aaron/ispc
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

+1

Browse Source
This commit is contained in:
Evghenii
2014-01-29 16:01:23 +01:00
parent 1c8517947b
commit 3f641f487d
3 changed files with 150 additions and 142 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
examples_ptx/mergeSort/Makefile_cpu
View File

@@ -1,9 +1,11 @@
EXAMPLE=radixSort
CPP_SRC=radixSort.cpp
ISPC_SRC=radixSort.ispc
ISPC_IA_TARGETS=avx1-i32x8
EXAMPLE=mergeSort
CPP_SRC=mergeSort.cpp
ISPC_SRC=mergeSort.ispc
ISPC_IA_TARGETS=avx1-i32x16
ISPC_ARM_TARGETS=neon
#ISPC_FLAGS=-DDEBUG -g
CXXFLAGS=-g
CCFLAGS=-g
include ../common.mk

28
examples_ptx/mergeSort/mergeSort.cpp
View File

@@ -38,7 +38,7 @@ struct Key
int main (int argc, char *argv[])
{
int i, j, n = argc == 1 ? 1000000 : atoi(argv[1]), m = n < 100 ? 1 : 50, l = n < 100 ? n : RAND_MAX;
int i, j, n = argc == 1 ? 1024*1024: atoi(argv[1]), m = n < 100 ? 1 : 50, l = n < 100 ? n : RAND_MAX;
double tISPC1 = 0.0, tISPC2 = 0.0, tSerial = 0.0;
Key *keys = new Key[n];
@@ -46,7 +46,7 @@ int main (int argc, char *argv[])
#pragma omp parallel for
for (int i = 0; i < n; i++)
{
keys[i].key = ((int)(drand48() * (1<<30)));
keys[i].key = i; //((int)(drand48() * (1<<30)));
keys[i].val = i;
}
std::random_shuffle(keys, keys + n);
@@ -57,8 +57,8 @@ int main (int argc, char *argv[])
int *valsBuf = new int[n];
int *keysDst = new int[n];
int *valsDst = new int[n];
int *keysGld = new int [n];
int *valsGld = new int [n];
int *keysGld = new int[n];
int *valsGld = new int[n];
#pragma omp parallel for
for (int i = 0; i < n; i++)
{
@@ -77,8 +77,8 @@ int main (int argc, char *argv[])
tISPC2 = 1e30;
for (i = 0; i < m; i ++)
{
ispcMemcpy(keysSrc, keysGld, n*sizeof(Key));
ispcMemcpy(valsSrc, keysGld, n*sizeof(Key));
ispcMemcpy(keysSrc, keysGld, n*sizeof(int));
ispcMemcpy(valsSrc, keysGld, n*sizeof(int));
reset_and_start_timer();
ispc::mergeSort(keysDst, valsDst, keysBuf, valsBuf, keysSrc, valsSrc, n);
@@ -92,6 +92,22 @@ int main (int argc, char *argv[])
printf("[sort ispc + tasks]:\t[%.3f] msec [%.3f Mpair/s]\n", tISPC2, 1.0e-3*n/tISPC2);
printf("\n Buf: \n");
for (int i = 0 ; i < 128; i++)
{
if ((i % 16) == 0)
printf("\n");
printf(" %d ", keysBuf[i]);
}
printf("\n Dst: \n");
for (int i = 0 ; i < 128; i++)
{
if ((i % 16) == 0)
printf("\n");
printf(" %d ", keysDst[i]);
}
printf("\n");
std::sort(keysGld, keysGld + n);
for (int i = 0; i < n; i++)
assert(keysDst[i] == keysGld[i]);

254
examples_ptx/mergeSort/mergeSort.ispc
View File

@@ -1,45 +1,24 @@
#define SAMPLE_STRIDE programCount
static inline
int iDivUp(int a, int b)
{
int div = a/b;
return ((a % b) == 0) ? div : (div + 1);
}
static inline
uniform int iDivUp(uniform int a, uniform int b)
{
uniform int div = a/b;
return ((a % b) == 0) ? div : (div + 1);
}
static inline
int getSampleCount(int dividend)
{
return iDivUp(dividend, SAMPLE_STRIDE);
}
static inline
uniform int getSampleCount(uniform int dividend)
{
return iDivUp(dividend, SAMPLE_STRIDE);
}
#define iDivUp(a,b) ((a) + (b) - 1)/(b)
#define getSampleCount(dividend) iDivUp((dividend), SAMPLE_STRIDE)
#define W (/*sizeof(int)=*/4 * 8)
static inline
int nextPowerOfTwo(int x)
{
/*
--x;
x |= x >> 1;
x |= x >> 2;
x |= x >> 4;
x |= x >> 8;
x |= x >> 16;
return ++x;
*/
#if 1
--x;
x |= x >> 1;
x |= x >> 2;
x |= x >> 4;
x |= x >> 8;
x |= x >> 16;
return ++x;
#else
return 1U << (W - count_leading_zeros(x - 1));
#endif
}
static inline
@@ -157,9 +136,7 @@ void generateSampleRanksKernel(
uniform int totalProgramCount)
{
const int pos = taskIndex * programCount + programIndex;
if (pos >= totalProgramCount)
return;
assert(pos < totalProgramCount);
const int i = pos & ((stride / SAMPLE_STRIDE) - 1);
const int segmentBase = (pos - i) * (2 * SAMPLE_STRIDE);
@@ -199,7 +176,11 @@ void generateSampleRanks(
uniform int N)
{
uniform int lastSegmentElements = N % (2 * stride);
uniform int threadCount = (lastSegmentElements > stride) ? (N + 2 * stride - lastSegmentElements) / (2 * SAMPLE_STRIDE) : (N - lastSegmentElements) / (2 * SAMPLE_STRIDE);
assert(lastSegmentElements == 0);
uniform int threadCount = (lastSegmentElements > stride) ?
(N + 2 * stride - lastSegmentElements) / (2 * SAMPLE_STRIDE) :
(N - lastSegmentElements) / (2 * SAMPLE_STRIDE);
uniform int nTasks = (threadCount + programCount - 1) / programCount;
launch [nTasks] generateSampleRanksKernel(ranksA, ranksB, srcKey, stride, N, threadCount);
@@ -216,32 +197,30 @@ void mergeRanksAndIndicesKernel(
uniform int N,
uniform int totalProgramCount)
{
int pos = taskIndex * programCount + programIndex;
int pos = taskIndex * programCount + programIndex;
assert(pos < totalProgramCount);
if (pos >= totalProgramCount)
return;
const int i = pos & ((stride / SAMPLE_STRIDE) - 1);
const int segmentBase = (pos - i) * (2 * SAMPLE_STRIDE);
int * ranks = in_Ranks + (pos - i) * 2;
int * limits = in_Limits + (pos - i) * 2;
const int i = pos & ((stride / SAMPLE_STRIDE) - 1);
const int segmentBase = (pos - i) * (2 * SAMPLE_STRIDE);
int * ranks = in_Ranks + (pos - i) * 2;
int * limits = in_Limits + (pos - i) * 2;
const int segmentElementsA = stride;
const int segmentElementsB = min(stride, N - segmentBase - stride);
const int segmentSamplesA = getSampleCount(segmentElementsA);
const int segmentSamplesB = getSampleCount(segmentElementsB);
const int segmentElementsA = stride;
const int segmentElementsB = min(stride, N - segmentBase - stride);
const int segmentSamplesA = getSampleCount(segmentElementsA);
const int segmentSamplesB = getSampleCount(segmentElementsB);
if (i < segmentSamplesA)
{
int dstPos = binarySearchExclusive(ranks[i], ranks + segmentSamplesA, segmentSamplesB, nextPowerOfTwo(segmentSamplesB)) + i;
limits[dstPos] = ranks[i];
}
if (i < segmentSamplesA)
{
int dstPos = binarySearchExclusive(ranks[i], ranks + segmentSamplesA, segmentSamplesB, nextPowerOfTwo(segmentSamplesB)) + i;
limits[dstPos] = ranks[i];
}
if (i < segmentSamplesB)
{
int dstPos = binarySearchInclusive(ranks[segmentSamplesA + i], ranks, segmentSamplesA, nextPowerOfTwo(segmentSamplesA)) + i;
limits[dstPos] = ranks[segmentSamplesA + i];
}
if (i < segmentSamplesB)
{
int dstPos = binarySearchInclusive(ranks[segmentSamplesA + i], ranks, segmentSamplesA, nextPowerOfTwo(segmentSamplesA)) + i;
limits[dstPos] = ranks[segmentSamplesA + i];
}
}
static inline
void mergeRanksAndIndices(
@@ -253,8 +232,12 @@ void mergeRanksAndIndices(
uniform int N)
{
const uniform int lastSegmentElements = N % (2 * stride);
const uniform int threadCount = (lastSegmentElements > stride) ? (N + 2 * stride - lastSegmentElements) / (2 * SAMPLE_STRIDE) : (N - lastSegmentElements) / (2 * SAMPLE_STRIDE);
const uniform int nTasks = (threadCount + programCount -1 ) / programCount;
assert(lastSegmentElements == 0);
const uniform int threadCount = (lastSegmentElements > stride) ?
(N + 2 * stride - lastSegmentElements) / (2 * SAMPLE_STRIDE) :
(N - lastSegmentElements) / (2 * SAMPLE_STRIDE);
const uniform int nTasks = (threadCount + programCount - 1 ) / programCount;
launch [nTasks] mergeRanksAndIndicesKernel(
limitsA,
@@ -324,78 +307,79 @@ void mergeElementaryIntervalsKernel(
uniform int limitsA[],
uniform int limitsB[],
uniform int stride,
uniform int N
)
uniform int N)
{
uniform int s_key[2 * SAMPLE_STRIDE];
uniform int s_val[2 * SAMPLE_STRIDE];
uniform int s_key[2 * SAMPLE_STRIDE];
uniform int s_val[2 * SAMPLE_STRIDE];
const int uniform intervalI = taskIndex & ((2 * stride) / SAMPLE_STRIDE - 1);
const int uniform segmentBase = (taskIndex - intervalI) * SAMPLE_STRIDE;
srcKey += segmentBase;
srcVal += segmentBase;
dstKey += segmentBase;
dstVal += segmentBase;
const int uniform intervalI = taskIndex & ((2 * stride) / SAMPLE_STRIDE - 1);
const int uniform segmentBase = (taskIndex - intervalI) * SAMPLE_STRIDE;
srcKey += segmentBase;
srcVal += segmentBase;
dstKey += segmentBase;
dstVal += segmentBase;
//Set up threadblock-wide parameters
uniform int startSrcA, startSrcB, lenSrcA, lenSrcB, startDstA, startDstB;
//Set up threadblock-wide parameters
uniform int startSrcA, startSrcB, lenSrcA, lenSrcB, startDstA, startDstB;
{
uniform int segmentElementsA = stride;
uniform int segmentElementsB = min(stride, N - segmentBase - stride);
uniform int segmentSamplesA = getSampleCount(segmentElementsA);
uniform int segmentSamplesB = getSampleCount(segmentElementsB);
uniform int segmentSamples = segmentSamplesA + segmentSamplesB;
{
uniform int segmentElementsA = stride;
uniform int segmentElementsB = min(stride, N - segmentBase - stride);
uniform int segmentSamplesA = getSampleCount(segmentElementsA);
uniform int segmentSamplesB = getSampleCount(segmentElementsB);
uniform int segmentSamples = segmentSamplesA + segmentSamplesB;
startSrcA = limitsA[taskIndex];
startSrcB = limitsB[taskIndex];
uniform int endSrcA = (intervalI + 1 < segmentSamples) ? limitsA[taskIndex+ 1] : segmentElementsA;
uniform int endSrcB = (intervalI + 1 < segmentSamples) ? limitsB[taskIndex + 1] : segmentElementsB;
lenSrcA = endSrcA - startSrcA;
lenSrcB = endSrcB - startSrcB;
startDstA = startSrcA + startSrcB;
startDstB = startDstA + lenSrcA;
}
//Load main input data
startSrcA = limitsA[taskIndex];
startSrcB = limitsB[taskIndex];
uniform int endSrcA = (intervalI + 1 < segmentSamples) ? limitsA[taskIndex + 1] : segmentElementsA;
uniform int endSrcB = (intervalI + 1 < segmentSamples) ? limitsB[taskIndex + 1] : segmentElementsB;
lenSrcA = endSrcA - startSrcA;
lenSrcB = endSrcB - startSrcB;
startDstA = startSrcA + startSrcB;
startDstB = startDstA + lenSrcA;
}
if (programIndex < lenSrcA)
{
s_key[programIndex + 0] = srcKey[0 + startSrcA + programIndex];
s_val[programIndex + 0] = srcVal[0 + startSrcA + programIndex];
}
//Load main input data
if (programIndex < lenSrcB)
{
s_key[programIndex + SAMPLE_STRIDE] = srcKey[stride + startSrcB + programIndex];
s_val[programIndex + SAMPLE_STRIDE] = srcVal[stride + startSrcB + programIndex];
}
if (programIndex < lenSrcA)
{
s_key[programIndex + 0] = srcKey[0 + startSrcA + programIndex];
s_val[programIndex + 0] = srcVal[0 + startSrcA + programIndex];
}
//Merge data in shared memory
merge(
s_key,
s_val,
s_key + 0,
s_val + 0,
s_key + SAMPLE_STRIDE,
s_val + SAMPLE_STRIDE,
lenSrcA, SAMPLE_STRIDE,
lenSrcB, SAMPLE_STRIDE
);
//Store merged data
if (programIndex < lenSrcB)
{
s_key[programIndex + SAMPLE_STRIDE] = srcKey[stride + startSrcB + programIndex];
s_val[programIndex + SAMPLE_STRIDE] = srcVal[stride + startSrcB + programIndex];
}
if (programIndex < lenSrcA)
{
dstKey[startDstA + programIndex] = s_key[programIndex];
dstVal[startDstA + programIndex] = s_val[programIndex];
}
//Merge data in shared memory
merge(
s_key,
s_val,
s_key + 0,
s_val + 0,
s_key + SAMPLE_STRIDE,
s_val + SAMPLE_STRIDE,
lenSrcA, SAMPLE_STRIDE,
lenSrcB, SAMPLE_STRIDE
);
if (programIndex < lenSrcB)
{
dstKey[startDstB + programIndex] = s_key[lenSrcA + programIndex];
dstVal[startDstB + programIndex] = s_val[lenSrcA + programIndex];
}
//Store merged data
assert(startDstA < N);
assert(startDstB < N);
if (programIndex < lenSrcA)
{
dstKey[startDstA + programIndex] = s_key[programIndex];
dstVal[startDstA + programIndex] = s_val[programIndex];
}
if (programIndex < lenSrcB)
{
dstKey[startDstB + programIndex] = s_key[lenSrcA + programIndex];
dstVal[startDstB + programIndex] = s_val[lenSrcA + programIndex];
}
}
static inline
void mergeElementaryIntervals(
@@ -455,6 +439,7 @@ void closeMergeSort()
export
void copyKernel(uniform int dst[], uniform int src[], uniform int size)
{
assert(size == 0);
foreach (i = 0 ... size)
dst[i] = src[i];
}
@@ -490,12 +475,15 @@ void mergeSort(
oVal = bufVal;
}
assert(N <= SAMPLE_STRIDE * MAX_SAMPLE_COUNT);
assert(N % (programCount*2) == 0);
mergeSortGang(iKey, iVal, srcKey, srcVal, N/(2*programCount));
for (uniform int stride = 2*programCount; stride < N; stride <<= 1)
{
print ("stride= % N= % \n", stride, N);
uniform int lastSegmentElements = N % (2 * stride);
//Find sample ranks and prepare for limiters merge
@@ -509,19 +497,21 @@ void mergeSort(
if (lastSegmentElements <= stride)
{
assert(0);
//Last merge segment consists of a single array which just needs to be passed through
copyKernel(oKey + (N - lastSegmentElements), iKey + (N - lastSegmentElements), lastSegmentElements);
copyKernel(oVal + (N - lastSegmentElements), iVal + (N - lastSegmentElements), lastSegmentElements);
}
uniform int * uniform tmpKey = iKey;
iKey = oKey;
oKey = tmpKey;
{
uniform int * uniform tmpKey = iKey;
iKey = oKey;
oKey = tmpKey;
}
uniform int * uniform tmpVal = iVal;
iVal = oVal;
oVal = tmpVal;
{
uniform int * uniform tmpVal = iVal;
iVal = oVal;
oVal = tmpVal;
}
}
}