Merge branch 'cl20_trunk'

test_conformance/CMakeLists.txt

@@ -65,6 +65,9 @@ set(CSV_FILES
     opencl_conformance_tests_full_no_math_or_conversions.csv
     opencl_conformance_tests_math.csv
     opencl_conformance_tests_quick.csv
+    opencl_conformance_tests_20_full.csv
+    opencl_conformance_tests_20_full_no_math_or_conversions.csv
+    opencl_conformance_tests_20_quick.csv
 )
 set(PY_FILES
     run_conformance.py

test_conformance/api/test_api_min_max.c

@@ -1327,14 +1327,19 @@ int test_min_max_constant_buffer_size(cl_device_id deviceID, cl_context context,
 
     log_info("Reported max constant buffer size of %lld bytes.\n", maxSize);
 
+    // Limit test buffer size to 1/8 of CL_DEVICE_GLOBAL_MEM_SIZE
     error = clGetDeviceInfo(deviceID, CL_DEVICE_GLOBAL_MEM_SIZE, sizeof(maxGlobalSize), &maxGlobalSize, 0);
     test_error(error, "Unable to get CL_DEVICE_GLOBAL_MEM_SIZE");
+
     if (maxSize > maxGlobalSize / 8)
         maxSize = maxGlobalSize / 8;
+
     error = clGetDeviceInfo(deviceID, CL_DEVICE_MAX_MEM_ALLOC_SIZE , sizeof(maxAllocSize), &maxAllocSize, 0);
     test_error(error, "Unable to get CL_DEVICE_MAX_MEM_ALLOC_SIZE ");
+    
     if (maxSize > maxAllocSize)
         maxSize = maxAllocSize;
+    
     /* Create a kernel to test with */
     if( create_single_kernel_helper( context, &program, &kernel, 1, sample_const_arg_kernel, "sample_test" ) != 0 )
     {
@@ -1426,7 +1431,7 @@ int test_min_max_constant_buffer_size(cl_device_id deviceID, cl_context context,
 
     if (allocPassed) {
         if (currentSize < maxSize/PASSING_FRACTION) {
-            log_error("Failed to allocate at least 1/4 of the reported constant size.\n");
+            log_error("Failed to allocate at least 1/8 of the reported constant size.\n");
             return -1;
         } else if (currentSize != maxSize) {
             log_info("Passed at reduced size. (%lld of %lld bytes)\n", currentSize, maxSize);

test_conformance/basic/test_constant.c

@@ -111,14 +111,20 @@ test_constant(cl_device_id device, cl_context context, cl_command_queue queue, i
     test_error( err, "Unable to get max constant buffer size" );
 
   log_info("Device reports CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE %llu bytes.\n", maxSize);
+  
+  // Limit test buffer size to 1/4 of CL_DEVICE_GLOBAL_MEM_SIZE
   err = clGetDeviceInfo(device, CL_DEVICE_GLOBAL_MEM_SIZE, sizeof(maxGlobalSize), &maxGlobalSize, 0);
   test_error(err, "Unable to get CL_DEVICE_GLOBAL_MEM_SIZE");
+
   if (maxSize > maxGlobalSize / 4)
     maxSize = maxGlobalSize / 4;
+
   err = clGetDeviceInfo(device, CL_DEVICE_MAX_MEM_ALLOC_SIZE , sizeof(maxAllocSize), &maxAllocSize, 0);
   test_error(err, "Unable to get CL_DEVICE_MAX_MEM_ALLOC_SIZE ");
+
   if (maxSize > maxAllocSize)
     maxSize = maxAllocSize;
+  
   maxSize/=4;
   num_ints = (size_t)maxSize/sizeof(cl_int);
   num_floats = (size_t)maxSize/sizeof(cl_float);

test_conformance/compatibility/test_conformance/api/test_api_min_max.c

@@ -496,6 +496,7 @@ int test_min_max_mem_alloc_size(cl_device_id deviceID, cl_context context, cl_co
     if (memSize > (cl_ulong)SIZE_MAX) {
       memSize = (cl_ulong)SIZE_MAX;
     }
+
     if( maxAllocSize < requiredAllocSize)
     {
         log_error( "ERROR: Reported max allocation size is less than required %lldMB! (%llu or %lluMB, from a total mem size of %lldMB)\n", (requiredAllocSize / 1024) / 1024, maxAllocSize, (maxAllocSize / 1024)/1024, (memSize / 1024)/1024 );
@@ -503,10 +504,12 @@ int test_min_max_mem_alloc_size(cl_device_id deviceID, cl_context context, cl_co
     }
 
     requiredAllocSize = ((memSize / 4) > (1024 * 1024 * 1024)) ? 1024 * 1024 * 1024 : memSize / 4;
+
     if (gIsEmbedded)
         requiredAllocSize = (requiredAllocSize < 1 * 1024 * 1024) ? 1 * 1024 * 1024 : requiredAllocSize;
     else
     requiredAllocSize = (requiredAllocSize < 128 * 1024 * 1024) ? 128 * 1024 * 1024 : requiredAllocSize;
+
     if( maxAllocSize < requiredAllocSize )
     {
         log_error( "ERROR: Reported max allocation size is less than required of total memory! (%llu or %lluMB, from a total mem size of %lluMB)\n", maxAllocSize, (maxAllocSize / 1024)/1024, (requiredAllocSize / 1024)/1024 );
@@ -519,7 +522,6 @@ int test_min_max_mem_alloc_size(cl_device_id deviceID, cl_context context, cl_co
     if ( memSize < maxAllocSize ) {
         log_info("Global memory size is less than max allocation size, using that.\n");
         maxAllocSize = memSize;
-
     }
 
     minSizeToTry = maxAllocSize/16;
@@ -1320,14 +1322,19 @@ int test_min_max_constant_buffer_size(cl_device_id deviceID, cl_context context,
 
     log_info("Reported max constant buffer size of %lld bytes.\n", maxSize);
 
+    // Limit test buffer size to 1/8 of CL_DEVICE_GLOBAL_MEM_SIZE
     error = clGetDeviceInfo(deviceID, CL_DEVICE_GLOBAL_MEM_SIZE, sizeof(maxGlobalSize), &maxGlobalSize, 0);
     test_error(error, "Unable to get CL_DEVICE_GLOBAL_MEM_SIZE");
+
     if (maxSize > maxGlobalSize / 8)
         maxSize = maxGlobalSize / 8;
+
     error = clGetDeviceInfo(deviceID, CL_DEVICE_MAX_MEM_ALLOC_SIZE , sizeof(maxAllocSize), &maxAllocSize, 0);
     test_error(error, "Unable to get CL_DEVICE_MAX_MEM_ALLOC_SIZE ");
+    
     if (maxSize > maxAllocSize)
         maxSize = maxAllocSize;
+    
     /* Create a kernel to test with */
     if( create_single_kernel_helper( context, &program, &kernel, 1, sample_const_arg_kernel, "sample_test" ) != 0 )
     {
@@ -1418,7 +1425,7 @@ int test_min_max_constant_buffer_size(cl_device_id deviceID, cl_context context,
 
     if (allocPassed) {
         if (currentSize < maxSize/PASSING_FRACTION) {
-            log_error("Failed to allocate at least 1/4 of the reported constant size.\n");
+            log_error("Failed to allocate at least 1/8 of the reported constant size.\n");
             return -1;
         } else if (currentSize != maxSize) {
             log_info("Passed at reduced size. (%lld of %lld bytes)\n", currentSize, maxSize);

test_conformance/compatibility/test_conformance/basic/test_constant.c

@@ -110,14 +110,20 @@ test_constant(cl_device_id device, cl_context context, cl_command_queue queue, i
     test_error( err, "Unable to get max constant buffer size" );
 
   log_info("Device reports CL_DEVICE_MAX_CONSTANT_BUFFER_SIZE %llu bytes.\n", maxSize);
+  
+  // Limit test buffer size to 1/4 of CL_DEVICE_GLOBAL_MEM_SIZE
   err = clGetDeviceInfo(device, CL_DEVICE_GLOBAL_MEM_SIZE, sizeof(maxGlobalSize), &maxGlobalSize, 0);
   test_error(err, "Unable to get CL_DEVICE_GLOBAL_MEM_SIZE");
+
   if (maxSize > maxGlobalSize / 4)
     maxSize = maxGlobalSize / 4;
+
   err = clGetDeviceInfo(device, CL_DEVICE_MAX_MEM_ALLOC_SIZE , sizeof(maxAllocSize), &maxAllocSize, 0);
   test_error(err, "Unable to get CL_DEVICE_MAX_MEM_ALLOC_SIZE ");
+
   if (maxSize > maxAllocSize)
     maxSize = maxAllocSize;
+  
   maxSize/=4;
   num_ints = (size_t)maxSize/sizeof(cl_int);
   num_floats = (size_t)maxSize/sizeof(cl_float);

test_conformance/compatibility/test_conformance/basic/test_hostptr.c

@@ -223,7 +223,7 @@ test_hostptr(cl_device_id device, cl_context context, cl_command_queue queue, in
         log_info("Mapping the CL_MEM_USE_HOST_PTR image with clEnqueueMapImage...\n");
         size_t row_pitch;
         lock_buffer = clEnqueueMapImage(queue, streams[5], CL_TRUE,
-                                        0, origin, region,
+                                        CL_MAP_READ, origin, region,
                                         &row_pitch, NULL,
                                         0, NULL, NULL, &err);
         test_error(err, "clEnqueueMapImage failed");

test_conformance/gl/test_images_read_common.cpp

@@ -253,7 +253,7 @@ int test_cl_image_read( cl_context context, cl_command_queue queue,
 
   programPtr = kernelSource;
   if( create_single_kernel_helper( context, &program, &kernel, 1,
-    (const char **)&programPtr, "sample_test" ) )
+    (const char **)&programPtr, "sample_test", "-cl-std=CL2.0" ) )
   {
     return -1;
   }

test_conformance/opencl_conformance_tests_20_full.csv

@@ -0,0 +1,117 @@
+#
+# OpenCL Conformance Test Suite (full version)
+#
+
+# #########################################
+# Basic Information on the compute device
+# #########################################
+Compute Info,computeinfo/computeinfo
+
+# #########################################
+# Basic operation tests
+# #########################################
+Basic,basic/test_basic
+API,api/test_api
+Compiler,compiler/test_compiler
+
+# #########################################
+# Common mathematical functions
+# #########################################
+Common Functions,commonfns/test_commonfns
+Geometric Functions,geometrics/test_geometrics
+Relationals,relationals/test_relationals
+
+# #########################################
+# General operation
+# #########################################
+Thread Dimensions,thread_dimensions/test_thread_dimensions full*
+Multiple Device/Context,multiple_device_context/test_multiples
+Atomics,atomics/test_atomics
+Profiling,profiling/test_profiling
+Events,events/test_events
+Allocations (single maximum),allocations/test_allocations single 5 all
+Allocations (total maximum),allocations/test_allocations multiple 5 all 
+VecAlign, vec_align/test_vecalign
+VecStep, vec_step/test_vecstep
+Printf,printf/test_printf
+Device Partitioning,device_partition/test_device_partition
+
+# #########################################
+# Buffers and images
+# #########################################
+Buffers,buffers/test_buffers
+Images (API Info),images/clGetInfo/test_cl_get_info 
+Images (Kernel Methods),images/kernel_image_methods/test_kernel_image_methods 
+Images (Kernel),images/kernel_read_write/test_image_streams CL_FILTER_NEAREST
+Images (Kernel pitch),images/kernel_read_write/test_image_streams use_pitches CL_FILTER_NEAREST
+Images (Kernel max size),images/kernel_read_write/test_image_streams max_images CL_FILTER_NEAREST
+Images (clCopyImage),images/clCopyImage/test_cl_copy_images
+Images (clCopyImage small),images/clCopyImage/test_cl_copy_images small_images
+Images (clCopyImage max size),images/clCopyImage/test_cl_copy_images max_images 
+Images (clReadWriteImage),images/clReadWriteImage/test_cl_read_write_images 
+Images (clReadWriteImage pitch),images/clReadWriteImage/test_cl_read_write_images use_pitches 
+Images (clReadWriteImage max size),images/clReadWriteImage/test_cl_read_write_images max_images 
+Images (clFillImage),images/clFillImage/test_cl_fill_images 
+Images (clFillImage pitch),images/clFillImage/test_cl_fill_images use_pitches 
+Images (clFillImage max size),images/clFillImage/test_cl_fill_images max_images 
+Images (Samplerless),images/samplerlessReads/test_samplerless_reads 
+Images (Samplerless pitch),images/samplerlessReads/test_samplerless_reads use_pitches 
+Images (Samplerless max size),images/samplerlessReads/test_samplerless_reads max_images 
+Mem (Host Flags),mem_host_flags/test_mem_host_flags
+
+# #########################################
+# Headers
+# #########################################
+Headers (cl_typen), headers/test_headers
+Headers (cl.h standalone), headers/test_cl_h
+Headers (cl_platform.h standalone), headers/test_cl_platform_h
+Headers (cl_gl.h standalone), headers/test_cl_gl_h
+Headers (opencl.h standalone), headers/test_opencl_h
+
+# #########################################
+# CPU is required to pass linear and normalized image filtering
+# #########################################
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR),images/kernel_read_write/test_image_streams CL_FILTER_LINEAR
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR pitch),images/kernel_read_write/test_image_streams use_pitches CL_FILTER_LINEAR
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR max size),images/kernel_read_write/test_image_streams max_images CL_FILTER_LINEAR
+
+# #########################################
+# OpenGL/CL interaction
+# #########################################
+OpenCL-GL Sharing,gl/test_gl
+
+# #########################################
+# Thorough math and conversions tests
+# #########################################
+Select,select/test_select
+Conversions,conversions/test_conversions
+Contractions,contractions/contractions
+Math,math_brute_force/bruteforce
+Integer Ops,integer_ops/test_integer_ops
+Half Ops,half/Test_half
+
+# #########################################
+# Compatibility with Previous Versions
+# #########################################
+Basic 1.2,compatibility/test_conformance/basic/test_basic
+API 1.2,compatibility/test_conformance/api/test_api
+
+#####################################
+# OpenCL 2.0 tests
+#####################################
+C11 Atomics,c11_atomics/test_c11_atomics
+Execution Model,device_execution/test_device_execution
+Generic Address Space,generic_address_space/test_generic_address_space
+Non Uniform Work Groups,non_uniform_work_group/test_non_uniform_work_group
+Pipes,pipes/test_pipes
+SVM,SVM/test_SVM
+Workgroups,workgroups/test_workgroups
+
+#########################################
+# Extensions
+#########################################
+SPIR,spir/test_spir
+Mipmaps (Kernel),images/kernel_read_write/test_image_streams test_mipmaps CL_FILTER_NEAREST
+Mipmaps (clCopyImage),images/clCopyImage/test_cl_copy_images test_mipmaps
+Mipmaps (clReadWriteImage),images/clReadWriteImage/test_cl_read_write_images test_mipmaps
+Subgroups,subgroups/test_subgroups

test_conformance/opencl_conformance_tests_20_full_no_math_or_conversions.csv

@@ -0,0 +1,92 @@
+#
+# OpenCL Conformance Test Suite (full version)
+#
+
+# #########################################
+# Basic Information on the compute device
+# #########################################
+Compute Info,computeinfo/computeinfo
+
+# #########################################
+# Basic operation tests
+# #########################################
+Basic,basic/test_basic
+API,api/test_api
+Compiler,compiler/test_compiler
+
+# #########################################
+# Common mathematical functions
+# #########################################
+Common Functions,commonfns/test_commonfns
+Geometric Functions,geometrics/test_geometrics
+Relationals,relationals/test_relationals
+
+# #########################################
+# General operation
+# #########################################
+Thread Dimensions,thread_dimensions/test_thread_dimensions full*
+Multiple Device/Context,multiple_device_context/test_multiples
+Atomics,atomics/test_atomics
+Profiling,profiling/test_profiling
+Events,events/test_events
+Allocations (single maximum),allocations/test_allocations single 5 all
+Allocations (total maximum),allocations/test_allocations multiple 5 all 
+VecAlign, vec_align/test_vecalign
+VecStep, vec_step/test_vecstep
+
+# #########################################
+# Buffers and images
+# #########################################
+Buffers,buffers/test_buffers
+Images (API Info),images/clGetInfo/test_cl_get_info 
+Images (Kernel Methods),images/kernel_image_methods/test_kernel_image_methods 
+Images (Kernel),images/kernel_read_write/test_image_streams CL_FILTER_NEAREST
+Images (Kernel pitch),images/kernel_read_write/test_image_streams use_pitches CL_FILTER_NEAREST
+Images (Kernel max size),images/kernel_read_write/test_image_streams max_images CL_FILTER_NEAREST
+Images (clCopyImage),images/clCopyImage/test_cl_copy_images
+Images (clCopyImage small),images/clCopyImage/test_cl_copy_images small_images
+Images (clCopyImage max size),images/clCopyImage/test_cl_copy_images max_images 
+Images (clReadWriteImage),images/clReadWriteImage/test_cl_read_write_images 
+Images (clReadWriteImage pitch),images/clReadWriteImage/test_cl_read_write_images use_pitches 
+Images (clReadWriteImage max size),images/clReadWriteImage/test_cl_read_write_images max_images 
+Images (clFillImage),images/clFillImage/test_cl_fill_images 
+Images (clFillImage pitch),images/clFillImage/test_cl_fill_images use_pitches 
+Images (clFillImage max size),images/clFillImage/test_cl_fill_images max_images 
+Images (Samplerless),images/samplerlessReads/test_samplerless_reads 
+Images (Samplerless pitch),images/samplerlessReads/test_samplerless_reads use_pitches 
+Images (Samplerless max size),images/samplerlessReads/test_samplerless_reads max_images 
+Mem (Host Flags),mem_host_flags/test_mem_host_flags
+
+# #########################################
+# Headers
+# #########################################
+Headers (cl_typen), headers/test_headers
+Headers (cl.h standalone), headers/test_cl_h
+Headers (cl_platform.h standalone), headers/test_cl_platform_h
+Headers (cl_gl.h standalone), headers/test_cl_gl_h
+Headers (opencl.h standalone), headers/test_opencl_h
+Headers (cl.h standalone C99), headers/test_cl_h_c99
+Headers (cl_platform.h standalone C99), headers/test_cl_platform_h_c99
+Headers (cl_gl.h standalone C99), headers/test_cl_gl_h_c99
+Headers (opencl.h standalone C99), headers/test_opencl_h_c99
+
+# #########################################
+# CPU is required to pass linear and normalized image filtering
+# #########################################
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR),images/kernel_read_write/test_image_streams CL_FILTER_LINEAR
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR pitch),images/kernel_read_write/test_image_streams use_pitches CL_FILTER_LINEAR
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR max size),images/kernel_read_write/test_image_streams max_images CL_FILTER_LINEAR
+
+# #########################################
+# OpenGL/CL interaction
+# #########################################
+OpenCL-GL Sharing,gl/test_gl
+
+# #########################################
+# Thorough math and conversions tests
+# #########################################
+Select,select/test_select
+Contractions,contractions/contractions
+Integer Ops,integer_ops/test_integer_ops
+Half Ops,half/Test_half
+

test_conformance/opencl_conformance_tests_20_quick.csv

@@ -0,0 +1,90 @@
+#
+# OpenCL Conformance Test Suite (quick version)
+# The quick version skips some long-running image tests, runs a shorter math test,
+# and does not run the conversion test.
+#
+
+# #########################################
+# Basic Information on the compute device
+# #########################################
+Compute Info,computeinfo/computeinfo
+
+# #########################################
+# Basic operation tests
+# #########################################
+Basic,basic/test_basic
+API,api/test_api
+Compiler,compiler/test_compiler
+
+# #########################################
+# Common mathematical functions
+# #########################################
+Common Functions,commonfns/test_commonfns
+Geometric Functions,geometrics/test_geometrics
+Relationals,relationals/test_relationals
+
+# #########################################
+# General operation
+# #########################################
+Thread Dimensions,thread_dimensions/test_thread_dimensions quick*
+#Multiple Device/Context,multiple_device_context/test_multiples
+Atomics,atomics/test_atomics
+Profiling,profiling/test_profiling
+Events,events/test_events
+Allocations (single maximum),allocations/test_allocations single 5 all
+Allocations (total maximum),allocations/test_allocations multiple 5 all 
+Printf,printf/test_printf
+Device Partitioning,device_partition/test_device_partition
+
+# #########################################
+# Buffers and images
+# #########################################
+Buffers,buffers/test_buffers
+Images (API Info),images/clGetInfo/test_cl_get_info 
+Images (Kernel Methods),images/kernel_image_methods/test_kernel_image_methods 
+Images (Kernel),images/kernel_read_write/test_image_streams CL_FILTER_NEAREST
+Images (Kernel pitch),images/kernel_read_write/test_image_streams use_pitches CL_FILTER_NEAREST
+Images (Kernel max size),images/kernel_read_write/test_image_streams max_images CL_FILTER_NEAREST
+Images (clCopyImage),images/clCopyImage/test_cl_copy_images
+Images (clCopyImage max size),images/clCopyImage/test_cl_copy_images max_images 
+Images (clReadWriteImage),images/clReadWriteImage/test_cl_read_write_images 
+Images (clReadWriteImage pitch),images/clReadWriteImage/test_cl_read_write_images use_pitches 
+Images (clReadWriteImage max size),images/clReadWriteImage/test_cl_read_write_images max_images 
+Images (clFillImage),images/clFillImage/test_cl_fill_images 
+Images (clFillImage pitch),images/clFillImage/test_cl_fill_images use_pitches 
+Images (clFillImage max size),images/clFillImage/test_cl_fill_images max_images 
+Images (Samplerless),images/samplerlessReads/test_samplerless_reads 
+Images (Samplerless pitch),images/samplerlessReads/test_samplerless_reads use_pitches 
+Images (Samplerless max size),images/samplerlessReads/test_samplerless_reads max_images 
+Mem (Host Flags),mem_host_flags/test_mem_host_flags
+
+# #########################################
+# Headers
+# #########################################
+Headers (cl_typen), headers/test_headers
+Headers (cl.h standalone), headers/test_cl_h
+Headers (cl_platform.h standalone), headers/test_cl_platform_h
+Headers (cl_gl.h standalone), headers/test_cl_gl_h
+Headers (opencl.h standalone), headers/test_opencl_h
+
+# #########################################
+# CPU is required to pass linear and normalized image filtering
+# #########################################
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR),images/kernel_read_write/test_image_streams CL_FILTER_LINEAR
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR pitch),images/kernel_read_write/test_image_streams use_pitches CL_FILTER_LINEAR
+CL_DEVICE_TYPE_CPU, Images (Kernel CL_FILTER_LINEAR max size),images/kernel_read_write/test_image_streams max_images CL_FILTER_LINEAR
+
+# #########################################
+# OpenGL/CL interaction
+# #########################################
+OpenCL-GL Sharing,gl/test_gl
+
+# #########################################
+# Thorough math and conversions tests
+# #########################################
+Select,select/test_select
+#Conversions,conversions/test_conversions
+Contractions,contractions/contractions
+Math,math_brute_force/bruteforce -w
+Integer Ops,integer_ops/test_integer_ops integer_* quick_*
+Half Ops,half/Test_half -w