Pitched Memory Pool

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi,

I would like to implement a memory pool for pitched memory as it is implemented by pyCuda for normally allocated memory. What do you think about it?

Is there a possibility to find out the pitch and alignment rules for a special width and height of bytes?

Regards,

Kwyjibo

2

Hi,

I would like to implement a memory pool for pitched memory as it is implemented by pyCuda for normally allocated memory. What do you think about it?

Is there a possibility to find out the pitch and alignment rules for a special width and height of bytes?

Regards,

Kwyjibo

3

So I guess nobody knows how to find out these parameters without reserving memory and there is no assistance by the runtime environment.

Do you think preallocating memory makes sense at all? I have about 150+ (pitched) memory allocations per second and want to circumvent this.

Regards,

Kwyjibo

4

So I guess nobody knows how to find out these parameters without reserving memory and there is no assistance by the runtime environment.

Do you think preallocating memory makes sense at all? I have about 150+ (pitched) memory allocations per second and want to circumvent this.

Regards,

Kwyjibo