This program is still to be converted. Because the Windows source uses ThreadTimer::SetThreadAffinityMask(), this will take some time.
ssepackedintegerthreshold.asm
; Name: ssepackedintegerthreshold.asm ; ; Build: g++ -c -m32 main.cpp -o main.o ; nasm -f elf32 -o ssepackedintegerthreshold.o ssepackedintegerthreshold.asm ; g++ -m32 -o ssepackedintegerthreshold ssepackedintegerthreshold.o main.o ../../commonfiles/xmmval.o ; ; Source: Modern x86 Assembly Language Programming p. 279 ; ; Remark: Until ImageBuffer.cpp is converted, this example is of little use. extern NUM_PIXELS_MAX global SsePiThreshold global SsePiCalcMean ; Image threshold data structure (see SsePackedIntegerThreshold.h) struc ITD .PbSrc: resd 1 .PbMask: resd 1 .NumPixels: resd 1 .Threshold: resb 1 .Pad: resb 3 .NumMaskedPixels: resd 1 .SumMaskedPixels: resd 1 .MeanMaskedPixels: resq 1 endstruc section .data align 16 PixelScale: times 16 db 0x80 ;uint8 to int8 scale value CountPixelsMask: times 16 db 0x01 ;mask to count pixels R8_MinusOne: dq -1.0 ;invalid mean value section .text ; extern "C" bool SsePiThreshold(ITD* itd); ; ; Description: The following function performs image thresholding ; of an 8 bits-per-pixel grayscale image. ; ; Returns: 0 = invalid size or unaligned image buffer ; 1 = success ; ; Requires: SSSE3 %define itd [ebp+8] SsePiThreshold: push ebp mov ebp,esp push esi push edi ; Load and verify the argument values in ITD structure mov edx,itd ;edx = 'itd' xor eax,eax ;set error return code mov ecx,[edx+ITD.NumPixels] ;ecx = NumPixels test ecx,ecx jz .done ;jump if num_pixels == 0 cmp ecx,[NUM_PIXELS_MAX] ja .done ;jump if num_pixels too big test ecx,0fh jnz .done ;jump if num_pixels % 16 != 0 shr ecx,4 ;ecx = number of packed pixels mov esi,[edx+ITD.PbSrc] ;esi = PbSrc test esi,0fh jnz .done ;jump if misaligned mov edi,[edx+ITD.PbMask] ;edi = PbMask test edi,0fh jnz .done ;jump if misaligned ; Initialize packed threshold movzx eax,byte[edx+ITD.Threshold] ;eax = threshold movd xmm1,eax ;xmm1[7:0] = threshold pxor xmm0,xmm0 ;mask for pshufb pshufb xmm1,xmm0 ;xmm1 = packed threshold movdqa xmm2,[PixelScale] psubb xmm1,xmm2 ;xmm1 = scaled threshold ; Create the mask image .@1: movdqa xmm0,[esi] ;load next packed pixel psubb xmm0,xmm2 ;xmm0 = scaled image pixels pcmpgtb xmm0,xmm1 ;compare against threshold movdqa [edi],xmm0 ;save packed threshold mask add esi,16 add edi,16 dec ecx jnz .@1 ;repeat until done mov eax,1 ;set return code .done: pop edi pop esi pop ebp ret ; extern "C" bool SsePiCalcMean(ITD* itd); ; ; Description: The following function calculates the mean value all ; above-threshold image pixels using the mask created by ; the function SsePiThreshold_. ; ; Returns: 0 = invalid image size or unaligned image buffer ; 1 = success ; ; Requires: SSSE3 %define itd [ebp+8] SsePiCalcMean: push ebp mov ebp,esp push ebx push esi push edi ; Load and verify the argument values in ITD structure mov eax,itd ;eax = 'itd' mov ecx,[eax+ITD.NumPixels] ;ecx = NumPixels test ecx,ecx jz .error ;jump if num_pixels == 0 cmp ecx,[NUM_PIXELS_MAX] ja .error ;jump if num_pixels too big test ecx,0x0f jnz .error ;jump if num_pixels % 16 != 0 shr ecx,4 ;ecx = number of packed pixels mov edi,[eax+ITD.PbMask] ;edi = PbMask test edi,0x0f jnz .error ;jump if PbMask not aligned mov esi,[eax+ITD.PbSrc] ;esi = PbSrc test esi,0x0f jnz .error ;jump if PbSrc not aligned ; Initialize values for mean calculation xor edx,edx ;edx = update counter pxor xmm7,xmm7 ;xmm7 = packed zero pxor xmm2,xmm2 ;xmm2 = sum_masked_pixels (8 words) pxor xmm3,xmm3 ;xmm3 = sum_masked_pixels (8 words) pxor xmm4,xmm4 ;xmm4 = sum_masked_pixels (4 dwords) pxor xmm6,xmm6 ;xmm6 = num_masked_pixels (8 bytes) xor ebx,ebx ;ebx = num_masked_pixels (1 dword) ; Register usage for processing loop ; esi = PbSrc, edi = PbMask, eax = itd ; ebx = num_pixels_masked, ecx = NumPixels / 16, edx = update counter ; ; xmm0 = packed pixel, xmm1 = packed mask ; xmm3:xmm2 = sum_masked_pixels (16 words) ; xmm4 = sum_masked_pixels (4 dwords) ; xmm5 = scratch register ; xmm6 = packed num_masked_pixels ; xmm7 = packed zero .@1: movdqa xmm0,[esi] ;load next packed pixel movdqa xmm1,[edi] ;load next packed mask ; Update sum_masked_pixels (word values) movdqa xmm5,[CountPixelsMask] pand xmm5,xmm1 paddb xmm6,xmm5 ;update num_masked_pixels pand xmm0,xmm1 ;set non-masked pixels to zero movdqa xmm1,xmm0 punpcklbw xmm0,xmm7 punpckhbw xmm1,xmm7 ;xmm1:xmm0 = masked pixels (words) paddw xmm2,xmm0 paddw xmm3,xmm1 ;xmm3:xmm2 = sum_masked_pixels ; Check and see if it's necessary to update the dword sum_masked_pixels ; in xmm4 and num_masked_pixels in ebx inc edx cmp edx,255 jb .noUpdate call SsePiCalcMeanUpdateSums .noUpdate: add esi,16 add edi,16 dec ecx jnz .@1 ;repeat loop until done ; Main processing loop is finished. If necessary, perform final update ; of sum_masked_pixels in xmm4 & num_masked_pixels in ebx. test edx,edx jz .@2 call SsePiCalcMeanUpdateSums ; Compute and save final sum_masked_pixels & num_masked_pixels .@2: phaddd xmm4,xmm7 phaddd xmm4,xmm7 movd edx,xmm4 ;edx = final sum_mask_pixels mov [eax+ITD.SumMaskedPixels],edx ;save final sum_masked_pixels mov [eax+ITD.NumMaskedPixels],ebx ;save final num_masked_pixels ; Compute mean of masked pixels test ebx,ebx ;is num_mask_pixels zero? jz .noMean ;if yes, skip calc of mean cvtsi2sd xmm0,edx ;xmm0 = sum_masked_pixels cvtsi2sd xmm1,ebx ;xmm1 = num_masked_pixels divsd xmm0,xmm1 ;xmm0 = mean_masked_pixels jmp .@3 .noMean: movsd xmm0,[R8_MinusOne] ;use -1.0 for no mean .@3: movsd [eax+ITD.MeanMaskedPixels],xmm0 ;save mean mov eax,1 ;set return code .done: pop edi pop esi pop ebx pop ebp ret .error: xor eax,eax ;set error return code jmp .done ; void SsePiCalcMeanUpdateSums ; ; Description: The following function updates sum_masked_pixels in xmm4 ; and num_masked_pixels in ebx. It also resets any ; necessary intermediate values in order to prevent an ; overflow condition. ; ; Register contents: ; xmm3:xmm2 = packed word sum_masked_pixels ; xmm4 = packed dword sum_masked_pixels ; xmm6 = packed num_masked_pixels ; xmm7 = packed zero ; ebx = num_masked_pixels ; ; Temp registers: ; xmm0, xmm1, xmm5, edx SsePiCalcMeanUpdateSums: ; Promote packed word sum_masked_pixels to dword movdqa xmm0,xmm2 movdqa xmm1,xmm3 punpcklwd xmm0,xmm7 punpcklwd xmm1,xmm7 punpckhwd xmm2,xmm7 punpckhwd xmm3,xmm7 ; Update packed dword sums in sum_masked_pixels paddd xmm0,xmm1 paddd xmm2,xmm3 paddd xmm4,xmm0 paddd xmm4,xmm2 ;xmm4 = packed sum_masked_pixels ; Sum num_masked_pixel counts (bytes) in xmm6, then add to total in ebx. movdqa xmm5,xmm6 punpcklbw xmm5,xmm7 punpckhbw xmm6,xmm7 ;xmm6:xmm5 = packed num_masked_pixels paddw xmm6,xmm5 ;xmm6 = packed num_masked_pixels phaddw xmm6,xmm7 phaddw xmm6,xmm7 phaddw xmm6,xmm7 ;xmm6[15:0] = final word sum movd edx,xmm6 add ebx,edx ;ebx = num_masked_pixels ; Reset intermediate values xor edx,edx pxor xmm2,xmm2 pxor xmm3,xmm3 pxor xmm6,xmm6 ret
build
nasm -f elf32 -o ssepackedintegerthreshold.o ssepackedintegerthreshold.asm