ahmed/OpenCL-CTS
1
0
Fork 0
mirror of https://github.com/KhronosGroup/OpenCL-CTS.git synced 2026-03-19 22:19:02 +00:00
Code Packages Projects Releases Wiki Activity
Files
ba2b6d5a249ffb00a4dee1263cca5fc4bb134bce
OpenCL-CTS/test_conformance/images/kernel_read_write/test_iterations.cpp
Kevin Petit d8733efc0f Synchronise with Khronos-private Gitlab branch 
The maintenance of the conformance tests is moving to Github.

This commit contains all the changes that have been done in
Gitlab since the first public release of the conformance tests.

Signed-off-by: Kevin Petit <kevin.petit@arm.com>
2019-03-05 16:23:49 +00:00

947 lines
49 KiB
C++
Raw Blame History

//
// 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 "../testBase.h"
#include <float.h>
#if defined( __APPLE__ )
#include <signal.h>
#include <sys/signal.h>
#include <setjmp.h>
#endif
#define MAX_ERR 0.005f
#define MAX_HALF_LINEAR_ERR 0.3f
extern cl_command_queue queue;
extern cl_context context;
extern bool gDebugTrace, gExtraValidateInfo, gDisableOffsets, gTestSmallImages, gEnablePitch, gTestMaxImages, gTestRounding;
extern cl_device_type gDeviceType;
extern bool gUseKernelSamplers;
extern cl_filter_mode gFilterModeToUse;
extern cl_addressing_mode gAddressModeToUse;
extern uint64_t gRoundingStartValue;
extern cl_mem_flags gMemFlagsToUse;
#define MAX_TRIES 1
#define MAX_CLAMPED 1
const char *read2DKernelSourcePattern =
"__kernel void sample_kernel( read_only image2d_t input,%s __global float *xOffsets, __global float *yOffsets, __global %s4 *results )\n"
"{\n"
"%s"
" int tidX = get_global_id(0), tidY = get_global_id(1);\n"
" int offset = tidY*get_image_width(input) + tidX;\n"
"%s"
" results[offset] = read_image%s( input, imageSampler, coords );\n"
"}";
const char *intCoordKernelSource =
" int2 coords = (int2)( xOffsets[offset], yOffsets[offset]);\n";
const char *floatKernelSource =
" float2 coords = (float2)( (float)( xOffsets[offset] ), (float)( yOffsets[offset] ) );\n";
static const char *samplerKernelArg = " sampler_t imageSampler,";
#define ABS_ERROR( result, expected ) ( fabsf( (float)expected - (float)result ) )
extern void read_image_pixel_float( void *imageData, image_descriptor *imageInfo,
int x, int y, int z, float *outData );
template <class T> int determine_validation_error( void *imagePtr, image_descriptor *imageInfo, image_sampler_data *imageSampler,
T *resultPtr, T * expected, float error,
float x, float y, float xAddressOffset, float yAddressOffset, size_t j, int &numTries, int &numClamped, bool printAsFloat )
{
int actualX, actualY;
int found = debug_find_pixel_in_image( imagePtr, imageInfo, resultPtr, &actualX, &actualY, NULL );
bool clampingErr = false, clamped = false, otherClampingBug = false;
int clampedX, clampedY, ignoreMe;
clamped = get_integer_coords_offset( x, y, 0.f, xAddressOffset, yAddressOffset, 0.0f, imageInfo->width, imageInfo->height, 0, imageSampler, imageInfo, clampedX, clampedY, ignoreMe );
if( found )
{
// Is it a clamping bug?
if( clamped && clampedX == actualX && clampedY == actualY )
{
if( (--numClamped) == 0 )
{
log_error( "ERROR: TEST FAILED: Read is erroneously clamping coordinates for image size %ld x %ld!\n", imageInfo->width, imageInfo->height );
if( printAsFloat )
{
log_error( "Sample %d: coord {%f(%a), %f(%a)} did not validate!\n\tExpected (%g,%g,%g,%g),\n\tgot (%g,%g,%g,%g),\n\terror of %g\n",
(int)j, x, x, y, y, (float)expected[ 0 ], (float)expected[ 1 ], (float)expected[ 2 ], (float)expected[ 3 ],
(float)resultPtr[ 0 ], (float)resultPtr[ 1 ], (float)resultPtr[ 2 ], (float)resultPtr[ 3 ], error );
}
else
{
log_error( "Sample %d: coord {%f(%a), %f(%a)} did not validate!\n\tExpected (%x,%x,%x,%x),\n\tgot (%x,%x,%x,%x)\n",
(int)j, x, x, y, y, (int)expected[ 0 ], (int)expected[ 1 ], (int)expected[ 2 ], (int)expected[ 3 ],
(int)resultPtr[ 0 ], (int)resultPtr[ 1 ], (int)resultPtr[ 2 ], (int)resultPtr[ 3 ] );
}
return 1;
}
clampingErr = true;
otherClampingBug = true;
}
}
if( clamped && !otherClampingBug )
{
// If we are in clamp-to-edge mode and we're getting zeroes, it's possible we're getting border erroneously
if( resultPtr[ 0 ] == 0 && resultPtr[ 1 ] == 0 && resultPtr[ 2 ] == 0 && resultPtr[ 3 ] == 0 )
{
if( (--numClamped) == 0 )
{
log_error( "ERROR: TEST FAILED: Clamping is erroneously returning border color for image size %ld x %ld!\n", imageInfo->width, imageInfo->height );
if( printAsFloat )
{
log_error( "Sample %d: coord {%f(%a), %f(%a)} did not validate!\n\tExpected (%g,%g,%g,%g),\n\tgot (%g,%g,%g,%g),\n\terror of %g\n",
(int)j, x, x, y, y, (float)expected[ 0 ], (float)expected[ 1 ], (float)expected[ 2 ], (float)expected[ 3 ],
(float)resultPtr[ 0 ], (float)resultPtr[ 1 ], (float)resultPtr[ 2 ], (float)resultPtr[ 3 ], error );
}
else
{
log_error( "Sample %d: coord {%f(%a), %f(%a)} did not validate!\n\tExpected (%x,%x,%x,%x),\n\tgot (%x,%x,%x,%x)\n",
(int)j, x, x, y, y, (int)expected[ 0 ], (int)expected[ 1 ], (int)expected[ 2 ], (int)expected[ 3 ],
(int)resultPtr[ 0 ], (int)resultPtr[ 1 ], (int)resultPtr[ 2 ], (int)resultPtr[ 3 ] );
}
return 1;
}
clampingErr = true;
}
}
if( !clampingErr )
{
if( printAsFloat )
{
log_error( "Sample %d: coord {%f(%a), %f(%a)} did not validate!\n\tExpected (%g,%g,%g,%g),\n\tgot (%g,%g,%g,%g), error of %g\n",
(int)j, x, x, y, y, (float)expected[ 0 ], (float)expected[ 1 ], (float)expected[ 2 ], (float)expected[ 3 ],
(float)resultPtr[ 0 ], (float)resultPtr[ 1 ], (float)resultPtr[ 2 ], (float)resultPtr[ 3 ], error );
}
else
{
log_error( "Sample %d: coord {%f(%a), %f(%a)} did not validate!\n\tExpected (%x,%x,%x,%x),\n\tgot (%x,%x,%x,%x)\n",
(int)j, x, x, y, y, (int)expected[ 0 ], (int)expected[ 1 ], (int)expected[ 2 ], (int)expected[ 3 ],
(int)resultPtr[ 0 ], (int)resultPtr[ 1 ], (int)resultPtr[ 2 ], (int)resultPtr[ 3 ] );
}
log_error( "img size %ld,%ld (pitch %ld)", imageInfo->width, imageInfo->height, imageInfo->rowPitch );
if( clamped )
{
log_error( " which would clamp to %d,%d\n", clampedX, clampedY );
}
if( printAsFloat && gExtraValidateInfo)
{
log_error( "Nearby values:\n" );
log_error( "\t%d\t%d\t%d\t%d\n", clampedX - 2, clampedX - 1, clampedX, clampedX + 1 );
for( int yOff = -2; yOff <= 1; yOff++ )
{
float top[ 4 ], real[ 4 ], bot[ 4 ], bot2[ 4 ];
read_image_pixel_float( imagePtr, imageInfo, clampedX - 2 , clampedY + yOff, 0, top );
read_image_pixel_float( imagePtr, imageInfo, clampedX - 1 ,clampedY + yOff, 0, real );
read_image_pixel_float( imagePtr, imageInfo, clampedX, clampedY + yOff, 0, bot );
read_image_pixel_float( imagePtr, imageInfo, clampedX + 1, clampedY + yOff, 0, bot2 );
log_error( "%d\t(%g,%g,%g,%g)",clampedY + yOff, top[0], top[1], top[2], top[3] );
log_error( " (%g,%g,%g,%g)", real[0], real[1], real[2], real[3] );
log_error( " (%g,%g,%g,%g)",bot[0], bot[1], bot[2], bot[3] );
log_error( " (%g,%g,%g,%g)\n",bot2[0], bot2[1], bot2[2], bot2[3] );
}
if( clampedY < 1 )
{
log_error( "Nearby values:\n" );
log_error( "\t%d\t%d\t%d\t%d\n", clampedX - 2, clampedX - 1, clampedX, clampedX + 1 );
for( int yOff = (int)imageInfo->height - 2; yOff <= (int)imageInfo->height + 1; yOff++ )
{
float top[ 4 ], real[ 4 ], bot[ 4 ], bot2[ 4 ];
read_image_pixel_float( imagePtr, imageInfo, clampedX - 2 , clampedY + yOff, 0, top );
read_image_pixel_float( imagePtr, imageInfo, clampedX - 1 ,clampedY + yOff, 0, real );
read_image_pixel_float( imagePtr, imageInfo, clampedX, clampedY + yOff, 0, bot );
read_image_pixel_float( imagePtr, imageInfo, clampedX + 1, clampedY + yOff, 0, bot2 );
log_error( "%d\t(%g,%g,%g,%g)",clampedY + yOff, top[0], top[1], top[2], top[3] );
log_error( " (%g,%g,%g,%g)", real[0], real[1], real[2], real[3] );
log_error( " (%g,%g,%g,%g)",bot[0], bot[1], bot[2], bot[3] );
log_error( " (%g,%g,%g,%g)\n",bot2[0], bot2[1], bot2[2], bot2[3] );
}
}
}
if( imageSampler->filter_mode != CL_FILTER_LINEAR )
{
if( found )
log_error( "\tValue really found in image at %d,%d (%s)\n", actualX, actualY, ( found > 1 ) ? "NOT unique!!" : "unique" );
else
log_error( "\tValue not actually found in image\n" );
}
log_error( "\n" );
numClamped = -1; // We force the clamped counter to never work
if( ( --numTries ) == 0 )
{
return 1;
}
}
return 0;
}
#define CLAMP( _val, _min, _max ) ((_val) < (_min) ? (_min) : (_val) > (_max) ? (_max) : (_val))
static void InitFloatCoords( image_descriptor *imageInfo, image_sampler_data *imageSampler, float *xOffsets, float *yOffsets, float xfract, float yfract, int normalized_coords, MTdata d )
{
size_t i = 0;
if( gDisableOffsets )
{
for( size_t y = 0; y < imageInfo->height; y++ )
{
for( size_t x = 0; x < imageInfo->width; x++, i++ )
{
xOffsets[ i ] = (float) (xfract + (double) x);
yOffsets[ i ] = (float) (yfract + (double) y);
}
}
}
else
{
for( size_t y = 0; y < imageInfo->height; y++ )
{
for( size_t x = 0; x < imageInfo->width; x++, i++ )
{
xOffsets[ i ] = (float) (xfract + (double) ((int) x + random_in_range( -10, 10, d )));
yOffsets[ i ] = (float) (yfract + (double) ((int) y + random_in_range( -10, 10, d )));
}
}
}
if( imageSampler->addressing_mode == CL_ADDRESS_NONE )
{
i = 0;
for( size_t y = 0; y < imageInfo->height; y++ )
{
for( size_t x = 0; x < imageInfo->width; x++, i++ )
{
xOffsets[ i ] = (float) CLAMP( (double) xOffsets[ i ], 0.0, (double) imageInfo->width - 1.0);
yOffsets[ i ] = (float) CLAMP( (double) yOffsets[ i ], 0.0, (double)imageInfo->height - 1.0);
}
}
}
if( normalized_coords )
{
i = 0;
for( size_t y = 0; y < imageInfo->height; y++ )
{
for( size_t x = 0; x < imageInfo->width; x++, i++ )
{
xOffsets[ i ] = (float) ((double) xOffsets[ i ] / (double) imageInfo->width);
yOffsets[ i ] = (float) ((double) yOffsets[ i ] / (double) imageInfo->height);
}
}
}
}
#ifndef MAX
#define MAX( _a, _b ) ((_a) > (_b) ? (_a) : (_b))
#endif
int test_read_image_2D( cl_device_id device, cl_context context, cl_command_queue queue, cl_kernel kernel,
image_descriptor *imageInfo, image_sampler_data *imageSampler,
bool useFloatCoords, ExplicitType outputType, MTdata d )
{
int error;
static int initHalf = 0;
size_t threads[2];
clMemWrapper xOffsets, yOffsets, results;
clSamplerWrapper actualSampler;
BufferOwningPtr<char> maxImageUseHostPtrBackingStore;
// The DataBuffer template class really does use delete[], not free -- IRO
BufferOwningPtr<cl_float> xOffsetValues(malloc(sizeof(cl_float) * imageInfo->width * imageInfo->height));
BufferOwningPtr<cl_float> yOffsetValues(malloc(sizeof(cl_float) * imageInfo->width * imageInfo->height));
if( imageInfo->format->image_channel_data_type == CL_HALF_FLOAT )
if( DetectFloatToHalfRoundingMode(queue) )
return 1;
// generate_random_image_data allocates with malloc, so we use a MallocDataBuffer here
BufferOwningPtr<char> imageValues;
generate_random_image_data( imageInfo, imageValues, d );
if( gDebugTrace )
log_info( " - Creating image %d by %d...\n", (int)imageInfo->width, (int)imageInfo->height );
// Construct testing sources
clProtectedImage protImage;
clMemWrapper unprotImage;
cl_mem image;
if( gMemFlagsToUse == CL_MEM_USE_HOST_PTR )
{
// clProtectedImage uses USE_HOST_PTR, so just rely on that for the testing (via Ian)
// Do not use protected images for max image size test since it rounds the row size to a page size
if (gTestMaxImages) {
generate_random_image_data( imageInfo, maxImageUseHostPtrBackingStore, d );
unprotImage = create_image_2d( context, CL_MEM_READ_ONLY | CL_MEM_USE_HOST_PTR, imageInfo->format,
imageInfo->width, imageInfo->height, ( gEnablePitch ? imageInfo->rowPitch : 0 ),
maxImageUseHostPtrBackingStore, &error );
} else {
error = protImage.Create( context, (cl_mem_flags)(CL_MEM_READ_ONLY), imageInfo->format, imageInfo->width, imageInfo->height );
}
if( error != CL_SUCCESS )
{
log_error( "ERROR: Unable to create 2D image of size %d x %d pitch %d (%s)\n", (int)imageInfo->width, (int)imageInfo->height, (int)imageInfo->rowPitch, IGetErrorString( error ) );
return error;
}
if (gTestMaxImages)
image = (cl_mem)unprotImage;
else
image = (cl_mem)protImage;
}
else if( gMemFlagsToUse == CL_MEM_COPY_HOST_PTR )
{
// Don't use clEnqueueWriteImage; just use copy host ptr to get the data in
unprotImage = create_image_2d( context, CL_MEM_READ_ONLY | CL_MEM_COPY_HOST_PTR, imageInfo->format,
imageInfo->width, imageInfo->height, ( gEnablePitch ? imageInfo->rowPitch : 0 ),
imageValues, &error );
if( error != CL_SUCCESS )
{
log_error( "ERROR: Unable to create 2D image of size %d x %d pitch %d (%s)\n", (int)imageInfo->width, (int)imageInfo->height, (int)imageInfo->rowPitch, IGetErrorString( error ) );
return error;
}
image = unprotImage;
}
else // Either CL_MEM_ALLOC_HOST_PTR or none
{
// Note: if ALLOC_HOST_PTR is used, the driver allocates memory that can be accessed by the host, but otherwise
// it works just as if no flag is specified, so we just do the same thing either way
unprotImage = create_image_2d( context, CL_MEM_READ_ONLY | gMemFlagsToUse, imageInfo->format,
imageInfo->width, imageInfo->height, ( gEnablePitch ? imageInfo->rowPitch : 0 ),
imageValues, &error );
if( error != CL_SUCCESS )
{
log_error( "ERROR: Unable to create 2D image of size %d x %d pitch %d (%s)\n", (int)imageInfo->width, (int)imageInfo->height, (int)imageInfo->rowPitch, IGetErrorString( error ) );
return error;
}
image = unprotImage;
}
if( gMemFlagsToUse != CL_MEM_COPY_HOST_PTR )
{
if( gDebugTrace )
log_info( " - Writing image...\n" );
size_t origin[ 3 ] = { 0, 0, 0 };
size_t region[ 3 ] = { imageInfo->width, imageInfo->height, 1 };
error = clEnqueueWriteImage(queue, image, CL_TRUE,
origin, region, ( gEnablePitch ? imageInfo->rowPitch : 0 ), 0,
imageValues, 0, NULL, NULL);
if (error != CL_SUCCESS)
{
log_error( "ERROR: Unable to write to 2D image of size %d x %d\n", (int)imageInfo->width, (int)imageInfo->height );
return error;
}
}
if( gDebugTrace )
log_info( " - Creating kernel arguments...\n" );
xOffsets = clCreateBuffer( context, (cl_mem_flags)( CL_MEM_COPY_HOST_PTR ), sizeof( cl_float ) * imageInfo->width * imageInfo->height, xOffsetValues, &error );
test_error( error, "Unable to create x offset buffer" );
yOffsets = clCreateBuffer( context, (cl_mem_flags)( CL_MEM_COPY_HOST_PTR ), sizeof( cl_float ) * imageInfo->width * imageInfo->height, yOffsetValues, &error );
test_error( error, "Unable to create y offset buffer" );
results = clCreateBuffer( context, (cl_mem_flags)(CL_MEM_READ_WRITE), get_explicit_type_size( outputType ) * 4 * imageInfo->width * imageInfo->height, NULL, &error );
test_error( error, "Unable to create result buffer" );
// Create sampler to use
actualSampler = clCreateSampler( context, (cl_bool)imageSampler->normalized_coords, imageSampler->addressing_mode, imageSampler->filter_mode, &error );
test_error( error, "Unable to create image sampler" );
// Set arguments
int idx = 0;
error = clSetKernelArg( kernel, idx++, sizeof( cl_mem ), &image );
test_error( error, "Unable to set kernel arguments" );
if( !gUseKernelSamplers )
{
error = clSetKernelArg( kernel, idx++, sizeof( cl_sampler ), &actualSampler );
test_error( error, "Unable to set kernel arguments" );
}
error = clSetKernelArg( kernel, idx++, sizeof( cl_mem ), &xOffsets );
test_error( error, "Unable to set kernel arguments" );
error = clSetKernelArg( kernel, idx++, sizeof( cl_mem ), &yOffsets );
test_error( error, "Unable to set kernel arguments" );
error = clSetKernelArg( kernel, idx++, sizeof( cl_mem ), &results );
test_error( error, "Unable to set kernel arguments" );
// A cast of troublesome offsets. The first one has to be zero.
const float float_offsets[] = { 0.0f, MAKE_HEX_FLOAT(0x1.0p-30f, 0x1L, -30), 0.25f, 0.3f, 0.5f - FLT_EPSILON/4.0f, 0.5f, 0.9f, 1.0f - FLT_EPSILON/2 };
int float_offset_count = sizeof( float_offsets) / sizeof( float_offsets[0] );
int numTries = MAX_TRIES, numClamped = MAX_CLAMPED;
int loopCount = 2 * float_offset_count;
if( ! useFloatCoords )
loopCount = 1;
if (gTestMaxImages) {
loopCount = 1;
log_info("Testing each size only once with pixel offsets of %g for max sized images.\n", float_offsets[0]);
}
// Get the maximum absolute error for this format
double formatAbsoluteError = get_max_absolute_error(imageInfo->format, imageSampler);
if (gDebugTrace) log_info("\tformatAbsoluteError is %e\n", formatAbsoluteError);
if (0 == initHalf && imageInfo->format->image_channel_data_type == CL_HALF_FLOAT ) {
initHalf = CL_SUCCESS == DetectFloatToHalfRoundingMode( queue );
if (initHalf) {
log_info("Half rounding mode successfully detected.\n");
}
}
for( int q = 0; q < loopCount; q++ )
{
float offset = float_offsets[ q % float_offset_count ];
// Init the coordinates
InitFloatCoords( imageInfo, imageSampler, xOffsetValues, yOffsetValues,
q>=float_offset_count ? -offset: offset,
q>=float_offset_count ? offset: -offset, imageSampler->normalized_coords, d );
error = clEnqueueWriteBuffer( queue, xOffsets, CL_TRUE, 0, sizeof(cl_float) * imageInfo->height * imageInfo->width, xOffsetValues, 0, NULL, NULL );
test_error( error, "Unable to write x offsets" );
error = clEnqueueWriteBuffer( queue, yOffsets, CL_TRUE, 0, sizeof(cl_float) * imageInfo->height * imageInfo->width, yOffsetValues, 0, NULL, NULL );
test_error( error, "Unable to write y offsets" );
// Get results
size_t resultValuesSize = imageInfo->width * imageInfo->height * get_explicit_type_size( outputType ) * 4;
BufferOwningPtr<char> resultValues(malloc(resultValuesSize));
memset( resultValues, 0xff, resultValuesSize );
clEnqueueWriteBuffer( queue, results, CL_TRUE, 0, resultValuesSize, resultValues, 0, NULL, NULL );
// Run the kernel
threads[0] = (size_t)imageInfo->width;
threads[1] = (size_t)imageInfo->height;
error = clEnqueueNDRangeKernel( queue, kernel, 2, NULL, threads, NULL, 0, NULL, NULL );
test_error( error, "Unable to run kernel" );
if( gDebugTrace )
log_info( " reading results, %ld kbytes\n", (unsigned long)( imageInfo->width * imageInfo->height * get_explicit_type_size( outputType ) * 4 / 1024 ) );
error = clEnqueueReadBuffer( queue, results, CL_TRUE, 0, imageInfo->width * imageInfo->height * get_explicit_type_size( outputType ) * 4, resultValues, 0, NULL, NULL );
test_error( error, "Unable to read results from kernel" );
if( gDebugTrace )
log_info( " results read\n" );
// Validate results element by element
char *imagePtr = imageValues;
/*
* FLOAT output type
*/
if( outputType == kFloat )
{
// Validate float results
float *resultPtr = (float *)(char *)resultValues;
float expected[4], error=0.0f;
float maxErr = get_max_relative_error( imageInfo->format, imageSampler, 0 /*not 3D*/, CL_FILTER_LINEAR == imageSampler->filter_mode );
for( size_t y = 0, j = 0; y < imageInfo->height; y++ )
{
for( size_t x = 0; x < imageInfo->width; x++, j++ )
{
// Step 1: go through and see if the results verify for the pixel
// For the normalized case on a GPU we put in offsets to the X and Y to see if we land on the
// right pixel. This addresses the significant inaccuracy in GPU normalization in OpenCL 1.0.
int checkOnlyOnePixel = 0;
int found_pixel = 0;
float offset = NORM_OFFSET;
if (!imageSampler->normalized_coords || imageSampler->filter_mode != CL_FILTER_NEAREST || NORM_OFFSET == 0
#if defined( __APPLE__ )
// Apple requires its CPU implementation to do correctly rounded address arithmetic in all modes
|| gDeviceType != CL_DEVICE_TYPE_GPU
#endif
)
offset = 0.0f; // Loop only once
for (float norm_offset_x = -offset; norm_offset_x <= offset && !found_pixel; norm_offset_x += NORM_OFFSET) {
for (float norm_offset_y = -offset; norm_offset_y <= offset && !found_pixel; norm_offset_y += NORM_OFFSET) {
// Try sampling the pixel, without flushing denormals.
int containsDenormals = 0;
FloatPixel maxPixel = sample_image_pixel_float_offset( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.0f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, expected, 0, &containsDenormals );
float err1 = fabsf( resultPtr[0] - expected[0] );
float err2 = fabsf( resultPtr[1] - expected[1] );
float err3 = fabsf( resultPtr[2] - expected[2] );
float err4 = fabsf( resultPtr[3] - expected[3] );
// Clamp to the minimum absolute error for the format
if (err1 > 0 && err1 < formatAbsoluteError) { err1 = 0.0f; }
if (err2 > 0 && err2 < formatAbsoluteError) { err2 = 0.0f; }
if (err3 > 0 && err3 < formatAbsoluteError) { err3 = 0.0f; }
if (err4 > 0 && err4 < formatAbsoluteError) { err4 = 0.0f; }
float maxErr1 = MAX( maxErr * maxPixel.p[0], FLT_MIN );
float maxErr2 = MAX( maxErr * maxPixel.p[1], FLT_MIN );
float maxErr3 = MAX( maxErr * maxPixel.p[2], FLT_MIN );
float maxErr4 = MAX( maxErr * maxPixel.p[3], FLT_MIN );
// Check if the result matches.
if( ! (err1 <= maxErr1) || ! (err2 <= maxErr2) ||
! (err3 <= maxErr3) || ! (err4 <= maxErr4) )
{
//try flushing the denormals, if there is a failure.
if( containsDenormals )
{
// If implementation decide to flush subnormals to zero,
// max error needs to be adjusted
maxErr1 += 4 * FLT_MIN;
maxErr2 += 4 * FLT_MIN;
maxErr3 += 4 * FLT_MIN;
maxErr4 += 4 * FLT_MIN;
maxPixel = sample_image_pixel_float_offset( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, expected, 0, NULL );
err1 = fabsf( resultPtr[0] - expected[0] );
err2 = fabsf( resultPtr[1] - expected[1] );
err3 = fabsf( resultPtr[2] - expected[2] );
err4 = fabsf( resultPtr[3] - expected[3] );
}
}
// If the final result DOES match, then we've found a valid result and we're done with this pixel.
found_pixel = (err1 <= maxErr1) && (err2 <= maxErr2) && (err3 <= maxErr3) && (err4 <= maxErr4);
}//norm_offset_x
}//norm_offset_y
// Step 2: If we did not find a match, then print out debugging info.
if (!found_pixel) {
// For the normalized case on a GPU we put in offsets to the X and Y to see if we land on the
// right pixel. This addresses the significant inaccuracy in GPU normalization in OpenCL 1.0.
checkOnlyOnePixel = 0;
int shouldReturn = 0;
for (float norm_offset_x = -offset; norm_offset_x <= offset && !checkOnlyOnePixel; norm_offset_x += NORM_OFFSET) {
for (float norm_offset_y = -offset; norm_offset_y <= offset && !checkOnlyOnePixel; norm_offset_y += NORM_OFFSET) {
// If we are not on a GPU, or we are not normalized, then only test with offsets (0.0, 0.0)
// E.g., test one pixel.
if (!imageSampler->normalized_coords || gDeviceType != CL_DEVICE_TYPE_GPU || NORM_OFFSET == 0) {
norm_offset_x = 0.0f;
norm_offset_y = 0.0f;
checkOnlyOnePixel = 1;
}
int containsDenormals = 0;
FloatPixel maxPixel = sample_image_pixel_float_offset( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, expected, 0, &containsDenormals );
float err1 = fabsf( resultPtr[0] - expected[0] );
float err2 = fabsf( resultPtr[1] - expected[1] );
float err3 = fabsf( resultPtr[2] - expected[2] );
float err4 = fabsf( resultPtr[3] - expected[3] );
float maxErr1 = MAX( maxErr * maxPixel.p[0], FLT_MIN );
float maxErr2 = MAX( maxErr * maxPixel.p[1], FLT_MIN );
float maxErr3 = MAX( maxErr * maxPixel.p[2], FLT_MIN );
float maxErr4 = MAX( maxErr * maxPixel.p[3], FLT_MIN );
if( ! (err1 <= maxErr1) || ! (err2 <= maxErr2) ||
! (err3 <= maxErr3) || ! (err4 <= maxErr4) )
{
//try flushing the denormals, if there is a failure.
if( containsDenormals )
{
maxErr1 += 4 * FLT_MIN;
maxErr2 += 4 * FLT_MIN;
maxErr3 += 4 * FLT_MIN;
maxErr4 += 4 * FLT_MIN;
maxPixel = sample_image_pixel_float_offset( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, expected, 0, NULL );
err1 = fabsf( resultPtr[0] - expected[0] );
err2 = fabsf( resultPtr[1] - expected[1] );
err3 = fabsf( resultPtr[2] - expected[2] );
err4 = fabsf( resultPtr[3] - expected[3] );
}
}
if( ! (err1 <= maxErr1) || ! (err2 <= maxErr2) ||
! (err3 <= maxErr3) || ! (err4 <= maxErr4) )
{
log_error("FAILED norm_offsets: %g , %g:\n", norm_offset_x, norm_offset_y);
float tempOut[4];
shouldReturn |= determine_validation_error<float>( imagePtr, imageInfo, imageSampler, resultPtr,
expected, error, xOffsetValues[ j ], yOffsetValues[ j ], norm_offset_x, norm_offset_y, j, numTries, numClamped, true );
log_error( "Step by step:\n" );
FloatPixel temp = sample_image_pixel_float_offset( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, tempOut, 1 /* verbose */, &containsDenormals /*dont flush while error reporting*/ );
log_error( "\tulps: %2.2f, %2.2f, %2.2f, %2.2f (max allowed: %2.2f)\n\n",
Ulp_Error( resultPtr[0], expected[0] ),
Ulp_Error( resultPtr[1], expected[1] ),
Ulp_Error( resultPtr[2], expected[2] ),
Ulp_Error( resultPtr[3], expected[3] ),
Ulp_Error( MAKE_HEX_FLOAT(0x1.000002p0f, 0x1000002L, -24) + maxErr, MAKE_HEX_FLOAT(0x1.000002p0f, 0x1000002L, -24) ) );
} else {
log_error("Test error: we should have detected this passing above.\n");
}
}//norm_offset_x
}//norm_offset_y
if( shouldReturn )
return 1;
} // if (!found_pixel)
resultPtr += 4;
}
}
}
/*
* UINT output type
*/
else if( outputType == kUInt )
{
// Validate unsigned integer results
unsigned int *resultPtr = (unsigned int *)(char *)resultValues;
unsigned int expected[4];
float error;
for( size_t y = 0, j = 0; y < imageInfo->height; y++ )
{
for( size_t x = 0; x < imageInfo->width; x++, j++ )
{
// Step 1: go through and see if the results verify for the pixel
// For the normalized case on a GPU we put in offsets to the X and Y to see if we land on the
// right pixel. This addresses the significant inaccuracy in GPU normalization in OpenCL 1.0.
int checkOnlyOnePixel = 0;
int found_pixel = 0;
for (float norm_offset_x = -NORM_OFFSET; norm_offset_x <= NORM_OFFSET && !found_pixel && !checkOnlyOnePixel; norm_offset_x += NORM_OFFSET) {
for (float norm_offset_y = -NORM_OFFSET; norm_offset_y <= NORM_OFFSET && !found_pixel && !checkOnlyOnePixel; norm_offset_y += NORM_OFFSET) {
// If we are not on a GPU, or we are not normalized, then only test with offsets (0.0, 0.0)
// E.g., test one pixel.
if (!imageSampler->normalized_coords || gDeviceType != CL_DEVICE_TYPE_GPU || NORM_OFFSET == 0) {
norm_offset_x = 0.0f;
norm_offset_y = 0.0f;
checkOnlyOnePixel = 1;
}
sample_image_pixel_offset<unsigned int>( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, expected );
error = errMax( errMax( abs_diff_uint(expected[ 0 ], resultPtr[ 0 ]), abs_diff_uint(expected[ 1 ], resultPtr[ 1 ]) ),
errMax( abs_diff_uint(expected[ 2 ], resultPtr[ 2 ]), abs_diff_uint(expected[ 3 ], resultPtr[ 3 ]) ) );
if (error <= MAX_ERR)
found_pixel = 1;
}//norm_offset_x
}//norm_offset_y
// Step 2: If we did not find a match, then print out debugging info.
if (!found_pixel) {
// For the normalized case on a GPU we put in offsets to the X and Y to see if we land on the
// right pixel. This addresses the significant inaccuracy in GPU normalization in OpenCL 1.0.
checkOnlyOnePixel = 0;
int shouldReturn = 0;
for (float norm_offset_x = -NORM_OFFSET; norm_offset_x <= NORM_OFFSET && !checkOnlyOnePixel; norm_offset_x += NORM_OFFSET) {
for (float norm_offset_y = -NORM_OFFSET; norm_offset_y <= NORM_OFFSET && !checkOnlyOnePixel; norm_offset_y += NORM_OFFSET) {
// If we are not on a GPU, or we are not normalized, then only test with offsets (0.0, 0.0)
// E.g., test one pixel.
if (!imageSampler->normalized_coords || gDeviceType != CL_DEVICE_TYPE_GPU || NORM_OFFSET == 0) {
norm_offset_x = 0.0f;
norm_offset_y = 0.0f;
checkOnlyOnePixel = 1;
}
sample_image_pixel_offset<unsigned int>( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, expected );
error = errMax( errMax( abs_diff_uint(expected[ 0 ], resultPtr[ 0 ]), abs_diff_uint(expected[ 1 ], resultPtr[ 1 ]) ),
errMax( abs_diff_uint(expected[ 2 ], resultPtr[ 2 ]), abs_diff_uint(expected[ 3 ], resultPtr[ 3 ]) ) );
if( error > MAX_ERR )
{
log_error("FAILED norm_offsets: %g , %g:\n", norm_offset_x, norm_offset_y);
shouldReturn |= determine_validation_error<unsigned int>( imagePtr, imageInfo, imageSampler, resultPtr,
expected, error, xOffsetValues[j], yOffsetValues[j], norm_offset_x, norm_offset_y, j, numTries, numClamped, false );
} else {
log_error("Test error: we should have detected this passing above.\n");
}
}//norm_offset_x
}//norm_offset_y
if( shouldReturn )
return 1;
} // if (!found_pixel)
resultPtr += 4;
}
}
}
/*
* INT output type
*/
else
{
// Validate integer results
int *resultPtr = (int *)(char *)resultValues;
int expected[4];
float error;
for( size_t y = 0, j = 0; y < imageInfo->height; y++ )
{
for( size_t x = 0; x < imageInfo->width; x++, j++ )
{
// Step 1: go through and see if the results verify for the pixel
// For the normalized case on a GPU we put in offsets to the X and Y to see if we land on the
// right pixel. This addresses the significant inaccuracy in GPU normalization in OpenCL 1.0.
int checkOnlyOnePixel = 0;
int found_pixel = 0;
for (float norm_offset_x = -NORM_OFFSET; norm_offset_x <= NORM_OFFSET && !found_pixel && !checkOnlyOnePixel; norm_offset_x += NORM_OFFSET) {
for (float norm_offset_y = -NORM_OFFSET; norm_offset_y <= NORM_OFFSET && !found_pixel && !checkOnlyOnePixel; norm_offset_y += NORM_OFFSET) {
// If we are not on a GPU, or we are not normalized, then only test with offsets (0.0, 0.0)
// E.g., test one pixel.
if (!imageSampler->normalized_coords || gDeviceType != CL_DEVICE_TYPE_GPU || NORM_OFFSET == 0) {
norm_offset_x = 0.0f;
norm_offset_y = 0.0f;
checkOnlyOnePixel = 1;
}
sample_image_pixel_offset<int>( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, expected );
error = errMax( errMax( abs_diff_int(expected[ 0 ], resultPtr[ 0 ]), abs_diff_int(expected[ 1 ], resultPtr[ 1 ]) ),
errMax( abs_diff_int(expected[ 2 ], resultPtr[ 2 ]), abs_diff_int(expected[ 3 ], resultPtr[ 3 ]) ) );
if (error <= MAX_ERR)
found_pixel = 1;
}//norm_offset_x
}//norm_offset_y
// Step 2: If we did not find a match, then print out debugging info.
if (!found_pixel) {
// For the normalized case on a GPU we put in offsets to the X and Y to see if we land on the
// right pixel. This addresses the significant inaccuracy in GPU normalization in OpenCL 1.0.
checkOnlyOnePixel = 0;
int shouldReturn = 0;
for (float norm_offset_x = -NORM_OFFSET; norm_offset_x <= NORM_OFFSET && !checkOnlyOnePixel; norm_offset_x += NORM_OFFSET) {
for (float norm_offset_y = -NORM_OFFSET; norm_offset_y <= NORM_OFFSET && !checkOnlyOnePixel; norm_offset_y += NORM_OFFSET) {
// If we are not on a GPU, or we are not normalized, then only test with offsets (0.0, 0.0)
// E.g., test one pixel.
if (!imageSampler->normalized_coords || gDeviceType != CL_DEVICE_TYPE_GPU || NORM_OFFSET == 0) {
norm_offset_x = 0.0f;
norm_offset_y = 0.0f;
checkOnlyOnePixel = 1;
}
sample_image_pixel_offset<int>( imageValues, imageInfo,
xOffsetValues[ j ], yOffsetValues[ j ], 0.f, norm_offset_x, norm_offset_y, 0.0f,
imageSampler, expected );
error = errMax( errMax( abs_diff_int(expected[ 0 ], resultPtr[ 0 ]), abs_diff_int(expected[ 1 ], resultPtr[ 1 ]) ),
errMax( abs_diff_int(expected[ 2 ], resultPtr[ 2 ]), abs_diff_int(expected[ 3 ], resultPtr[ 3 ]) ) );
if( error > MAX_ERR )
{
log_error("FAILED norm_offsets: %g , %g:\n", norm_offset_x, norm_offset_y);
shouldReturn |= determine_validation_error<int>( imagePtr, imageInfo, imageSampler, resultPtr,
expected, error, xOffsetValues[j], yOffsetValues[j], norm_offset_x, norm_offset_y, j, numTries, numClamped, false );
} else {
log_error("Test error: we should have detected this passing above.\n");
}
}//norm_offset_x
}//norm_offset_y
if( shouldReturn )
return 1;
} // if (!found_pixel)
resultPtr += 4;
}
}
}
}
return numTries != MAX_TRIES || numClamped != MAX_CLAMPED;
}
int test_read_image_set_2D( cl_device_id device, cl_image_format *format, image_sampler_data *imageSampler,
bool floatCoords, ExplicitType outputType )
{
char programSrc[10240];
const char *ptr;
const char *readFormat;
clProgramWrapper program;
clKernelWrapper kernel;
RandomSeed seed( gRandomSeed );
int error;
// Get our operating params
size_t maxWidth, maxHeight;
cl_ulong maxAllocSize, memSize;
image_descriptor imageInfo = { 0x0 };
size_t pixelSize;
imageInfo.format = format;
imageInfo.depth = imageInfo.arraySize = imageInfo.slicePitch = 0;
imageInfo.type = CL_MEM_OBJECT_IMAGE2D;
pixelSize = get_pixel_size( imageInfo.format );
error = clGetDeviceInfo( device, CL_DEVICE_IMAGE2D_MAX_WIDTH, sizeof( maxWidth ), &maxWidth, NULL );
error |= clGetDeviceInfo( device, CL_DEVICE_IMAGE2D_MAX_HEIGHT, sizeof( maxHeight ), &maxHeight, NULL );
error |= clGetDeviceInfo( device, CL_DEVICE_MAX_MEM_ALLOC_SIZE, sizeof( maxAllocSize ), &maxAllocSize, NULL );
error |= clGetDeviceInfo( device, CL_DEVICE_GLOBAL_MEM_SIZE, sizeof( memSize ), &memSize, NULL );
test_error( error, "Unable to get max image 2D size from device" );
// Determine types
if( outputType == kInt )
readFormat = "i";
else if( outputType == kUInt )
readFormat = "ui";
else // kFloat
readFormat = "f";
// Construct the source
const char *samplerArg = samplerKernelArg;
char samplerVar[ 1024 ] = "";
if( gUseKernelSamplers )
{
get_sampler_kernel_code( imageSampler, samplerVar );
samplerArg = "";
}
sprintf( programSrc, read2DKernelSourcePattern, samplerArg, get_explicit_type_name( outputType ),
samplerVar,
floatCoords ? floatKernelSource : intCoordKernelSource,
readFormat );
ptr = programSrc;
error = create_single_kernel_helper( context, &program, &kernel, 1, &ptr, "sample_kernel" );
test_error( error, "Unable to create testing kernel" );
if( gTestSmallImages )
{
for( imageInfo.width = 1; imageInfo.width < 13; imageInfo.width++ )
{
imageInfo.rowPitch = imageInfo.width * pixelSize;
for( imageInfo.height = 1; imageInfo.height < 9; imageInfo.height++ )
{
if( gDebugTrace )
log_info( " at size %d,%d\n", (int)imageInfo.width, (int)imageInfo.height );
int retCode = test_read_image_2D( device, context, queue, kernel, &imageInfo, imageSampler, floatCoords, outputType, seed );
if( retCode )
return retCode;
}
}
}
else if( gTestMaxImages )
{
// Try a specific set of maximum sizes
size_t numbeOfSizes;
size_t sizes[100][3];
get_max_sizes(&numbeOfSizes, 100, sizes, maxWidth, maxHeight, 1, 1, maxAllocSize, memSize, CL_MEM_OBJECT_IMAGE2D, imageInfo.format);
for( size_t idx = 0; idx < numbeOfSizes; idx++ )
{
imageInfo.width = sizes[ idx ][ 0 ];
imageInfo.height = sizes[ idx ][ 1 ];
imageInfo.rowPitch = imageInfo.width * pixelSize;
log_info("Testing %d x %d\n", (int)sizes[ idx ][ 0 ], (int)sizes[ idx ][ 1 ]);
if( gDebugTrace )
log_info( " at max size %d,%d\n", (int)sizes[ idx ][ 0 ], (int)sizes[ idx ][ 1 ] );
int retCode = test_read_image_2D( device, context, queue, kernel, &imageInfo, imageSampler, floatCoords, outputType, seed );
if( retCode )
return retCode;
}
}
else if( gTestRounding )
{
uint64_t typeRange = 1LL << ( get_format_type_size( imageInfo.format ) * 8 );
typeRange /= pixelSize / get_format_type_size( imageInfo.format );
imageInfo.height = (size_t)( ( typeRange + 255LL ) / 256LL );
imageInfo.width = (size_t)( typeRange / (cl_ulong)imageInfo.height );
while( imageInfo.height >= maxHeight / 2 )
{
imageInfo.width <<= 1;
imageInfo.height >>= 1;
}
while( imageInfo.width >= maxWidth / 2 )
imageInfo.width >>= 1;
imageInfo.rowPitch = imageInfo.width * pixelSize;
gRoundingStartValue = 0;
do
{
if( gDebugTrace )
log_info( " at size %d,%d, starting round ramp at %llu for range %llu\n", (int)imageInfo.width, (int)imageInfo.height, gRoundingStartValue, typeRange );
int retCode = test_read_image_2D( device, context, queue, kernel, &imageInfo, imageSampler, floatCoords, outputType, seed );
if( retCode )
return retCode;
gRoundingStartValue += imageInfo.width * imageInfo.height * pixelSize / get_format_type_size( imageInfo.format );
} while( gRoundingStartValue < typeRange );
}
else
{
for( int i = 0; i < NUM_IMAGE_ITERATIONS; i++ )
{
cl_ulong size;
// Loop until we get a size that a) will fit in the max alloc size and b) that an allocation of that
// image, the result array, plus offset arrays, will fit in the global ram space
do
{
imageInfo.width = (size_t)random_log_in_range( 16, (int)maxWidth / 32, seed );
imageInfo.height = (size_t)random_log_in_range( 16, (int)maxHeight / 32, seed );
imageInfo.rowPitch = imageInfo.width * pixelSize;
if( gEnablePitch )
{
size_t extraWidth = (int)random_log_in_range( 0, 64, seed );
imageInfo.rowPitch += extraWidth * pixelSize;
}
size = (size_t)imageInfo.rowPitch * (size_t)imageInfo.height * 4;
} while( size > maxAllocSize || ( size * 3 ) > memSize );
if( gDebugTrace )
log_info( " at size %d,%d (row pitch %d) out of %d,%d\n", (int)imageInfo.width, (int)imageInfo.height, (int)imageInfo.rowPitch, (int)maxWidth, (int)maxHeight );
int retCode = test_read_image_2D( device, context, queue, kernel, &imageInfo, imageSampler, floatCoords, outputType, seed );
if( retCode )
return retCode;
}
}
return 0;
}
View Git Blame Copy Permalink