2
	mov				edx, [ebp+16]		; get lx
	mov				ecx, [ebp+20]		; get rowsleft
	jmp				.next4row
align 32

.next4row:
	; First row
	
	movq				mm4, [edi]		; load first 8 bytes of pix2 row 
	movq				mm5, [edi+8]	; load last 8 bytes of pix2 row
	psadbw			mm4, [esi]		; SAD of first 8 bytes
	psadbw			mm5, [esi+8]	; SAD of last 8 bytes
	paddw				mm0, mm4			; Add to acumulators
	paddw				mm1, mm5
		
	; Second row	

	add				edi, edx;
	add				esi, edx;
	
	movq				mm6, [edi]		; load first 8 bytes of pix2 row 
	movq				mm7, [edi+8]	; load last 8 bytes of pix2 row
	psadbw			mm6, [esi]		; SAD of first 8 bytes
	psadbw			mm7, [esi+8]	; SAD of last 8 bytes
	paddw				mm0, mm6			; Add to acumulators
	paddw				mm1, mm7
		
	; Third row
	
	add				edi, edx;
	add				esi, edx;
	
	movq				mm4, [edi]		; load first 8 bytes of pix2 row 
	movq				mm5, [edi+8]	; load last 8 bytes of pix2 row
	psadbw			mm4, [esi]		; SAD of first 8 bytes
	psadbw			mm5, [esi+8]	; SAD of last 8 bytes
	paddw				mm0, mm4			; Add to acumulators
	paddw				mm1, mm5
		
	; Fourth row	

	add				edi, edx;
	add				esi, edx;
	
	movq				mm6, [edi]		; load first 8 bytes of pix2 row 
	movq				mm7, [edi+8]	; load last 8 bytes of pix2 row
	psadbw			mm6, [esi]		; SAD of first 8 bytes
	psadbw			mm7, [esi+8]	; SAD of last 8 bytes
	paddw				mm0, mm6			; Add to acumulators
	paddw				mm1, mm7
	
	; Loop termination
	
	add				esi, edx		; update pointers to next row
	add				edi, edx		
	sub				ecx,4
	test				ecx, ecx		; check rowsleft
	jnz				near .next4row
	
	paddd				mm0, mm1		; Sum acumulators
	movd				eax, mm0		; Store return value

	pop edi
	pop esi	
	pop edx			
	pop ecx			
	pop ebx			

	pop ebp							; restore stack pointer

	;emms								; clear mmx registers
	ret								; return
		
global pix_abs16x16_x2_mmx

; int  pix_abs16x16_x2_mmx(unsigned char *pix1,unsigned char *pix2, int lx, int h);
; esi = p1 (init:		blk1)
; edi = p2 (init:		blk2)
; ecx = rowsleft (init:	 h)
; edx = lx;

; mm0 = distance accumulators (4 words)
; mm1 = distance accumulators (4 words) 
; mm2 = temp 
; mm3 = temp
; mm4 = temp
; mm5 = temp 
; mm6 = 0
; mm7 = temp


align 32
pix_abs16x16_x2_mmx:
	push				ebp					; save frame pointer
	mov				ebp, esp

	push				ebx					; Saves registers (called saves convention in
	push				ecx					; x86 GCC it seems)
	push				edx					; 
	push				esi
	push				edi
		
	pxor				mm0, mm0				; zero acculumators
	pxor				mm1, mm1
	pxor				mm6, mm6
	mov				esi, [ebp+8]		; get pix1
	mov				edi, [ebp+12]		; get pix2
	mov				edx, [ebp+16]		; get lx
	mov				ecx, [ebp+20]		; get rowsleft
	jmp				.nextrow_x2
align 32

.nextrow_x2:
	; First 8 bytes of the row
	
	movq				mm4, [edi]			; load first 8 bytes of pix2 row 
	movq				mm5, [edi+1]		; load bytes 1-8 of pix2 row

	movq				mm2, mm4		; copy mm4 on mm2
	movq				mm3, mm5		; copy mm5 on mm3
	punpcklbw		mm4, mm6		; first 4 bytes of [edi] on mm4
	punpcklbw		mm5, mm6		; first 4 bytes of [edi+1] on mm5
	paddusw			mm4, mm5		; mm4 := first 4 bytes interpolated in words
	psrlw				mm4, 1

	punpckhbw		mm2, mm6		; last 4 bytes of [edi] on mm2
	punpckhbw		mm3, mm6		; last 4 bytes of [edi+1] on mm3
	paddusw			mm2, mm3		; mm2 := last 4 bytes interpolated in words
	psrlw				mm2, 1
	
	packuswb			mm4, mm2 	; pack 8 bytes interpolated on mm4
	movq				mm5,[esi]	; load first 8 bytes of pix1 row
	
	movq				mm3, mm4		; mm4 := abs(mm4-mm5)
	psubusb			mm4, mm5
	psubusb			mm5, mm3
	por				mm4, mm5
		
	; Last 8 bytes of the row	

	movq	mm7, [edi+8]			; load last 8 bytes of pix2 row 
	movq	mm5, [edi+9]			; load bytes 10-17 of pix2 row

	movq				mm2, mm7		; copy mm7 on mm2
	movq				mm3, mm5		; copy mm5 on mm3
	punpcklbw		mm7, mm6		; first 4 bytes of [edi+8] on mm7
	punpcklbw		mm5, mm6		; first 4 bytes of [edi+9] on mm5
	paddusw			mm7, mm5		; mm1 := first 4 bytes interpolated in words
	psrlw				mm7, 1

	punpckhbw		mm2, mm6		; last 4 bytes of [edi] on mm2
	punpckhbw		mm3, mm6		; last 4 bytes of [edi+1] on mm3
	paddusw			mm2, mm3		; mm2 := last 4 bytes interpolated in words
	psrlw				mm2, 1
	
	packuswb			mm7, mm2 	; pack 8 bytes interpolated on mm1
	movq				mm5,[esi+8]	; load last 8 bytes of pix1 row
	
	movq				mm3, mm7		; mm7 := abs(mm1-mm5)
	psubusb			mm7, mm5
	psubusb			mm5, mm3
	por				mm7, mm5
	
	; Now mm4 and mm7 have 16 absdiffs to add
	
	movq				mm3, mm4		; Make copies of these bytes
	movq				mm2, mm7
		
	punpcklbw		mm4, mm6		; Unpack to words and add
	punpcklbw		mm7, mm6
	paddusw			mm4, mm7
	paddusw			mm0, mm4		; Add to the acumulator (mm0)
	
	punpckhbw		mm3, mm6		; Unpack to words and add
	punpckhbw		mm2, mm6
	paddusw			mm3, mm2
	paddusw			mm1, mm3		; Add to the acumulator (mm1)
	
	; Loop termination
	
	add				esi, edx		; update pointers to next row
	add				edi, edx		
			
	sub				ecx,1
	test				ecx, ecx		; check rowsleft
	jnz				near .nextrow_x2
	
	paddusw			mm0, mm1 
	
	movq				mm1, mm0		; Copy mm0 to mm1
	psrlq				mm1, 32
	paddusw			mm0, mm1		; Add 
	movq				mm2, mm0
	psrlq				mm2, 16
	paddusw			mm0, mm2
	movd				eax, mm0		; Store return value
	and				eax, 0xffff

	pop edi
	pop esi	
	pop edx			
	pop ecx			
	pop ebx			

	pop ebp							; restore stack pointer

	emms								; clear mmx registers
	ret								; return
	
global pix_abs16x16_y2_mmx

; int  pix_abs16x16_y2_mmx(unsigned char *pix1,unsigned char *pix2, int lx, int h);
; esi = p1 (init:		blk1)
; edi = p2 (init:		blk2)
; ebx = p2 + lx
; ecx = rowsleft (init:	 h)
; edx = lx;

; mm0 = distance accumulators (4 words)
; mm1 = distance accumulators (4 words) 
; mm2 = temp 
; mm3 = temp
; mm4 = temp
; mm5 = temp 
; mm6 = 0
; mm7 = temp


align 32
pix_abs16x16_y2_mmx:
	push				ebp					; save frame pointer
	mov				ebp, esp

	push				ebx					; Saves registers (called saves convention in
	push				ecx					; x86 GCC it seems)
	push				edx					; 
	push				esi
	push				edi
		
	pxor				mm0, mm0				; zero acculumators
	pxor				mm1, mm1
	pxor				mm6, mm6
	mov				esi, [ebp+8]		; get pix1
	mov				edi, [ebp+12]		; get pix2
	mov				edx, [ebp+16]		; get lx
	mov				ecx, [ebp+20]		; get rowsleft
	mov				ebx, edi
	add				ebx, edx
	jmp				.nextrow_y2
align 32

.nextrow_y2:
	; First 8 bytes of the row
	
	movq				mm4, [edi]			; load first 8 bytes of pix2 row 
	movq				mm5, [ebx]			; load bytes 1-8 of pix2 row

	movq				mm2, mm4		; copy mm4 on mm2
	movq				mm3, mm5		; copy mm5 on mm3
	punpcklbw		mm4, mm6		; first 4 bytes of [edi] on mm4
	punpcklbw		mm5, mm6		; first 4 bytes of [ebx] on mm5
	paddusw			mm4, mm5		; mm4 := first 4 bytes interpolated in words
	psrlw				mm4, 1

	punpckhbw		mm2, mm6		; last 4 bytes of [edi] on mm2
	punpckhbw		mm3, mm6		; last 4 bytes of [edi+1] on mm3
	paddusw			mm2, mm3		; mm2 := last 4 bytes interpolated in words
	psrlw				mm2, 1
	
	packuswb			mm4, mm2 	; pack 8 bytes interpolated on mm4
	movq				mm5,[esi]	; load first 8 bytes of pix1 row
	
	movq				mm3, mm4		; mm4 := abs(mm4-mm5)
	psubusb			mm4, mm5
	psubusb			mm5, mm3
	por				mm4, mm5
		
	; Last 8 bytes of the row	

	movq	mm7, [edi+8]			; load last 8 bytes of pix2 row 
	movq	mm5, [ebx+8]			; load bytes 10-17 of pix2 row

	movq				mm2, mm7		; copy mm7 on mm2
	movq				mm3, mm5		; copy mm5 on mm3
	punpcklbw		mm7, mm6		; first 4 bytes of [edi+8] on mm7
	punpcklbw		mm5, mm6		; first 4 bytes of [ebx+8] on mm5
	paddusw			mm7, mm5		; mm1 := first 4 bytes interpolated in words
	psrlw				mm7, 1

	punpckhbw		mm2, mm6		; last 4 bytes of [edi+8] on mm2
	punpckhbw		mm3, mm6		; last 4 bytes of [ebx+8] on mm3
	paddusw			mm2, mm3		; mm2 := last 4 bytes interpolated in words
	psrlw				mm2, 1
	
	packuswb			mm7, mm2 	; pack 8 bytes interpolated on mm1
	movq				mm5,[esi+8]	; load last 8 bytes of pix1 row
	
	movq				mm3, mm7		; mm7 := abs(mm1-mm5)
	psubusb			mm7, mm5
	psubusb			mm5, mm3
	por				mm7, mm5
	
	; Now mm4 and mm7 have 16 absdiffs to add
	
	movq				mm3, mm4		; Make copies of these bytes
	movq				mm2, mm7
		
	punpcklbw		mm4, mm6		; Unpack to words and add
	punpcklbw		mm7, mm6
	paddusw			mm4, mm7
	paddusw			mm0, mm4		; Add to the acumulator (mm0)
	
	punpckhbw		mm3, mm6		; Unpack to words and add
	punpckhbw		mm2, mm6
	paddusw			mm3, mm2
	paddusw			mm1, mm3		; Add to the acumulator (mm1)
	
	; Loop termination
	
	add				esi, edx		; update pointers to next row
	add				edi, edx		
	add				ebx, edx		
	sub				ecx,1
	test				ecx, ecx		; check rowsleft
	jnz				near .nextrow_y2
	
	paddusw			mm0, mm1 
	
	movq				mm1, mm0		; Copy mm0 to mm1
	psrlq				mm1, 32
	paddusw			mm0, mm1		; Add 
	movq				mm2, mm0
	psrlq				mm2, 16
	paddusw			mm0, mm2
	movd				eax, mm0		; Store return value
	and				eax, 0xffff

	pop edi
	pop esi	
	pop edx			
	pop ecx			
	pop ebx			

	pop ebp							; restore stack pointer

	emms								; clear mmx registers
	ret								; return
	
global pix_abs16x16_xy2_mmx

; int pix_abs16x16_xy2_mmx(unsigned char *p1,unsigned char *p2,int lx,int h);

; esi = p1 (init:		blk1)
; edi = p2 (init:		blk2)
; ebx = p1+lx
; ecx = rowsleft (init:	 h)
; edx = lx;

; mm0 = distance accumulators (4 words)
; mm1 = bytes p2
; mm2 = bytes p1
; mm3 = bytes p1+lx
; I'd love to find someplace to stash p1+1 and p1+lx+1's bytes
; but I don't think thats going to happen in iA32-land...
; mm4 = temp 4 bytes in words interpolating p1, p1+1
; mm5 = temp 4 bytes in words from p2
; mm6 = temp comparison bit mask p1,p2
; mm7 = temp comparison bit mask p2,p1


align 32
pix_abs16x16_xy2_mmx:
	push ebp		; save stack pointer
	mov ebp, esp	; so that we can do this

	push ebx		; Saves registers (called saves convention in
	push ecx		; x86 GCC it seems)
	push edx		; 
	push esi
	push edi
		
	pxor mm0, mm0				; zero acculumators

	mov esi, [ebp+12]			; get p1
	mov edi, [ebp+8]			; get p2
	mov edx, [ebp+16]			; get lx
	mov ecx, [ebp+20]			; rowsleft := h
	mov ebx, esi
    add ebx, edx		
	jmp .nextrowmm11					; snap to it
align 32
.nextrowmm11:

		;; 
		;; First 8 bytes of row
		;; 
		
		;; First 4 bytes of 8

	movq mm4, [esi]             ; mm4 := first 4 bytes p1
	pxor mm7, mm7
	movq mm2, mm4				;  mm2 records all 8 bytes
	punpcklbw mm4, mm7            ;  First 4 bytes p1 in Words...
	
	movq mm6, [ebx]			    ;  mm6 := first 4 bytes p1+lx
	movq mm3, mm6               ;  mm3 records all 8 bytes
	punpcklbw mm6, mm7
	paddw mm4, mm6              


	movq mm5, [esi+1]			; mm5 := first 4 bytes p1+1
	punpcklbw mm5, mm7            ;  First 4 bytes p1 in Words...
	paddw mm4, mm5		
	movq mm6, [ebx+1]           ;  mm6 := first 4 bytes p1+lx+1
	punpcklbw mm6, mm7
	paddw mm4, mm6

	psrlw mm4, 2	            ; mm4 := First 4 bytes interpolated in words
		
	movq mm5, [edi]				; mm5:=first 4 bytes of p2 in words
	movq mm1, mm5
	punpcklbw mm5, mm7
			
	movq  mm7,mm4
	pcmpgtw mm7,mm5		; mm7 := [i : W0..3,mm4>mm5]

	movq  mm6,mm4		; mm6 := [i : W0..3, (mm4-mm5)*(mm4-mm5 > 0)]
 	psubw mm6,mm5
	pand  mm6, mm7

	paddw mm0, mm6				; Add to accumulator

	movq  mm6,mm5       ; mm6 := [i : W0..3,mm5>mm4]
	pcmpgtw mm6,mm4	    
 	psubw mm5,mm4		; mm5 := [i : B0..7, (mm5-mm4)*(mm5-mm4 > 0)]
	pand  mm5, mm6		

	paddw mm0, mm5				; Add to accumulator

		;; Second 4 bytes of 8
	
	movq mm4, mm2		    ; mm4 := Second 4 bytes p1 in words
	pxor  mm7, mm7
	punpckhbw mm4, mm7			
	movq mm6, mm3			; mm6 := Second 4 bytes p1+1 in words  
	punpckhbw mm6, mm7
	paddw mm4, mm6          

	movq mm5, [esi+1]			; mm5 := first 4 bytes p1+1
	punpckhbw mm5, mm7          ;  First 4 bytes p1 in Words...
	paddw mm4, mm5
	movq mm6, [ebx+1]           ;  mm6 := first 4 bytes p1+lx+1
	punpckhbw mm6, mm7
	paddw mm4, mm6

	psrlw mm4, 2	            ; mm4 := First 4 bytes interpolated in words
		
	movq mm5, mm1			; mm5:= second 4 bytes of p2 in words
	punpckhbw mm5, mm7
			
	movq  mm7,mm4
	pcmpgtw mm7,mm5		; mm7 := [i : W0..3,mm4>mm5]

	movq  mm6,mm4		; mm6 := [i : W0..3, (mm4-mm5)*(mm4-mm5 > 0)]
 	psubw mm6,mm5
	pand  mm6, mm7

	paddw mm0, mm6				; Add to accumulator

	movq  mm6,mm5       ; mm6 := [i : W0..3,mm5>mm4]
	pcmpgtw mm6,mm4	    
 	psubw mm5,mm4		; mm5 := [i : B0..7, (mm5-mm4)*(mm5-mm4 > 0)]
	pand  mm5, mm6		

	paddw mm0, mm5				; Add to accumulator


 		;;
		;; Second 8 bytes of row
		;; 
		;; First 4 bytes of 8

	movq mm4, [esi+8]             ; mm4 := first 4 bytes p1+8
	pxor mm7, mm7
	movq mm2, mm4				;  mm2 records all 8 bytes
	punpcklbw mm4, mm7            ;  First 4 bytes p1 in Words...
	
	movq mm6, [ebx+8]			    ;  mm6 := first 4 bytes p1+lx+8
	movq mm3, mm6               ;  mm3 records all 8 bytes
	punpcklbw mm6, mm7
	paddw mm4, mm6              


	movq mm5, [esi+9]			; mm5 := first 4 bytes p1+9
	punpcklbw mm5, mm7            ;  First 4 bytes p1 in Words...
	paddw mm4, mm5
	movq mm6, [ebx+9]           ;  mm6 := first 4 bytes p1+lx+9
	punpcklbw mm6, mm7
	paddw mm4, mm6

	psrlw mm4, 2	            ; mm4 := First 4 bytes interpolated in words
		
	movq mm5, [edi+8]				; mm5:=first 4 bytes of p2+8 in words
	movq mm1, mm5
	punpcklbw mm5, mm7
			
	movq  mm7,mm4
	pcmpgtw mm7,mm5		; mm7 := [i : W0..3,mm4>mm5]

	movq  mm6,mm4		; mm6 := [i : W0..3, (mm4-mm5)*(mm4-mm5 > 0)]
 	psubw mm6,mm5
	pand  mm6, mm7

	paddw mm0, mm6				; Add to accumulator

	movq  mm6,mm5       ; mm6 := [i : W0..3,mm5>mm4]
	pcmpgtw mm6,mm4	    
 	psubw mm5,mm4		; mm5 := [i : B0..7, (mm5-mm4)*(mm5-mm4 > 0)]
	pand  mm5, mm6		

	paddw mm0, mm5				; Add to accumulator

		;; Second 4 bytes of 8
	
	movq mm4, mm2		    ; mm4 := Second 4 bytes p1 in words
	pxor  mm7, mm7
	punpckhbw mm4, mm7			
	movq mm6, mm3			; mm6 := Second 4 bytes p1+1 in words  
	punpckhbw mm6, mm7
	paddw mm4, mm6          

	movq mm5, [esi+9]			; mm5 := first 4 bytes p1+1
	punpckhbw mm5, mm7          ;  First 4 bytes p1 in Words...
	paddw mm4, mm5	
	movq mm6, [ebx+9]           ;  mm6 := first 4 bytes p1+lx+1
	punpckhbw mm6, mm7
	paddw mm4, mm6
		
	psrlw mm4, 2	            ; mm4 := First 4 bytes interpolated in words

	movq mm5, mm1			; mm5:= second 4 bytes of p2 in words
	punpckhbw mm5, mm7
			
	movq  mm7,mm4
	pcmpgtw mm7,mm5		; mm7 := [i : W0..3,mm4>mm5]

	movq  mm6,mm4		; mm6 := [i : W0..3, (mm4-mm5)*(mm4-mm5 > 0)]
 	psubw mm6,mm5
	pand  mm6, mm7

	paddw mm0, mm6				; Add to accumulator

	movq  mm6,mm5       ; mm6 := [i : W0..3,mm5>mm4]
	pcmpgtw mm6,mm4	    
 	psubw mm5,mm4		; mm5 := [i : B0..7, (mm5-mm4)*(mm5-mm4 > 0)]
	pand  mm5, mm6		

	paddw mm0, mm5				; Add to accumulator


		;;
		;;	Loop termination condition... and stepping
		;;		

	add esi, edx		; update pointer to next row
	add edi, edx		; ditto
	add ebx, edx

	sub  ecx,1
	test ecx, ecx		; check rowsleft
	jnz near .nextrowmm11
		
		;; Sum the Accumulators
	movq  mm4, mm0
	psrlq mm4, 32
	paddw mm0, mm4
	movq  mm6, mm0
	psrlq mm6, 16
	paddw mm0, mm6
	movd eax, mm0		; store return value
	and  eax, 0xffff
		
	pop edi
	pop esi	
	pop edx			
	pop ecx			
	pop ebx			

	pop ebp			; restore stack pointer

	emms			; clear mmx registers
	ret			; we now return you to your regular programming