• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 //
2 // Copyright (c) 2017 The Khronos Group Inc.
3 //
4 // Licensed under the Apache License, Version 2.0 (the "License");
5 // you may not use this file except in compliance with the License.
6 // You may obtain a copy of the License at
7 //
8 //    http://www.apache.org/licenses/LICENSE-2.0
9 //
10 // Unless required by applicable law or agreed to in writing, software
11 // distributed under the License is distributed on an "AS IS" BASIS,
12 // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 // See the License for the specific language governing permissions and
14 // limitations under the License.
15 //
16 #include "../testBase.h"
17 
18 #if !defined(_WIN32)
19 #include <sys/mman.h>
20 #endif
21 
22 extern cl_mem_flags gMemFlagsToUse;
23 extern int gtestTypesToRun;
24 
25 extern bool validate_float_write_results( float *expected, float *actual, image_descriptor *imageInfo );
26 extern bool validate_half_write_results( cl_half *expected, cl_half *actual, image_descriptor *imageInfo );
27 
28 // Utility function to clamp down image sizes for certain tests to avoid
29 // using too much memory.
reduceImageSizeRange(size_t maxDimSize)30 static size_t reduceImageSizeRange(size_t maxDimSize) {
31   size_t DimSize = maxDimSize/32;
32   if (DimSize < (size_t) 16)
33     return 16;
34   else if (DimSize > (size_t) 128)
35     return 128;
36   else
37     return DimSize;
38 }
39 
reduceImageDepth(size_t maxDepth)40 static size_t reduceImageDepth(size_t maxDepth) {
41   size_t Depth = maxDepth/32;
42   if (Depth < (size_t) 8)
43     return 8;
44   else if (Depth > (size_t) 32)
45     return 32;
46   else
47     return Depth;
48 }
49 
50 const char *write2DArrayKernelSourcePattern =
51 "__kernel void sample_kernel( __global %s%s *input, write_only %s output %s)\n"
52 "{\n"
53 "   int tidX = get_global_id(0), tidY = get_global_id(1), tidZ = get_global_id(2);\n"
54 "%s"
55 "   write_image%s( output, (int4)( tidX, tidY, tidZ, 0 ) %s, input[ offset ]);\n"
56 "}";
57 
58 const char *readwrite2DArrayKernelSourcePattern =
59 "__kernel void sample_kernel( __global %s%s *input, read_write %s output %s)\n"
60 "{\n"
61 "   int tidX = get_global_id(0), tidY = get_global_id(1), tidZ = get_global_id(2);\n"
62 "%s"
63 "   write_image%s( output, (int4)( tidX, tidY, tidZ, 0 ) %s, input[ offset ] );\n"
64 "}";
65 
66 const char *offset2DArrayKernelSource =
67 "   int offset = tidZ*get_image_width(output)*get_image_height(output) + tidY*get_image_width(output) + tidX;\n";
68 
69 const char *offset2DArrayLodKernelSource =
70 "   int width_lod = ( get_image_width(output) >> lod ) ? ( get_image_width(output) >> lod ) : 1;\n"
71 "   int height_lod = ( get_image_height(output) >> lod ) ? ( get_image_height(output) >> lod ) : 1;\n"
72 "   int offset = tidZ*width_lod*height_lod + tidY*width_lod + tidX;\n";
73 
test_write_image_2D_array(cl_device_id device,cl_context context,cl_command_queue queue,cl_kernel kernel,image_descriptor * imageInfo,ExplicitType inputType,MTdata d)74 int test_write_image_2D_array( cl_device_id device, cl_context context, cl_command_queue queue, cl_kernel kernel,
75                         image_descriptor *imageInfo, ExplicitType inputType, MTdata d )
76 {
77     int                 totalErrors = 0;
78 
79     size_t              num_flags   = 0;
80     const cl_mem_flags  *mem_flag_types = NULL;
81     const char *        *mem_flag_names = NULL;
82     const cl_mem_flags  write_only_mem_flag_types[2] = {  CL_MEM_WRITE_ONLY,   CL_MEM_READ_WRITE };
83     const char *        write_only_mem_flag_names[2] = { "CL_MEM_WRITE_ONLY", "CL_MEM_READ_WRITE" };
84     const cl_mem_flags  read_write_mem_flag_types[1] = {  CL_MEM_READ_WRITE};
85     const char *        read_write_mem_flag_names[1] = { "CL_MEM_READ_WRITE"};
86     if(gtestTypesToRun & kWriteTests)
87     {
88         mem_flag_types = write_only_mem_flag_types;
89         mem_flag_names = write_only_mem_flag_names;
90         num_flags      = sizeof( write_only_mem_flag_types ) / sizeof( write_only_mem_flag_types[0] );
91     }
92     else
93     {
94         mem_flag_types = read_write_mem_flag_types;
95         mem_flag_names = read_write_mem_flag_names;
96         num_flags      = sizeof( read_write_mem_flag_types ) / sizeof( read_write_mem_flag_types[0] );
97     }
98 
99     size_t pixelSize = get_pixel_size( imageInfo->format );
100 
101     for( size_t mem_flag_index = 0; mem_flag_index < num_flags; mem_flag_index++ )
102     {
103         int error;
104         size_t threads[3];
105         bool verifyRounding = false;
106         int forceCorrectlyRoundedWrites = 0;
107 
108 #if defined( __APPLE__ )
109         // Require Apple's CPU implementation to be correctly rounded, not just within 0.6
110         if( GetDeviceType(device) == CL_DEVICE_TYPE_CPU )
111             forceCorrectlyRoundedWrites = 1;
112 #endif
113 
114         if( imageInfo->format->image_channel_data_type == CL_HALF_FLOAT )
115             if( DetectFloatToHalfRoundingMode(queue) )
116                 return 1;
117 
118         BufferOwningPtr<char> maxImageUseHostPtrBackingStore, imageValues;
119 
120         create_random_image_data( inputType, imageInfo, imageValues, d );
121 
122         if(!gTestMipmaps)
123         {
124             if( inputType == kFloat && imageInfo->format->image_channel_data_type != CL_FLOAT )
125             {
126                 /* Pilot data for sRGB images */
127                 if(is_sRGBA_order(imageInfo->format->image_channel_order))
128                 {
129                     // We want to generate ints (mostly) in range of the target format which should be [0,255]
130                     // However the range chosen here is [-test_range_ext, 255 + test_range_ext] so that
131                     // it can test some out-of-range data points
132                     const unsigned int test_range_ext = 16;
133                     int formatMin = 0 - test_range_ext;
134                     int formatMax = 255 + test_range_ext;
135                     int pixel_value = 0;
136 
137                     // First, fill with arbitrary floats
138                     for( size_t z = 0; z < imageInfo->arraySize; z++ )
139                     {
140                         for( size_t y = 0; y < imageInfo->height; y++ )
141                         {
142                             float *inputValues = (float *)(char*)imageValues + imageInfo->width * y * 4 + imageInfo->height * imageInfo->width * z * 4;
143                             for( size_t i = 0; i < imageInfo->width * 4; i++ )
144                             {
145                                 pixel_value = random_in_range( formatMin, (int)formatMax, d );
146                                 inputValues[ i ] = (float)(pixel_value/255.0f);
147                             }
148                         }
149                     }
150 
151                     // Throw a few extra test values in there
152                     float *inputValues = (float *)(char*)imageValues;
153                     size_t i = 0;
154                     // Piloting some debug inputs.
155                     inputValues[ i++ ] = -0.5f;
156                     inputValues[ i++ ] = 0.5f;
157                     inputValues[ i++ ] = 2.f;
158                     inputValues[ i++ ] = 0.5f;
159 
160                     // Also fill in the first few vectors with some deliberate tests to determine the rounding mode
161                     // is correct
162                     if( imageInfo->width > 12 )
163                     {
164                         float formatMax = (float)get_format_max_int( imageInfo->format );
165                         inputValues[ i++ ] = 4.0f / formatMax;
166                         inputValues[ i++ ] = 4.3f / formatMax;
167                         inputValues[ i++ ] = 4.5f / formatMax;
168                         inputValues[ i++ ] = 4.7f / formatMax;
169                         inputValues[ i++ ] = 5.0f / formatMax;
170                         inputValues[ i++ ] = 5.3f / formatMax;
171                         inputValues[ i++ ] = 5.5f / formatMax;
172                         inputValues[ i++ ] = 5.7f / formatMax;
173                     }
174                 }
175                 else
176                 {
177                     // First, fill with arbitrary floats
178                     for( size_t z = 0; z < imageInfo->arraySize; z++ )
179                     {
180                         for( size_t y = 0; y < imageInfo->height; y++ )
181                         {
182                             float *inputValues = (float *)(char*)imageValues + imageInfo->width * y * 4 + imageInfo->height * imageInfo->width * z * 4;
183                             for( size_t i = 0; i < imageInfo->width * 4; i++ )
184                                 inputValues[ i ] = get_random_float( -0.1f, 1.1f, d );
185                         }
186                     }
187 
188                     // Throw a few extra test values in there
189                     float *inputValues = (float *)(char*)imageValues;
190                     size_t i = 0;
191                     inputValues[ i++ ] = -0.0000000000009f;
192                     inputValues[ i++ ] = 1.f;
193                     inputValues[ i++ ] = -1.f;
194                     inputValues[ i++ ] = 2.f;
195 
196                     // Also fill in the first few vectors with some deliberate tests to determine the rounding mode
197                     // is correct
198                     if( imageInfo->width > 12 )
199                     {
200                         float formatMax = (float)get_format_max_int( imageInfo->format );
201                         inputValues[ i++ ] = 4.0f / formatMax;
202                         inputValues[ i++ ] = 4.3f / formatMax;
203                         inputValues[ i++ ] = 4.5f / formatMax;
204                         inputValues[ i++ ] = 4.7f / formatMax;
205                         inputValues[ i++ ] = 5.0f / formatMax;
206                         inputValues[ i++ ] = 5.3f / formatMax;
207                         inputValues[ i++ ] = 5.5f / formatMax;
208                         inputValues[ i++ ] = 5.7f / formatMax;
209                         verifyRounding = true;
210                     }
211                 }
212             }
213             else if( inputType == kUInt )
214             {
215                 unsigned int *inputValues = (unsigned int*)(char*)imageValues;
216                 size_t i = 0;
217                 inputValues[ i++ ] = 0;
218                 inputValues[ i++ ] = 65535;
219                 inputValues[ i++ ] = 7271820;
220                 inputValues[ i++ ] = 0;
221             }
222         }
223 
224         // Construct testing sources
225         clProtectedImage protImage;
226         clMemWrapper unprotImage;
227         cl_mem image;
228 
229         if( gMemFlagsToUse == CL_MEM_USE_HOST_PTR )
230         {
231             create_random_image_data( inputType, imageInfo, maxImageUseHostPtrBackingStore, d );
232 
233             unprotImage = create_image_2d_array( context, mem_flag_types[mem_flag_index] | CL_MEM_USE_HOST_PTR, imageInfo->format,
234                                           imageInfo->width, imageInfo->height, imageInfo->arraySize, 0, 0,
235                                           maxImageUseHostPtrBackingStore, &error );
236 
237             if( error != CL_SUCCESS )
238             {
239                 log_error( "ERROR: Unable to create 2D image array of size %ld x %ld x %ld pitch %ld (%s)\n", imageInfo->width, imageInfo->height, imageInfo->arraySize, imageInfo->rowPitch, IGetErrorString( error ) );
240                 return error;
241             }
242 
243             image = (cl_mem)unprotImage;
244         }
245         else // Either CL_MEM_ALLOC_HOST_PTR, CL_MEM_COPY_HOST_PTR or none
246         {
247             // Note: if ALLOC_HOST_PTR is used, the driver allocates memory that can be accessed by the host, but otherwise
248             // it works just as if no flag is specified, so we just do the same thing either way
249             // Note: if the flags is really CL_MEM_COPY_HOST_PTR, we want to remove it, because we don't want to copy any incoming data
250             if( gTestMipmaps )
251             {
252                 cl_image_desc image_desc = {0};
253                 image_desc.image_type = imageInfo->type;
254                 image_desc.num_mip_levels = imageInfo->num_mip_levels;
255                 image_desc.image_width = imageInfo->width;
256                 image_desc.image_height = imageInfo->height;
257                 image_desc.image_array_size = imageInfo->arraySize;
258 
259                 unprotImage = clCreateImage( context, mem_flag_types[mem_flag_index] | ( gMemFlagsToUse & ~(CL_MEM_COPY_HOST_PTR) ),
260                                              imageInfo->format, &image_desc, NULL, &error);
261                 if( error != CL_SUCCESS )
262                 {
263                     log_error( "ERROR: Unable to create %d level 2D image array of size %ld x %ld x %ld (%s, %s)\n", imageInfo->num_mip_levels, imageInfo->width, imageInfo->height, imageInfo->arraySize,
264                                IGetErrorString( error ), mem_flag_names[mem_flag_index] );
265                     return error;
266                 }
267             }
268             else
269             {
270                 unprotImage = create_image_2d_array( context, mem_flag_types[mem_flag_index] | ( gMemFlagsToUse & ~(CL_MEM_COPY_HOST_PTR) ), imageInfo->format,
271                                               imageInfo->width, imageInfo->height, imageInfo->arraySize, 0, 0, imageValues, &error );
272                 if( error != CL_SUCCESS )
273                 {
274                     log_error( "ERROR: Unable to create 2D image array of size %ld x %ld x %ld pitch %ld (%s)\n", imageInfo->width, imageInfo->height, imageInfo->arraySize, imageInfo->rowPitch, IGetErrorString( error ) );
275                     return error;
276                 }
277             }
278             image = unprotImage;
279         }
280 
281         error = clSetKernelArg( kernel, 1, sizeof( cl_mem ), &image );
282         test_error( error, "Unable to set kernel arguments" );
283 
284         size_t width_lod = imageInfo->width, height_lod = imageInfo->height, nextLevelOffset = 0;
285         size_t origin[ 4 ] = { 0, 0, 0, 0 };
286         size_t region[ 3 ] = { imageInfo->width, imageInfo->height, imageInfo->arraySize };
287         size_t resultSize;
288 
289         int num_lod_loops = (gTestMipmaps)? imageInfo->num_mip_levels : 1;
290         for( int lod = 0; lod < num_lod_loops; lod++)
291         {
292             if(gTestMipmaps)
293             {
294                 error = clSetKernelArg( kernel, 2, sizeof( int ), &lod );
295             }
296             // Run the kernel
297             threads[0] = (size_t)width_lod;
298             threads[1] = (size_t)height_lod;
299             threads[2] = (size_t)imageInfo->arraySize;
300 
301             clMemWrapper inputStream;
302 
303             char *imagePtrOffset = imageValues + nextLevelOffset;
304             inputStream =
305                 clCreateBuffer(context, CL_MEM_COPY_HOST_PTR,
306                                get_explicit_type_size(inputType) * 4 * width_lod
307                                    * height_lod * imageInfo->arraySize,
308                                imagePtrOffset, &error);
309             test_error( error, "Unable to create input buffer" );
310 
311             // Set arguments
312             error = clSetKernelArg( kernel, 0, sizeof( cl_mem ), &inputStream );
313             test_error( error, "Unable to set kernel arguments" );
314 
315             error = clEnqueueNDRangeKernel( queue, kernel, 3, NULL, threads, NULL, 0, NULL, NULL );
316             test_error( error, "Unable to run kernel" );
317 
318             // Get results
319             if( gTestMipmaps )
320                 resultSize = width_lod * height_lod *imageInfo->arraySize * pixelSize;
321             else
322                 resultSize = imageInfo->slicePitch *imageInfo->arraySize;
323             clProtectedArray PA(resultSize);
324             char *resultValues = (char *)((void *)PA);
325 
326             if( gDebugTrace )
327                 log_info( "    reading results, %ld kbytes\n", (unsigned long)( resultSize / 1024 ) );
328 
329             origin[3] = lod;
330             region[0] = width_lod;
331             region[1] = height_lod;
332 
333             error = clEnqueueReadImage( queue, image, CL_TRUE, origin, region, gEnablePitch ? imageInfo->rowPitch : 0, gEnablePitch ? imageInfo->slicePitch : 0, resultValues, 0, NULL, NULL );
334             test_error( error, "Unable to read results from kernel" );
335             if( gDebugTrace )
336                 log_info( "    results read\n" );
337 
338             // Validate results element by element
339             char *imagePtr = imageValues + nextLevelOffset;
340             int numTries = 5;
341             for( size_t z = 0, i = 0; z < imageInfo->arraySize; z++ )
342             {
343                 for( size_t y = 0; y < height_lod; y++ )
344                 {
345                     char *resultPtr;
346                     if( gTestMipmaps )
347                         resultPtr = (char *)resultValues + y * width_lod * pixelSize + z * width_lod * height_lod * pixelSize;
348                     else
349                         resultPtr = (char*)resultValues + y * imageInfo->rowPitch + z * imageInfo->slicePitch;
350                     for( size_t x = 0; x < width_lod; x++, i++ )
351                     {
352                         char resultBuffer[ 16 ]; // Largest format would be 4 channels * 4 bytes (32 bits) each
353 
354                         // Convert this pixel
355                         if( inputType == kFloat )
356                             pack_image_pixel( (float *)imagePtr, imageInfo->format, resultBuffer );
357                         else if( inputType == kInt )
358                             pack_image_pixel( (int *)imagePtr, imageInfo->format, resultBuffer );
359                         else // if( inputType == kUInt )
360                             pack_image_pixel( (unsigned int *)imagePtr, imageInfo->format, resultBuffer );
361 
362                         // Compare against the results
363                         if(is_sRGBA_order(imageInfo->format->image_channel_order))
364                         {
365                             // Compare sRGB-mapped values
366                             cl_float expected[4]    = {0};
367                             cl_float* input_values  = (float*)imagePtr;
368                             cl_uchar *actual        = (cl_uchar*)resultPtr;
369                             float max_err           = MAX_lRGB_TO_sRGB_CONVERSION_ERROR;
370                             float err[4]            = {0.0f};
371 
372                             for( unsigned int j = 0; j < get_format_channel_count( imageInfo->format ); j++ )
373                             {
374                                 if(j < 3)
375                                 {
376                                     expected[j] = sRGBmap(input_values[j]);
377                                 }
378                                 else // there is no sRGB conversion for alpha component if it exists
379                                 {
380                                     expected[j] = NORMALIZE(input_values[j], 255.0f);
381                                 }
382 
383                                 err[j] = fabsf( expected[ j ] - actual[ j ] );
384                             }
385 
386                             if ((err[0] > max_err) ||
387                                 (err[1] > max_err) ||
388                                 (err[2] > max_err) ||
389                                 (err[3] > 0)) // there is no conversion for alpha so the error should be zero
390                             {
391                                 log_error( "       Error:     %g %g %g %g\n", err[0], err[1], err[2], err[3]);
392                                 log_error( "       Input:     %g %g %g %g\n", *((float *)imagePtr), *((float *)imagePtr + 1), *((float *)imagePtr + 2), *((float *)imagePtr + 3));
393                                 log_error( "       Expected: %g %g %g %g\n", expected[ 0 ], expected[ 1 ], expected[ 2 ], expected[ 3 ] );
394                                 log_error( "       Actual:   %d %d %d %d\n", actual[ 0 ], actual[ 1 ], actual[ 2 ], actual[ 3 ] );
395                                 return 1;
396                             }
397                         }
398                         else if( imageInfo->format->image_channel_data_type == CL_FLOAT )
399                         {
400                             float *expected = (float *)resultBuffer;
401                             float *actual = (float *)resultPtr;
402 
403                             if( !validate_float_write_results( expected, actual, imageInfo ) )
404                             {
405                                 unsigned int *e = (unsigned int *)resultBuffer;
406                                 unsigned int *a = (unsigned int *)resultPtr;
407                                 log_error( "ERROR: Sample %ld (%ld,%ld,%ld) did not validate! (%s)\n", i, x, y, z, mem_flag_names[ mem_flag_index ] );
408                                 log_error( "       Expected: %a %a %a %a\n", expected[ 0 ], expected[ 1 ], expected[ 2 ], expected[ 3 ] );
409                                 log_error( "       Expected: %08x %08x %08x %08x\n", e[ 0 ], e[ 1 ], e[ 2 ], e[ 3 ] );
410                                 log_error( "       Actual:   %a %a %a %a\n", actual[ 0 ], actual[ 1 ], actual[ 2 ], actual[ 3 ] );
411                                 log_error( "       Actual:   %08x %08x %08x %08x\n", a[ 0 ], a[ 1 ], a[ 2 ], a[ 3 ] );
412                                 totalErrors++;
413                                 if( ( --numTries ) == 0 )
414                                     return 1;
415                             }
416                         }
417                         else if( imageInfo->format->image_channel_data_type == CL_HALF_FLOAT )
418                         {
419                             cl_half *e = (cl_half *)resultBuffer;
420                             cl_half *a = (cl_half *)resultPtr;
421                             if( !validate_half_write_results( e, a, imageInfo ) )
422                             {
423                                 totalErrors++;
424                                 log_error( "ERROR: Sample %ld (%ld,%ld,%ld) did not validate! (%s)\n", i, x, y, z, mem_flag_names[ mem_flag_index ] );
425                                 unsigned short *e = (unsigned short *)resultBuffer;
426                                 unsigned short *a = (unsigned short *)resultPtr;
427                                 log_error( "    Expected: 0x%04x 0x%04x 0x%04x 0x%04x\n", e[ 0 ], e[ 1 ], e[ 2 ], e[ 3 ] );
428                                 log_error( "    Actual:   0x%04x 0x%04x 0x%04x 0x%04x\n", a[ 0 ], a[ 1 ], a[ 2 ], a[ 3 ] );
429                                 if( inputType == kFloat )
430                                 {
431                                     float *p = (float *)imagePtr;
432                                     log_error( "    Source: %a %a %a %a\n", p[ 0 ], p[ 1 ], p[ 2 ], p[ 3 ] );
433                                     log_error( "          : %12.24f %12.24f %12.24f %12.24f\n", p[ 0 ], p[ 1 ], p[ 2 ], p[ 3 ] );
434                                 }
435                                 if( ( --numTries ) == 0 )
436                                     return 1;
437                             }
438                         }
439                         else
440                         {
441                             // Exact result passes every time
442                             if( memcmp( resultBuffer, resultPtr, get_pixel_size( imageInfo->format ) ) != 0 )
443                             {
444                                 // result is inexact.  Calculate error
445                                 int failure = 1;
446                                 float errors[4] = {NAN, NAN, NAN, NAN};
447                                 pack_image_pixel_error( (float *)imagePtr, imageInfo->format, resultBuffer, errors );
448 
449                                 // We are allowed 0.6 absolute error vs. infinitely precise for some normalized formats
450                                 if( 0 == forceCorrectlyRoundedWrites    &&
451                                    (
452                                     imageInfo->format->image_channel_data_type == CL_UNORM_INT8 ||
453                                     imageInfo->format->image_channel_data_type == CL_UNORM_INT_101010 ||
454                                     imageInfo->format->image_channel_data_type == CL_UNORM_INT16 ||
455                                     imageInfo->format->image_channel_data_type == CL_SNORM_INT8 ||
456                                     imageInfo->format->image_channel_data_type == CL_SNORM_INT16
457                                     ))
458                                 {
459                                     if( ! (fabsf( errors[0] ) > 0.6f) && ! (fabsf( errors[1] ) > 0.6f) &&
460                                        ! (fabsf( errors[2] ) > 0.6f) && ! (fabsf( errors[3] ) > 0.6f)  )
461                                         failure = 0;
462                                 }
463 
464 
465                                 if( failure )
466                                 {
467                                     totalErrors++;
468                                     // Is it our special rounding test?
469                                     if( verifyRounding && i >= 1 && i <= 2 )
470                                     {
471                                         // Try to guess what the rounding mode of the device really is based on what it returned
472                                         const char *deviceRounding = "unknown";
473                                         unsigned int deviceResults[8];
474                                         read_image_pixel<unsigned int>( resultPtr, imageInfo, 0, 0, 0, deviceResults, lod);
475                                         read_image_pixel<unsigned int>( resultPtr, imageInfo, 1, 0, 0, &deviceResults[ 4 ], lod );
476 
477                                         if( deviceResults[ 0 ] == 4 && deviceResults[ 1 ] == 4 && deviceResults[ 2 ] == 4 && deviceResults[ 3 ] == 4 &&
478                                            deviceResults[ 4 ] == 5 && deviceResults[ 5 ] == 5 && deviceResults[ 6 ] == 5 && deviceResults[ 7 ] == 5 )
479                                             deviceRounding = "truncate";
480                                         else if( deviceResults[ 0 ] == 4 && deviceResults[ 1 ] == 4 && deviceResults[ 2 ] == 5 && deviceResults[ 3 ] == 5 &&
481                                                 deviceResults[ 4 ] == 5 && deviceResults[ 5 ] == 5 && deviceResults[ 6 ] == 6 && deviceResults[ 7 ] == 6 )
482                                             deviceRounding = "round to nearest";
483                                         else if( deviceResults[ 0 ] == 4 && deviceResults[ 1 ] == 4 && deviceResults[ 2 ] == 4 && deviceResults[ 3 ] == 5 &&
484                                                 deviceResults[ 4 ] == 5 && deviceResults[ 5 ] == 5 && deviceResults[ 6 ] == 6 && deviceResults[ 7 ] == 6 )
485                                             deviceRounding = "round to even";
486 
487                                         log_error( "ERROR: Rounding mode sample (%ld) did not validate, probably due to the device's rounding mode being wrong (%s)\n", i, mem_flag_names[mem_flag_index] );
488                                         log_error( "       Actual values rounded by device: %d %d %d %d %d %d %d %d\n", deviceResults[ 0 ], deviceResults[ 1 ], deviceResults[ 2 ], deviceResults[ 3 ],
489                                                   deviceResults[ 4 ], deviceResults[ 5 ], deviceResults[ 6 ], deviceResults[ 7 ] );
490                                         log_error( "       Rounding mode of device appears to be %s\n", deviceRounding );
491                                         return 1;
492                                     }
493                                     log_error( "ERROR: Sample %d (%d,%d) did not validate!\n", (int)i, (int)x, (int)y );
494                                     switch(imageInfo->format->image_channel_data_type)
495                                     {
496                                         case CL_UNORM_INT8:
497                                         case CL_SNORM_INT8:
498                                         case CL_UNSIGNED_INT8:
499                                         case CL_SIGNED_INT8:
500                                             log_error( "    Expected: 0x%2.2x 0x%2.2x 0x%2.2x 0x%2.2x\n", ((cl_uchar*)resultBuffer)[0], ((cl_uchar*)resultBuffer)[1], ((cl_uchar*)resultBuffer)[2], ((cl_uchar*)resultBuffer)[3] );
501                                             log_error( "    Actual:   0x%2.2x 0x%2.2x 0x%2.2x 0x%2.2x\n", ((cl_uchar*)resultPtr)[0], ((cl_uchar*)resultPtr)[1], ((cl_uchar*)resultPtr)[2], ((cl_uchar*)resultPtr)[3] );
502                                             log_error( "    Error:    %f %f %f %f\n", errors[0], errors[1], errors[2], errors[3] );
503                                             break;
504                                         case CL_UNORM_INT16:
505                                         case CL_SNORM_INT16:
506                                         case CL_UNSIGNED_INT16:
507                                         case CL_SIGNED_INT16:
508 #ifdef CL_SFIXED14_APPLE
509                                         case CL_SFIXED14_APPLE:
510 #endif
511                                             log_error( "    Expected: 0x%4.4x 0x%4.4x 0x%4.4x 0x%4.4x\n", ((cl_ushort*)resultBuffer)[0], ((cl_ushort*)resultBuffer)[1], ((cl_ushort*)resultBuffer)[2], ((cl_ushort*)resultBuffer)[3] );
512                                             log_error( "    Actual:   0x%4.4x 0x%4.4x 0x%4.4x 0x%4.4x\n", ((cl_ushort*)resultPtr)[0], ((cl_ushort*)resultPtr)[1], ((cl_ushort*)resultPtr)[2], ((cl_ushort*)resultPtr)[3] );
513                                             log_error( "    Error:    %f %f %f %f\n", errors[0], errors[1], errors[2], errors[3] );
514                                             break;
515                                         case CL_HALF_FLOAT:
516                                             log_error(
517                                                 "    Expected: 0x%4.4x 0x%4.4x "
518                                                 "0x%4.4x 0x%4.4x\n",
519                                                 ((cl_half *)resultBuffer)[0],
520                                                 ((cl_half *)resultBuffer)[1],
521                                                 ((cl_half *)resultBuffer)[2],
522                                                 ((cl_half *)resultBuffer)[3]);
523                                             log_error(
524                                                 "    Actual:   0x%4.4x 0x%4.4x "
525                                                 "0x%4.4x 0x%4.4x\n",
526                                                 ((cl_half *)resultPtr)[0],
527                                                 ((cl_half *)resultPtr)[1],
528                                                 ((cl_half *)resultPtr)[2],
529                                                 ((cl_half *)resultPtr)[3]);
530                                             log_error( "    Ulps:     %f %f %f %f\n", errors[0], errors[1], errors[2], errors[3] );
531                                             break;
532                                         case CL_UNSIGNED_INT32:
533                                         case CL_SIGNED_INT32:
534                                             log_error( "    Expected: 0x%8.8x 0x%8.8x 0x%8.8x 0x%8.8x\n", ((cl_uint*)resultBuffer)[0], ((cl_uint*)resultBuffer)[1], ((cl_uint*)resultBuffer)[2], ((cl_uint*)resultBuffer)[3] );
535                                             log_error( "    Actual:   0x%8.8x 0x%8.8x 0x%8.8x 0x%8.8x\n", ((cl_uint*)resultPtr)[0], ((cl_uint*)resultPtr)[1], ((cl_uint*)resultPtr)[2], ((cl_uint*)resultPtr)[3] );
536                                             break;
537                                         case CL_FLOAT:
538                                             log_error( "    Expected: %a %a %a %a\n", ((cl_float*)resultBuffer)[0], ((cl_float*)resultBuffer)[1], ((cl_float*)resultBuffer)[2], ((cl_float*)resultBuffer)[3] );
539                                             log_error( "    Actual:   %a %a %a %a\n", ((cl_float*)resultPtr)[0], ((cl_float*)resultPtr)[1], ((cl_float*)resultPtr)[2], ((cl_float*)resultPtr)[3] );
540                                             log_error( "    Ulps:     %f %f %f %f\n", errors[0], errors[1], errors[2], errors[3] );
541                                             break;
542                                     }
543 
544                                     float *v = (float *)(char *)imagePtr;
545                                     log_error( "   src: %g %g %g %g\n", v[ 0 ], v[ 1], v[ 2 ], v[ 3 ] );
546                                     log_error( "      : %a %a %a %a\n", v[ 0 ], v[ 1], v[ 2 ], v[ 3 ] );
547                                     log_error( "   src: %12.24f %12.24f %12.24f %12.24f\n", v[0 ], v[  1], v[ 2 ], v[ 3 ] );
548 
549                                     if( ( --numTries ) == 0 )
550                                         return 1;
551                                 }
552                             }
553                         }
554                         imagePtr += get_explicit_type_size( inputType ) * (( imageInfo->format->image_channel_order == CL_DEPTH ) ? 1 : 4);
555                         resultPtr += get_pixel_size( imageInfo->format );
556                     }
557                 }
558             }
559             {
560                 nextLevelOffset += width_lod*height_lod*imageInfo->arraySize*pixelSize;
561                 width_lod = (width_lod >> 1) ? (width_lod >> 1) : 1;
562                 height_lod = (height_lod >> 1) ? (height_lod >> 1) : 1;
563             }
564         }
565     }
566     // All done!
567     return totalErrors;
568 }
569 
570 
test_write_image_2D_array_set(cl_device_id device,cl_context context,cl_command_queue queue,const cl_image_format * format,ExplicitType inputType,MTdata d)571 int test_write_image_2D_array_set(cl_device_id device, cl_context context,
572                                   cl_command_queue queue,
573                                   const cl_image_format *format,
574                                   ExplicitType inputType, MTdata d)
575 {
576     char programSrc[10240];
577     const char *ptr;
578     const char *readFormat;
579     clProgramWrapper program;
580     clKernelWrapper kernel;
581     const char *KernelSourcePattern = NULL;
582     int error;
583 
584     // Get our operating parameters
585     size_t maxWidth, maxHeight, maxArraySize;
586     cl_ulong maxAllocSize, memSize;
587 
588     image_descriptor imageInfo = { 0x0 };
589 
590     imageInfo.format = format;
591     imageInfo.type = CL_MEM_OBJECT_IMAGE2D_ARRAY;
592     imageInfo.depth = 1;
593     imageInfo.slicePitch = 0;
594 
595     error = clGetDeviceInfo( device, CL_DEVICE_IMAGE2D_MAX_WIDTH, sizeof( maxWidth ), &maxWidth, NULL );
596     error |= clGetDeviceInfo( device, CL_DEVICE_IMAGE2D_MAX_HEIGHT, sizeof( maxHeight ), &maxHeight, NULL );
597     error |= clGetDeviceInfo( device, CL_DEVICE_IMAGE_MAX_ARRAY_SIZE, sizeof( maxArraySize ), &maxArraySize, NULL );
598     error |= clGetDeviceInfo( device, CL_DEVICE_MAX_MEM_ALLOC_SIZE, sizeof( maxAllocSize ), &maxAllocSize, NULL );
599     error |= clGetDeviceInfo( device, CL_DEVICE_GLOBAL_MEM_SIZE, sizeof( memSize ), &memSize, NULL );
600     test_error( error, "Unable to get max image 3D size from device" );
601 
602     if (memSize > (cl_ulong)SIZE_MAX) {
603       memSize = (cl_ulong)SIZE_MAX;
604     }
605 
606     // Determine types
607     if( inputType == kInt )
608         readFormat = "i";
609     else if( inputType == kUInt )
610         readFormat = "ui";
611     else // kFloat
612         readFormat = "f";
613 
614     if(gtestTypesToRun & kWriteTests)
615     {
616         KernelSourcePattern = write2DArrayKernelSourcePattern;
617     }
618     else
619     {
620         KernelSourcePattern = readwrite2DArrayKernelSourcePattern;
621     }
622     // Construct the source
623     // Construct the source
624     sprintf( programSrc,
625              KernelSourcePattern,
626              get_explicit_type_name( inputType ),
627              (format->image_channel_order == CL_DEPTH) ? "" : "4",
628              (format->image_channel_order == CL_DEPTH) ? "image2d_array_depth_t" : "image2d_array_t",
629              gTestMipmaps ? " , int lod" : "",
630              gTestMipmaps ? offset2DArrayLodKernelSource : offset2DArrayKernelSource,
631              readFormat,
632              gTestMipmaps ? ", lod" : "" );
633 
634     ptr = programSrc;
635     error = create_single_kernel_helper(context, &program, &kernel, 1, &ptr,
636                                         "sample_kernel");
637     test_error( error, "Unable to create testing kernel" );
638 
639     // Run tests
640     if( gTestSmallImages )
641     {
642         for( imageInfo.width = 1; imageInfo.width < 13; imageInfo.width++ )
643         {
644             imageInfo.rowPitch = imageInfo.width * get_pixel_size( imageInfo.format );
645             for( imageInfo.height = 1; imageInfo.height < 9; imageInfo.height++ )
646             {
647                 imageInfo.slicePitch = imageInfo.height * imageInfo.rowPitch;
648                 for( imageInfo.arraySize = 2; imageInfo.arraySize < 7; imageInfo.arraySize++ )
649                 {
650                     if( gTestMipmaps )
651                         imageInfo.num_mip_levels = (size_t) random_in_range(2, compute_max_mip_levels(imageInfo.width, imageInfo.height, 0)-1, d);
652 
653                     if( gDebugTrace )
654                         log_info( "   at size %d,%d,%d\n", (int)imageInfo.width, (int)imageInfo.height, (int)imageInfo.arraySize );
655                     int retCode = test_write_image_2D_array( device, context, queue, kernel, &imageInfo, inputType, d );
656                     if( retCode )
657                         return retCode;
658                 }
659             }
660         }
661     }
662     else if( gTestMaxImages )
663     {
664         // Try a specific set of maximum sizes
665         size_t numbeOfSizes;
666         size_t sizes[100][3];
667 
668         get_max_sizes(&numbeOfSizes, 100, sizes, maxWidth, maxHeight, 1, maxArraySize, maxAllocSize, memSize, CL_MEM_OBJECT_IMAGE2D_ARRAY, imageInfo.format, CL_TRUE);
669 
670         for( size_t idx = 0; idx < numbeOfSizes; idx++ )
671         {
672             imageInfo.width = sizes[ idx ][ 0 ];
673             imageInfo.height = sizes[ idx ][ 1 ];
674             imageInfo.arraySize = sizes[ idx ][ 2 ];
675             imageInfo.rowPitch = imageInfo.width * get_pixel_size( imageInfo.format );
676             imageInfo.slicePitch = imageInfo.height * imageInfo.rowPitch;
677             if( gTestMipmaps )
678                 imageInfo.num_mip_levels = (size_t) random_in_range(2, compute_max_mip_levels(imageInfo.width, imageInfo.height, 0)-1, d);
679             log_info("Testing %d x %d x %d\n", (int)imageInfo.width, (int)imageInfo.height, (int)imageInfo.arraySize);
680             int retCode = test_write_image_2D_array( device, context, queue, kernel, &imageInfo, inputType, d );
681             if( retCode )
682                 return retCode;
683         }
684     }
685     else if( gTestRounding )
686     {
687         size_t typeRange = 1 << ( get_format_type_size( imageInfo.format ) * 8 );
688         imageInfo.height = typeRange / 256;
689         imageInfo.width = (size_t)( typeRange / (cl_ulong)imageInfo.height );
690         imageInfo.arraySize = 2;
691 
692         imageInfo.rowPitch = imageInfo.width * get_pixel_size( imageInfo.format );
693         imageInfo.slicePitch = imageInfo.height * imageInfo.rowPitch;
694         int retCode = test_write_image_2D_array( device, context, queue, kernel, &imageInfo, inputType, d );
695         if( retCode )
696             return retCode;
697     }
698     else
699     {
700         for( int i = 0; i < NUM_IMAGE_ITERATIONS; i++ )
701         {
702             int maxWidthRange = (int) reduceImageSizeRange(maxWidth);
703             int maxHeighthRange = (int) reduceImageSizeRange(maxHeight);
704             int maxArraySizeRange = (int) reduceImageDepth(maxArraySize);
705 
706             cl_ulong size, buffSize;
707             // Loop until we get a size that a) will fit in the max alloc size and b) that an allocation of that
708             // image, the result array, plus offset arrays, will fit in the global ram space
709             do
710             {
711                 imageInfo.width = (size_t)random_log_in_range( 16, maxWidthRange, d );
712                 imageInfo.height = (size_t)random_log_in_range( 16, maxHeighthRange, d );
713                 imageInfo.arraySize = (size_t)random_log_in_range( 8, maxArraySizeRange, d );
714 
715                 if(gTestMipmaps)
716                 {
717                     imageInfo.num_mip_levels = (size_t) random_in_range(2,(compute_max_mip_levels(imageInfo.width, imageInfo.height, 0) - 1), d);
718                     //Need to take into account the input buffer size, otherwise we will end up with input buffer that is exceeding MaxAlloc
719                     size = 4 * compute_mipmapped_image_size(imageInfo);
720                     buffSize = size * get_explicit_type_size( inputType );
721                 }
722                 else
723                 {
724                     imageInfo.rowPitch = imageInfo.width * get_pixel_size( imageInfo.format );
725                     imageInfo.slicePitch = imageInfo.height * imageInfo.rowPitch;
726                     if( gEnablePitch )
727                     {
728                         size_t extraWidth = (int)random_log_in_range( 0, 64, d );
729                         imageInfo.rowPitch += extraWidth * get_pixel_size( imageInfo.format );
730 
731                         imageInfo.slicePitch = imageInfo.height * imageInfo.rowPitch;
732                         extraWidth = (int)random_log_in_range( 0, 64, d );
733                         imageInfo.slicePitch += extraWidth * imageInfo.rowPitch;
734                     }
735 
736 
737                     // Image size and buffer size may differ due to different pixel size.
738                     // See creation of buffer at line ~153.
739                     size = (cl_ulong)imageInfo.slicePitch * (cl_ulong)imageInfo.arraySize * 4;
740                     buffSize = (cl_ulong)imageInfo.width * (cl_ulong)imageInfo.height * imageInfo.arraySize * get_explicit_type_size(inputType) * 4;
741                 }
742             } while(  size > maxAllocSize || buffSize > maxAllocSize || ( size * 3 ) > memSize );
743 
744             if( gDebugTrace )
745                 log_info( "   at size %ld,%ld,%ld (pitch %ld, slice %ld) out of %ld,%ld,%ld\n", imageInfo.width, imageInfo.height, imageInfo.arraySize,
746                          imageInfo.rowPitch, imageInfo.slicePitch, maxWidth, maxHeight, maxArraySize );
747 
748             int retCode = test_write_image_2D_array( device, context, queue, kernel, &imageInfo, inputType, d );
749             if( retCode )
750                 return retCode;
751         }
752     }
753 
754     return 0;
755 }
756