problem using texture memory getting errors while running debug mode

Accelerated Computing CUDA CUDA Programming and Performance
1

cudaArray *a_Data;

cudaChannelFormatDesc floatTex = cudaCreateChannelDesc();

CUDA_SAFE_CALL( cudaMallocArray(&a_Data, &floatTex, DATA_W, DATA_H) ); ==> line number 150

CUDA_SAFE_CALL( cudaMemcpyToArray(a_Data, 0, 0, h_DataA, DATA_SIZE, cudaMemcpyHostToDevice) );

CUDA_SAFE_CALL( cudaBindTextureToArray(texData, a_Data) );

when I am trying to compile code which uses texture memory, I am getting error "
Cuda error in file ‘convolutionTexture.cu’ in line 150 : feature is not yet implemented." in debug mode.

Please let me know if anybody has faced same issue and what is solution to it ?

thanks in advance

2

whats ur DATA_W and DATA_H ?

3

“feature is not yet implemented” is usually a result of an improper driver/toolkit pair. i.e., the CUDA 2.0 toolkit with only CUDA 1.1 supported in the driver.

4

CUDA 2.0
CUDA driver: NVIDIA Driver for Microsoft Windows XP with CUDA Support (178.08). Includes driver support for 8 series and higher GPUs.
CUDA toolkit: CUDA Toolkit version 2.0 for Windows XP
CUDA SDK: CUDA SDK version 2.0 for Windows XP

This is the configuration i am using .

@sarnath DATA_W and DATA_H represents width and height … i am taking 320 & 240