// // Copyright (c) 2017 The Khronos Group Inc. // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // You may obtain a copy of the License at // // http://www.apache.org/licenses/LICENSE-2.0 // // Unless required by applicable law or agreed to in writing, software // distributed under the License is distributed on an "AS IS" BASIS, // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. // See the License for the specific language governing permissions and // limitations under the License. // #include #include #include "harness/testHarness.h" #include "harness/typeWrappers.h" #include #include "procs.h" #include "utils.h" #include #ifdef CL_VERSION_2_0 extern int gWimpyMode; static const char* multi_queue_simple_block1[] = { NL, "void block_fn(size_t tid, int mul, __global int* res)" NL, "{" NL, " res[tid] = mul * 7 - 21;" NL, "}" NL, "" NL, "kernel void multi_queue_simple_block1(__global int* res)" NL, "{" NL, " int multiplier = 3;" NL, " size_t tid = get_global_id(0);" NL, "" NL, " void (^kernelBlock)(void) = ^{ block_fn(tid, multiplier, res); };" NL, "" NL, " res[tid] = -1;" NL, " queue_t def_q = get_default_queue();" NL, " ndrange_t ndrange = ndrange_1D(1);" NL, " int enq_res = enqueue_kernel(def_q, CLK_ENQUEUE_FLAGS_WAIT_KERNEL, ndrange, kernelBlock);" NL, " if(enq_res != CLK_SUCCESS) { res[tid] = -1; return; }" NL, "}" NL }; static const char* multi_queue_simple_block2[] = { NL, "void block_fn(size_t tid, int mul, __global int* res)" NL, "{" NL, " res[tid] = mul * 7 - 21;" NL, "}" NL, "" NL, "kernel void multi_queue_simple_block2(__global int* res)" NL, "{" NL, " int multiplier = 3;" NL, " size_t tid = get_global_id(0);" NL, "" NL, " void (^kernelBlock)(void) = ^{ block_fn(tid, multiplier, res); };" NL, "" NL, " res[tid] = -1;" NL, " queue_t def_q = get_default_queue();" NL, " ndrange_t ndrange = ndrange_1D(1);" NL, " int enq_res = enqueue_kernel(def_q, CLK_ENQUEUE_FLAGS_WAIT_KERNEL, ndrange, kernelBlock);" NL, " if(enq_res != CLK_SUCCESS) { res[tid] = -1; return; }" NL, "}" NL }; static const char* multi_queue_simple_block3[] = { NL, "void block_fn(size_t tid, int mul, __global int* res)" NL, "{" NL, " res[tid] = mul * 7 - 21;" NL, "}" NL, "" NL, "kernel void multi_queue_simple_block3(__global int* res)" NL, "{" NL, " int multiplier = 3;" NL, " size_t tid = get_global_id(0);" NL, "" NL, " void (^kernelBlock)(void) = ^{ block_fn(tid, multiplier, res); };" NL, "" NL, " res[tid] = -1;" NL, " queue_t def_q = get_default_queue();" NL, " ndrange_t ndrange = ndrange_1D(1);" NL, " int enq_res = enqueue_kernel(def_q, CLK_ENQUEUE_FLAGS_WAIT_KERNEL, ndrange, kernelBlock);" NL, " if(enq_res != CLK_SUCCESS) { res[tid] = -1; return; }" NL, "}" NL }; static const char* multi_queue_simple_block4[] = { NL, "void block_fn(size_t tid, int mul, __global int* res)" NL, "{" NL, " res[tid] = mul * 7 - 21;" NL, "}" NL, "" NL, "kernel void multi_queue_simple_block4(__global int* res)" NL, "{" NL, " int multiplier = 3;" NL, " size_t tid = get_global_id(0);" NL, "" NL, " void (^kernelBlock)(void) = ^{ block_fn(tid, multiplier, res); };" NL, "" NL, " res[tid] = -1;" NL, " queue_t def_q = get_default_queue();" NL, " ndrange_t ndrange = ndrange_1D(1);" NL, " int enq_res = enqueue_kernel(def_q, CLK_ENQUEUE_FLAGS_WAIT_KERNEL, ndrange, kernelBlock);" NL, " if(enq_res != CLK_SUCCESS) { res[tid] = -1; return; }" NL, "}" NL }; static const kernel_src sources_multi_queue_block[] = { KERNEL(multi_queue_simple_block1), KERNEL(multi_queue_simple_block2), KERNEL(multi_queue_simple_block3), KERNEL(multi_queue_simple_block4), }; static const size_t num_kernels_multi_queue_block = arr_size(sources_multi_queue_block); int test_host_multi_queue(cl_device_id device, cl_context context, cl_command_queue queue, int num_elements) { cl_uint i; cl_int err_ret, res = 0; clCommandQueueWrapper dev_queue; cl_int kernel_results[MAX_GWS] = {0}; size_t ret_len; cl_uint max_queues = 1; cl_uint maxQueueSize = 0; err_ret = clGetDeviceInfo(device, CL_DEVICE_QUEUE_ON_DEVICE_MAX_SIZE, sizeof(maxQueueSize), &maxQueueSize, 0); test_error(err_ret, "clGetDeviceInfo(CL_DEVICE_QUEUE_ON_DEVICE_MAX_SIZE) failed"); err_ret = clGetDeviceInfo(device, CL_DEVICE_MAX_ON_DEVICE_QUEUES, sizeof(max_queues), &max_queues, &ret_len); test_error(err_ret, "clGetDeviceInfo(CL_DEVICE_MAX_ON_DEVICE_QUEUES) failed"); size_t max_local_size = 1; err_ret = clGetDeviceInfo(device, CL_DEVICE_MAX_WORK_GROUP_SIZE, sizeof(max_local_size), &max_local_size, &ret_len); test_error(err_ret, "clGetDeviceInfo(CL_DEVICE_MAX_WORK_GROUP_SIZE) failed"); cl_queue_properties queue_prop_def[] = { CL_QUEUE_PROPERTIES, CL_QUEUE_OUT_OF_ORDER_EXEC_MODE_ENABLE|CL_QUEUE_ON_DEVICE|CL_QUEUE_ON_DEVICE_DEFAULT, CL_QUEUE_SIZE, maxQueueSize, 0 }; dev_queue = clCreateCommandQueueWithProperties(context, device, queue_prop_def, &err_ret); test_error(err_ret, "clCreateCommandQueueWithProperties(CL_QUEUE_DEVICE|CL_QUEUE_DEFAULT) failed"); cl_uint n = num_kernels_multi_queue_block; // Number of host queues std::vector queues(n); std::vector q(n); std::vector program(n); std::vector kernel(n); std::vector mem(n); std::vector event(n); for(i = 0; i < n; ++i) { queues[i] = clCreateCommandQueueWithProperties(context, device, NULL, &err_ret); if(check_error(err_ret, "clCreateCommandQueueWithProperties() failed")) { res = -1; break; } q[i] = queues[i]; } if(err_ret == CL_SUCCESS) { for(i = 0; i < n; ++i) { size_t global = MAX_GWS; if(gWimpyMode) { global = 16; } err_ret |= create_single_kernel_helper( context, &program[i], &kernel[i], sources_multi_queue_block[i].num_lines, sources_multi_queue_block[i].lines, sources_multi_queue_block[i].kernel_name); if(check_error(err_ret, "Create single kernel failed")) { res = -1; break; } mem[i] = clCreateBuffer(context, CL_MEM_READ_WRITE | CL_MEM_COPY_HOST_PTR, sizeof(kernel_results), kernel_results, &err_ret); if(check_error(err_ret, "clCreateBuffer() failed")) { res = -1; break; } err_ret |= clSetKernelArg(kernel[i], 0, sizeof(cl_mem), &mem[i]); if(check_error(err_ret, "clSetKernelArg(0) failed")) { res = -1; break; } err_ret |= clEnqueueNDRangeKernel(q[i], kernel[i], 1, NULL, &global, 0, 0, NULL, &event[i]); if(check_error(err_ret, "clEnqueueNDRangeKernel() failed")) { res = -1; break; } } } if(err_ret == CL_SUCCESS) { for(i = 0; i < n; ++i) { cl_int status; err_ret = clEnqueueReadBuffer(q[i], mem[i], CL_TRUE, 0, sizeof(kernel_results), kernel_results, 0, NULL, NULL); if(check_error(err_ret, "clEnqueueReadBuffer() failed")) { res = -1; break; } err_ret = clGetEventInfo(event[i], CL_EVENT_COMMAND_EXECUTION_STATUS, sizeof(status), &status, &ret_len); if(check_error(err_ret, "clGetEventInfo() failed")) { res = -1; break; } #if CL_COMPLETE != CL_SUCCESS #error Fix me! #endif // This hack is possible because both CL_COMPLETE and CL_SUCCESS defined as 0x00 if(check_error(status, "Kernel execution status %d", status)) { err_ret = status; res = -1; break; } else if(kernel_results[0] != 0 && check_error(-1, "'%s' kernel results validation failed = %d", sources_multi_queue_block[i].kernel_name, kernel_results[0])) { res = -1; break; } } } return res; } #endif