TEST( testCopyBlock, basic ) {
EasyCL *cl = EasyCL::createForFirstGpuOtherwiseCpu();
float a[] = { 1,2,3,4,
5,6,7,8,
9,10,11,12 };
float b[10];
memset(b, 0, sizeof(float)*10);
CLKernel *kernel = makeBasicKernel( cl );
kernel->in( 12, a )->out( 6, b )->in( ( 3<<10)|4)->in( (0<<10)|1)->in((2<<10)|3);
kernel->localFloats( 2 * 3 );
kernel->run_1d(12,4);
// kernel->run_1d(12,12);
cl->finish();
float expected[] = { 2,3,4,
6,7,8 };
for( int i = 0; i < 2; i++ ) {
for( int j = 0; j < 3; j++ ) {
cout << b[i*3+j] << " ";
EXPECT_EQ( expected[i*3+j], b[i*3+j] );
}
cout << endl;
}
cout << endl;
for( int i = 6; i < 10; i++ ) {
cout << b[i] << " ";
EXPECT_EQ( 0, b[i] );
}
cout << endl;
cout << endl;
kernel->in( 12, a )->out( 6, b )->in( ( 3<<10)|4)->in( (1<<10)|0)->in((2<<10)|3);
kernel->localFloats( 2 * 3 );
// kernel->run_1d(12,4);
kernel->run_1d(12,4);
cl->finish();
float expected2[] = { 5,6,7,
9,10,11 };
for( int i = 0; i < 2; i++ ) {
for( int j = 0; j < 3; j++ ) {
cout << b[i*3+j] << " ";
EXPECT_EQ( expected2[i*3+j], b[i*3+j] );
}
cout << endl;
}
cout << endl;
for( int i = 6; i < 10; i++ ) {
cout << b[i] << " ";
EXPECT_EQ( 0, b[i] );
}
cout << endl;
cout << endl;
delete kernel;
delete cl;
}
TEST(SLOW_testlocal, selfdot_3levels_withoutscratch) {
EasyCL *cl = EasyCL::createForFirstGpuOtherwiseCpu();
CLKernel *kernel = cl->buildKernel("testlocal.cl", "selfdot_ints_withoutscratch");
int workgroupSize = min(512, cl->getMaxWorkgroupSize());
const int numWorkgroups = workgroupSize;
const int level3size = numWorkgroups / 4;
const int N = workgroupSize * numWorkgroups * level3size;
cout << "numworkgroups " << numWorkgroups << " workgroupsize " << workgroupSize << " N " << N << endl;
int *myarray = new int[N];
for(int i = 0; i < N; i++) {
myarray[i] = ((i + 7) * 3) % 5;
}
// Timer timer;
CLWrapper *a1wrapper = cl->wrap(N, myarray);
a1wrapper->copyToDevice();
// timer.timeCheck("copied array to device");
int *second = new int[N];
CLWrapper *secondwrapper = cl->wrap(N, second);
int *a2 = new int[numWorkgroups*level3size];
CLWrapper *a2wrapper = cl->wrap(numWorkgroups * level3size, a2);
kernel->in(a1wrapper);
kernel->out(secondwrapper);
kernel->out(a2wrapper);
kernel->run_1d(N, workgroupSize);
cl->finish();
int *a3 = new int[numWorkgroups];
CLWrapper *a3wrapper = cl->wrap(level3size, a3);
kernel->in(a2wrapper);
kernel->out(secondwrapper);
kernel->out(a3wrapper);
kernel->run_1d(workgroupSize * level3size, workgroupSize);
cl->finish();
int finalSum;
kernel->in(a3wrapper);
kernel->out(secondwrapper);
kernel->out(1, &finalSum);
kernel->run_1d(level3size, level3size);
// timer.timeCheck("finished 3-level reduce");
EXPECT_EQ(-1306309159, finalSum);
delete a1wrapper;
delete a2wrapper;
delete a3wrapper;
delete secondwrapper;
delete[] a3;
delete[] second;
delete[] a2;
delete[]myarray;
delete kernel;
delete cl;
}
TEST(testqueues, defaultqueue) {
cout << "start" << endl;
if(!EasyCL::isOpenCLAvailable()) {
cout << "opencl library not found" << endl;
exit(-1);
}
cout << "found opencl library" << endl;
EasyCL *cl = EasyCL::createForFirstGpuOtherwiseCpu();
CLKernel *kernel = cl->buildKernelFromString(getKernel(), "test", "");
float in[5];
float out[5];
for(int i = 0; i < 5; i++) {
in[i] = i * 3;
out[i] = 0;
}
kernel->input(5, in);
kernel->output(5, out);
size_t global = 5;
size_t local = 5;
kernel->run(1, &global, &local);
cl->finish();
assertEquals(out[0] , 7);
assertEquals(out[1] , 10);
assertEquals(out[2] , 13);
assertEquals(out[3] , 16);
assertEquals(out[4] , 19);
for(int i = 0; i < 5; i++) {
in[i] = i * 3;
out[i] = 0;
}
cout << "cl->queue " << (void *)cl->queue << endl;
cout << "*cl->queue " << (void *)*cl->queue << endl;
cout << "cl->default_queue->queue" << (void *)cl->default_queue->queue << endl;
kernel->input(5, in);
kernel->output(5, out);
kernel->run(cl->default_queue, 1, &global, &local);
cl->finish();
assertEquals(out[0] , 7);
assertEquals(out[1] , 10);
assertEquals(out[2] , 13);
assertEquals(out[3] , 16);
assertEquals(out[4] , 19);
cout << "tests completed ok" << endl;
}
TEST( testCopyLocal, basic ) {
EasyCL *cl = EasyCL::createForFirstGpuOtherwiseCpu();
float a[] = { 1,2,3,4,
5,6,7,8,
9,10,11,12 };
float b[16];
memset(b, 0, sizeof(float)*16);
CLKernel *kernel = makeKernel( cl );
kernel->in( 12, a )->out( 16, b )->in( 12 );
kernel->localFloats( 12 );
kernel->run_1d(12,12);
cl->finish();
for( int i = 0; i < 3; i++ ) {
for( int j = 0; j < 4; j++ ) {
cout << b[i*4+j] << " ";
EXPECT_EQ( i * 4 + j + 1, b[i*4+j] );
}
cout << endl;
}
cout << endl;
for( int i = 12; i < 16; i++ ) {
cout << b[i] << " ";
EXPECT_EQ( 0, b[i] );
}
cout << endl;
delete kernel;
delete cl;
}
// this runs an entire kernel to get one value. Clearly this is going to be pretty slow, but
// at least it's more or less compatible, and comparable, to how cutorch does it
// lgfgs expects a working implementation of this method
float THClStorage_get(THClState *state, const THClStorage *self, long index)
{
//// printf("THClStorage_get\n");
THArgCheck((index >= 0) && (index < self->size), 2, "index out of bounds");
THArgCheck(self->wrapper != 0, 1, "storage argument not initialized, is empty");
// if( self->wrapper->isDeviceDirty() ) {
// if(state->trace) cout << "wrapper->copyToHost()" << endl;
// self->wrapper->copyToHost();
// }
// return self->data[index];
const char *uniqueName = __FILE__ ":get";
EasyCL *cl = self->cl; // cant remember if this is a good idea or not :-P
CLKernel *kernel = 0;
if(cl->kernelExists(uniqueName)) {
kernel = cl->getKernel(uniqueName);
} else {
TemplatedKernel kernelBuilder(cl);
kernel = kernelBuilder.buildKernel( uniqueName, __FILE__, getGetKernelSource(), "THClStorageGet" );
}
float res;
kernel->out(1, &res);
kernel->in(self->wrapper);
kernel->in((int64_t)index);
kernel->run_1d(1, 1);
if(state->addFinish) cl->finish();
return res;
}
// this runs an entire kernel to get one value. Clearly this is going to be pretty slow, but
// at least it's more or less compatible, and comparable, to how cutorch does it
void THClStorage_set(THClState *state, THClStorage *self, long index, float value)
{
//// cout << "set size=" << self->size << " index=" << index << " value=" << value << endl;
THArgCheck((index >= 0) && (index < self->size), 2, "index out of bounds");
THArgCheck(self->wrapper != 0, 1, "storage argument not initialized, is empty");
// if( self->wrapper->isDeviceDirty() ) { // we have to do this, since we're going to copy it all back again
// // although I suppose we could set via a kernel perhaps
// // either way, this function is pretty inefficient right now :-P
// if(state->trace) cout << "wrapper->copyToHost() size " << self->size << endl;
// self->wrapper->copyToHost();
// }
// self->data[index] = value;
// if(state->trace) cout << "wrapper->copyToDevice() size " << self->size << endl;
// self->wrapper->copyToDevice();
const char *uniqueName = __FILE__ ":set";
EasyCL *cl = self->cl; // cant remember if this is a good idea or not :-P
CLKernel *kernel = 0;
if(cl->kernelExists(uniqueName)) {
kernel = cl->getKernel(uniqueName);
} else {
TemplatedKernel kernelBuilder(cl);
kernel = kernelBuilder.buildKernel( uniqueName, __FILE__, getSetKernelSource(), "THClStorageSet" );
}
kernel->inout(self->wrapper);
kernel->in((int64_t)index);
kernel->in(value);
kernel->run_1d(1, 1);
if(state->addFinish) cl->finish();
}
TEST(testClBlas, basic) {
EasyCL *cl = DeepCLGtestGlobals_createEasyCL();
float A[] = {1, 3,
2, 7,
9, 5};
float B[] = {3,
-1};
float C[3];
ClBlasInstance clblasInstance;
CLWrapper *AWrap = cl->wrap(6, A);
CLWrapper *BWrap = cl->wrap(2, B);
CLWrapper *CWrap = cl->wrap(3, C);
AWrap->copyToDevice();
BWrap->copyToDevice();
CWrap->createOnDevice();
ClBlasHelper::Gemm(
cl,
clblasRowMajor,
clblasNoTrans, clblasNoTrans,
3, 2, 1,
1,
AWrap, 0,
BWrap, 0,
0,
CWrap, 0
);
cl->finish();
CWrap->copyToHost();
EXPECT_EQ(0, C[0]);
EXPECT_EQ(-1, C[1]);
EXPECT_EQ(22, C[2]);
cl->finish();
delete CWrap;
delete BWrap;
delete AWrap;
cl->finish();
delete cl;
clblasTeardown();
}
TEST( testMemset, basic ) {
EasyCL *cl = DeepCLGtestGlobals_createEasyCL();
CLKernel *kMemset = 0;
// [[[cog
// import stringify
// stringify.write_kernel2( "kMemset", "cl/memset.cl", "cl_memset", '""' )
// ]]]
// generated using cog, from cl/memset.cl:
const char * kMemsetSource =
"// Copyright Hugh Perkins 2015 hughperkins at gmail\n"
"//\n"
"// This Source Code Form is subject to the terms of the Mozilla Public License,\n"
"// v. 2.0. If a copy of the MPL was not distributed with this file, You can\n"
"// obtain one at http://mozilla.org/MPL/2.0/.\n"
"\n"
"kernel void cl_memset(global float *target, const float value, const int N) {\n"
" #define globalId get_global_id(0)\n"
" if ((int)globalId < N) {\n"
" target[globalId] = value;\n"
" }\n"
"}\n"
"\n"
"";
kMemset = cl->buildKernelFromString(kMemsetSource, "cl_memset", "", "cl/memset.cl");
// [[[end]]]
int N = 10000;
float *myArray = new float[N];
CLWrapper *myArrayWrapper = cl->wrap( N, myArray );
myArrayWrapper->createOnDevice();
kMemset->out( myArrayWrapper )->in( 99.0f )->in( N );
int workgroupSize = 64;
kMemset->run_1d( ( N + workgroupSize - 1 ) / workgroupSize * workgroupSize, workgroupSize );
cl->finish();
myArrayWrapper->copyToHost();
for( int i = 0; i < 10; i++ ) {
// cout << "myArray[" << i << "]=" << myArray[i] << endl;
}
for( int i = 0; i < N; i++ ) {
EXPECT_EQ( 99.0f, myArray[i] );
}
delete kMemset;
delete cl;
}
TEST( testCopyBlock, testPos ) {
EasyCL *cl = EasyCL::createForFirstGpuOtherwiseCpu();
float in[12];
float res[12];
in[0] = ( 3 << 10 ) | 4;
in[1] = 8;
in[2] = 14;
for( int i = 0; i < 3; i++ ) {
cout << "in[" << i << "]=" << in[i] << endl;
}
CLKernel *kernel = makeTestPosKernel( cl );
kernel->in( 12, in )->out( 12, res );
kernel->run_1d(1,1);
cl->finish();
for( int i = 0; i < 5; i++ ) {
cout << "res[" << i << "]=" << res[i] << endl;
}
}
void kernelLaunch_THClTensor_reduceNoncontigDim(
THClState *state,
dim3 &grid,
dim3 &block,
int ADims,
int BDims,
TensorInfo<IndexType> out,
TensorInfo<IndexType> in,
IndexType reductionStride,
IndexType reductionSize,
IndexType totalSlices,
float init,
HasOperator2 const*modifyOp,
HasOperator3 const*reduceOp) {
// cl->finish();
StatefulTimer::timeCheck("Reduce-noncontig START");
std::string uniqueName = "THClTensor_reduceNoncontigDim_" + easycl::toString(ADims) + "_" + easycl::toString(BDims) + "_" +
TypeParseTraits<IndexType>::name + "_" + modifyOp->operator2() + "_" + reduceOp->operator3();
EasyCL *cl = in.wrapper->getCl();
CLKernel *kernel = 0;
if(cl->kernelExists(uniqueName)) {
kernel = cl->getKernel(uniqueName);
StatefulTimer::timeCheck("Apply3 1aa");
} else {
// launch kernel here....
TemplatedKernel kernelBuilder(cl);
std::set<int> dims_set;
if(ADims >= 0) {
dims_set.insert(ADims);
}
if(BDims >= 0) {
dims_set.insert(BDims);
}
std::vector<int> dims;
for( std::set<int>::iterator it = dims_set.begin(); it != dims_set.end(); it++ ) {
dims.push_back(*it);
}
std::string indexType = TypeParseTraits<IndexType>::name;
kernelBuilder
.set("include_THClReduceApplyUtils", THClReduceApplyUtils_getKernelTemplate())
.set("dims", dims)
.set("dim1", ADims)
.set("dim2", BDims)
.set("defreduceblock", 1)
.set("WarpSize", 32) // probably can do like 'if nvidia 32 else 64' ?
.set("MAX_CLTORCH_DIMS", MAX_CLTORCH_DIMS)
.set("IndexType", indexType)
.set("modify_operation", modifyOp->operator2())
.set("reduce_operation", reduceOp->operator3())
;
kernel = kernelBuilder.buildKernel(uniqueName, "THClReduce.cl", THClReduce_getKernelSource(), "THClTensor_reduceNoncontigDim");
}
THClKernels k(state, kernel);
k.out(out);
k.in(in);
k.in((int)reductionStride);
k.in((int)reductionSize);
k.in((int)totalSlices);
k.in(init);
k.run(grid, block);
if(state->addFinish) cl->finish();
StatefulTimer::timeCheck("Reduce-noncontig END");
}