int lineCap ( void)
{
// This function requires calibration per applicaton
const double constanttime = 8.30565e-10;
__s64 time1,time2;
unsigned short sum=0 , i , sixtyVolts, fiftyFiveVolts;
#ifdef PCM_SLIC_FLOAT
double finaltime,capValue;
#endif
// doReset();
// changeCID(0);
// initialize();
writeRam(VCM,0);
writeReg(ILIM,5);
writeRam(SBIAS, 0x61); // TURN OFF SLOW FEED
writeReg(LINEFEED,5);
for (i=0;i<12;i++){
writeRam(VOC,(unsigned short) ((i/2)*TENVOLTINC));
delay(50);
}
writeRam(VOC,6*TENVOLTINC);
delay(50);
sixtyVolts=readRam(VTIP);
writeRam(VOC,11*TENVOLTINC/2);
delay(50);
fiftyFiveVolts=readRam(VTIP);
time1=readTSC(); //record initial time
for(i=0;i<50;i++)
{
writeRam(VOC,6*TENVOLTINC);
while(readRam(VTIP)!=sixtyVolts);
writeRam(VOC,11*TENVOLTINC/2);
while (readRam(VTIP)!=fiftyFiveVolts);
}
time2=readTSC();//record final time
time2-=time1;
writeReg(LINEFEED,0);
delay(50);
printf(" \n Time = %I64i",time2);
#ifdef PCM_SLIC_FLOAT
time2-=beta; //subtract beta, see AN71
finaltime = (double) time2;
capValue = (finaltime/alpha)*1e6;//divide alpha, see AN71
//output capicatance
printf("\n Cap = %f uF ", capValue);
#endif
return(0);
}
__u64 delay( unsigned long wait)
{
//__s64 target = readTSC() + (ticksPerSecond * wait ) /1000 ;
__s64 target ;
while (readTSC() < target) ;
return ( readTSC() - target);
}
__s64 calibratePrecisionClock(void)
{
slic_sleep(1);
time0= readTSC();
slic_sleep (1800);
time1 = readTSC();
//ticksPerSecond=((time1-time0)/1800000)*1000000;
return ticksPerSecond;
}
double secondsPerTSC(void)
{
clock_t clock1;
tsc_type tsc1,tsc2;
double clocksPerSec,MHZ;
if ( haveSecondsPerClock )
{
return secondsPerClock;
}
do
{
clock1 = clock();
readTSC(tsc1);
while( (clock() - clock1) < CLOCKS_PER_SEC ) ;
readTSC(tsc2);
clocksPerSec = diffTSChz(tsc1,tsc2);
MHZ = clocksPerSec / (1000000.0);
// check for common errors
{
int Z50,Z83;
Z50 = ((int)((MHZ + 25)/50))*50; // round to the nearest fifty
Z83 = ((int)(((MHZ + 41)*12)/1000))*1000/12; // round to the nearest 83.333 == 1000/12
if ( ABS((double)Z50 - MHZ) < 14 )
MHZ = Z50;
else if ( ABS((double)Z83 - MHZ) < 14 )
MHZ = Z83;
else
{
MHZ = ((int)((MHZ + 5)/10))*10; // round to the nearest ten
}
TSC_MHZ = (int) MHZ;
}
} while( TSC_MHZ > 1000 );
clocksPerSec = MHZ * (1000000.0);
haveSecondsPerClock = true;
secondsPerClock = 1.0f / clocksPerSec;
return secondsPerClock;
}
void pushTSC(void)
{
tscNode *tn;
tn = new(tscNode);
if ( !tn ) return;
tn->next = head;
head = tn;
readTSC(tn->tsc);
}
double popTSChz(void)
{
tscNode *tn;
ulong tsc[2];
readTSC(tsc);
if ( ! head ) return 0.0;
tn = head;
head = head->next;
return diffTSChz(tn->tsc,tsc);
}
ppm_t* img_fast_brighten(ppm_t* src, float k, uint64_t* cycles)
{
if (k < 0) return src;
int i = 0, j = 0;
uint8_t* r = src->r;
uint8_t* g = src->g;
uint8_t* b = src->b;
int ni = 0;
__m128i xmm0, xmm1, xmm2, xmm4, xmm5, xmm6;
__m128i xmm3 = _mm_set1_epi8((int)ceil(k));
uint64_t t0 = readTSC();
for (j = 0; j < src->w; j++)
{
for (i = 0; i < src->h; i += 16)
{
ni += 16;
// Load 16 RGB pixels
xmm0 = _mm_load_si128((__m128i*)(r + ni));
xmm1 = _mm_load_si128((__m128i*)(g + ni));
xmm2 = _mm_load_si128((__m128i*)(b + ni));
// Multiply by the coefficient vector
xmm4 = _mm_scale_epu8(xmm0, xmm3);
xmm5 = _mm_scale_epu8(xmm1, xmm3);
xmm6 = _mm_scale_epu8(xmm2, xmm3);
// Store back
_mm_store_si128((__m128i*)(r + ni), xmm4);
_mm_store_si128((__m128i*)(g + ni), xmm5);
_mm_store_si128((__m128i*)(b + ni), xmm6);
}
}
if (cycles != NULL) *cycles = cyclesElapsed(readTSC(), t0);
return src;
}
ppm_t* img_fast_bw(ppm_t* src, uint64_t* cycles)
{
int i = 0, j = 0;
uint8_t* r = src->r;
uint8_t* g = src->g;
uint8_t* b = src->b;
int ni = 0;
__m128i xmm0, xmm1, xmm2, xmm3, xmm4, xmm5;
uint64_t t0 = readTSC();
for (j = 0; j < src->w; j++)
{
for (i = 0; i < src->h; i += 16)
{
ni += 16;
// Load 16 RGB pixels
xmm0 = _mm_load_si128((__m128i*)(r + ni));
xmm1 = _mm_load_si128((__m128i*)(g + ni));
xmm2 = _mm_load_si128((__m128i*)(b + ni));
// Average by pairs (RG, GB), then average pairs
xmm3 = _mm_avg_epu8(xmm0, xmm1);
xmm4 = _mm_avg_epu8(xmm1, xmm2);
xmm5 = _mm_avg_epu8(xmm0, xmm1);
// Store back
_mm_store_si128((__m128i*)(r + ni), xmm5);
_mm_store_si128((__m128i*)(g + ni), xmm5);
_mm_store_si128((__m128i*)(b + ni), xmm5);
}
}
if (cycles != NULL) *cycles = cyclesElapsed(readTSC(), t0);
return src;
}
double readTSChz(void)
{
tsc_type tsc;
readTSC(tsc);
return (tsc[0]*4294967296.0 + (double)tsc[1]);
}
/*
** Function: SYSTEM_GetTime
*/
int time_GetTimeWrapper (void *hTimer, void *time){
((timeStamp *)time)->time = readTSC();
return 0;
}
/*
** Function: SYSTEM_TimeElapsed
*/
int time_TimeElapsedWrapper (void *hTimer, void *startTime, int *timeInMs){
_int64 diff = readTSC() - ((timeStamp *)startTime)->time;
*timeInMs = (int)((int)diff / ((int)((systemTimer_S *)hTimer)->ticksPerSecond/1000));
//startTime->time = readTSC();
return 0;
}
ppm_t* img_fast_sharpen(ppm_t* src, ppm_t* dst, float k, uint64_t* cycles)
{
int i = 0, j = 0;
float temp = 0.0f;
float PSF[12] __attribute__((aligned(16)))
= {-k/8.0f, -k/8.0f, -k/8.0f, 0.0f, -k/8.0f, k+1.0f, -k/8.0f, 0.0f, -k/8.0f, -k/8.0f, -k/8.0f, 0.0f};
float tempr[4] __attribute__((aligned(16)))
= { 0.0f };
int mask[4] __attribute__((aligned(16)))
= { 0xFFFF0000, 0xFFFF0000, 0xFFFF0000, 0xFFFF0000 };
__m128 xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9;
__m128i xmm11;
memcpy(dst->r, src->r, src->h*src->w*sizeof(uint8_t));
memcpy(dst->g, src->g, src->h*src->w*sizeof(uint8_t));
memcpy(dst->b, src->b, src->h*src->w*sizeof(uint8_t));
uint64_t t0 = readTSC();
_mm_empty();
xmm0 = _mm_load_ps(PSF);
xmm1 = _mm_load_ps(PSF + 4);
xmm2 = _mm_load_ps(PSF + 8);
// Skip first and last row, no neighbors to convolve with
for (j = 1; j < src->w - 1; j++)
{
// Skip first and last column, no neighbors to convolve with
for (i = 1; i < src->h - 1; i++)
{
temp = 0.0f;
// Load 9 values
//xmm3 = _mm_cvtpu8_ps(*(__m64*)&src->r[i + (j-1)*src->h - 1]);
//xmm4 = _mm_cvtpu8_ps(*(__m64*)&src->r[i + (j)*src->h - 1]);
//xmm5 = _mm_cvtpu8_ps(*(__m64*)&src->r[i + (j+1)*src->h - 1]);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->r[i + (j-1)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm3 = _mm_cvtepi32_ps(xmm11);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->r[i + (j)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm4 = _mm_cvtepi32_ps(xmm11);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->r[i + (j+1)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm5 = _mm_cvtepi32_ps(xmm11);
// Multiply
xmm6 = _mm_mul_ps(xmm0, xmm3);
xmm7 = _mm_mul_ps(xmm1, xmm4);
xmm8 = _mm_mul_ps(xmm2, xmm5);
// Add
xmm9 = _mm_add_ps(xmm6, _mm_add_ps(xmm7, xmm8));
_mm_store_ps(tempr, xmm9);
temp = tempr[0] + tempr[1] + tempr[2];
if (temp < 0.0f) temp = 0.0f;
if (temp > (float)src->max) temp = (float)src->max;
dst->r[i + src->h*j] = (uint8_t)temp;
temp = 0.0f;
// Load 9 values
//xmm3 = _mm_cvtpu8_ps(*(__m64*)&src->g[i + (j-1)*src->h - 1]);
//xmm4 = _mm_cvtpu8_ps(*(__m64*)&src->g[i + (j)*src->h - 1]);
//xmm5 = _mm_cvtpu8_ps(*(__m64*)&src->g[i + (j+1)*src->h - 1]);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->g[i + (j-1)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm3 = _mm_cvtepi32_ps(xmm11);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->g[i + (j)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm4 = _mm_cvtepi32_ps(xmm11);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->g[i + (j+1)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm5 = _mm_cvtepi32_ps(xmm11);
// Multiply
xmm6 = _mm_mul_ps(xmm0, xmm3);
xmm7 = _mm_mul_ps(xmm1, xmm4);
xmm8 = _mm_mul_ps(xmm2, xmm5);
// Add
xmm9 = _mm_add_ps(xmm6, _mm_add_ps(xmm7, xmm8));
_mm_store_ps(tempr, xmm9);
temp = tempr[0] + tempr[1] + tempr[2];
if (temp < 0.0f) temp = 0.0f;
if (temp > (float)src->max) temp = (float)src->max;
dst->g[i + src->h*j] = (uint8_t)temp;
temp = 0.0f;
// Load 9 values
//xmm3 = _mm_cvtpu8_ps(*(__m64*)&src->b[i + (j-1)*src->h - 1]);
//xmm4 = _mm_cvtpu8_ps(*(__m64*)&src->b[i + (j)*src->h - 1]);
//xmm5 = _mm_cvtpu8_ps(*(__m64*)&src->b[i + (j+1)*src->h - 1]);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->b[i + (j-1)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm3 = _mm_cvtepi32_ps(xmm11);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->b[i + (j)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm4 = _mm_cvtepi32_ps(xmm11);
xmm11 = _mm_cvtsi32_si128(*(const int*)&src->b[i + (j+1)*src->h - 1]);
xmm11 = _mm_unpacklo_epi8(xmm11, _mm_setzero_si128());
xmm11 = _mm_unpacklo_epi16(xmm11, _mm_setzero_si128());
xmm5 = _mm_cvtepi32_ps(xmm11);
// Multiply
xmm6 = _mm_mul_ps(xmm0, xmm3);
xmm7 = _mm_mul_ps(xmm1, xmm4);
xmm8 = _mm_mul_ps(xmm2, xmm5);
// Add
xmm9 = _mm_add_ps(xmm6, _mm_add_ps(xmm7, xmm8));
_mm_store_ps(tempr, xmm9);
temp = tempr[0] + tempr[1] + tempr[2];
if (temp < 0.0f) temp = 0.0f;
if (temp > (float)src->max) temp = (float)src->max;
dst->b[i + src->h*j] = (uint8_t)temp;
}
}
_mm_empty();
if (cycles != NULL) *cycles = cyclesElapsed(readTSC(), t0);
return dst;
}
ppm_t* img_fast_sharpen_copy(ppm_t* src, ppm_t* dst, float k, uint64_t* cycles)
{
int i = 0, j = 0;
float temp = 0.0f;
float PSF[12] __attribute__((aligned(16)))
= {-k/8.0f, -k/8.0f, -k/8.0f, 0.0f, -k/8.0f, k+1.0f, -k/8.0f, 0.0f, -k/8.0f, -k/8.0f, -k/8.0f, 0.0f};
float tempr[4] __attribute__((aligned(16)))
= { 0.0f };
__m128 xmm0, xmm1, xmm2, xmm3, xmm4, xmm5, xmm6, xmm7, xmm8, xmm9;
// Initialize data
float* copyR = ppm_alloc_aligned_f(src->w, src->h);
float* copyG = ppm_alloc_aligned_f(src->w, src->h);
float* copyB = ppm_alloc_aligned_f(src->w, src->h);
for (j = 0; j < src->w; j++)
{
for (i = 0; i < src->h; i++)
{
copyR[i + src->h*j] = (float)src->r[i + src->h*j];
copyG[i + src->h*j] = (float)src->g[i + src->h*j];
copyB[i + src->h*j] = (float)src->b[i + src->h*j];
dst->r[i + src->h*j] = src->r[i + src->h*j];
dst->g[i + src->h*j] = src->g[i + src->h*j];
dst->b[i + src->h*j] = src->b[i + src->h*j];
}
}
uint64_t t0 = readTSC();
_mm_empty();
xmm0 = _mm_load_ps(PSF);
xmm1 = _mm_load_ps(PSF + 4);
xmm2 = _mm_load_ps(PSF + 8);
// Skip first and last row, no neighbors to convolve with
for (j = 1; j < src->w - 1; j++)
{
// Skip first and last column, no neighbors to convolve with
for (i = 1; i < src->h - 1; i++)
{
temp = 0.0f;
// Load 9 values
if ((i-1) % 4 == 0)
{
xmm3 = _mm_load_ps(©R[i + (j-1)*src->h] - 1);
xmm4 = _mm_load_ps(©R[i + (j)*src->h] - 1);
xmm5 = _mm_load_ps(©R[i + (j+1)*src->h] - 1);
}
else
{
xmm3 = _mm_loadu_ps(©R[i + (j-1)*src->h] - 1);
xmm4 = _mm_loadu_ps(©R[i + (j)*src->h] - 1);
xmm5 = _mm_loadu_ps(©R[i + (j+1)*src->h] - 1);
}
// Multiply
xmm6 = _mm_mul_ps(xmm0, xmm3);
xmm7 = _mm_mul_ps(xmm1, xmm4);
xmm8 = _mm_mul_ps(xmm2, xmm5);
// Add
xmm9 = _mm_add_ps(xmm6, _mm_add_ps(xmm7, xmm8));
_mm_store_ps(tempr, xmm9);
temp = tempr[0] + tempr[1] + tempr[2];
if (temp < 0.0f) temp = 0.0f;
if (temp > (float)src->max) temp = (float)src->max;
dst->r[i + src->h*j] = (uint8_t)temp;
temp = 0.0f;
// Load 9 values
if ((i-1) % 4 == 0)
{
xmm3 = _mm_load_ps(©G[i + (j-1)*src->h] - 1);
xmm4 = _mm_load_ps(©G[i + (j)*src->h] - 1);
xmm5 = _mm_load_ps(©G[i + (j+1)*src->h] - 1);
}
else
{
xmm3 = _mm_loadu_ps(©G[i + (j-1)*src->h] - 1);
xmm4 = _mm_loadu_ps(©G[i + (j)*src->h] - 1);
xmm5 = _mm_loadu_ps(©G[i + (j+1)*src->h] - 1);
}
// Multiply
xmm6 = _mm_mul_ps(xmm0, xmm3);
xmm7 = _mm_mul_ps(xmm1, xmm4);
xmm8 = _mm_mul_ps(xmm2, xmm5);
// Add
xmm9 = _mm_add_ps(xmm6, _mm_add_ps(xmm7, xmm8));
_mm_store_ps(tempr, xmm9);
temp = tempr[0] + tempr[1] + tempr[2];
if (temp < 0.0f) temp = 0.0f;
if (temp > (float)src->max) temp = (float)src->max;
dst->g[i + src->h*j] = (uint8_t)temp;
temp = 0.0f;
// Load 9 values
if ((i-1) % 4 == 0)
{
xmm3 = _mm_load_ps(©B[i + (j-1)*src->h] - 1);
xmm4 = _mm_load_ps(©B[i + (j)*src->h] - 1);
xmm5 = _mm_load_ps(©B[i + (j+1)*src->h] - 1);
}
else
{
xmm3 = _mm_loadu_ps(©B[i + (j-1)*src->h] - 1);
xmm4 = _mm_loadu_ps(©B[i + (j)*src->h] - 1);
xmm5 = _mm_loadu_ps(©B[i + (j+1)*src->h] - 1);
}
// Multiply
xmm6 = _mm_mul_ps(xmm0, xmm3);
xmm7 = _mm_mul_ps(xmm1, xmm4);
xmm8 = _mm_mul_ps(xmm2, xmm5);
// Add
xmm9 = _mm_add_ps(xmm6, _mm_add_ps(xmm7, xmm8));
_mm_store_ps(tempr, xmm9);
temp = tempr[0] + tempr[1] + tempr[2];
if (temp < 0.0f) temp = 0.0f;
if (temp > (float)src->max) temp = (float)src->max;
dst->b[i + src->h*j] = (uint8_t)temp;
}
}
_mm_empty();
if (cycles != NULL) *cycles = cyclesElapsed(readTSC(), t0);
#ifdef __INTEL_COMPILER
_mm_free(copyR);
#else
free(copyR);
#endif
#ifdef __INTEL_COMPILER
_mm_free(copyG);
#else
free(copyG);
#endif
#ifdef __INTEL_COMPILER
_mm_free(copyB);
#else
free(copyB);
#endif
return dst;
}
