AnsweredAssumed Answered

Memory Allocation and Estimation

Question asked by Rohan Valvekar on Jan 30, 2019
Latest reply on Feb 12, 2019 by Rohan Valvekar

Hello fellow NXP Users and Developers,

I have been researching into using i.MX Processors for Video Processing application and had few doubts about Memory Planning and Estimation.

Since it is hard to determine the exact pin point memory requirement of any application which involves a bit of researching for custom hardware and software, it is Hard to plan out the Memory Estimation using the development boards.

With respect to Graphical Processing especially with Image Processing and Camera Systems, Memory is the most important part of the system. 

I have been searching the forum for a Memory Planning and Estimation discussion, but have not go a satisfactory answer. 

 

Well I would like to know and get to know better, and understand better with a academic perspective, that how can we do a memory estimation with imx processors which Must have the functionality of GPU and VPU.

 

Most of the high end development boards have 4 GB RAM, on which most of the things do work efficiently. But what i want to know is what is the minimum threshold for memory when it comes to GPU and VPU based applications. 

 

  1. Is there any golden mean for memory estimations with GPU and VPU applications ?
  2. Does the memory estimation involves sharing of memory ? So the memory is shared between a GPU and the CPU ?
  3. Can we have a dedicated memory just for GPU and VPU ?
  4. Are there any other ways, like non conventional ways to have a memory estimation ?
  5. How to make sure that the memory planning is also thought about with respect to power consumption
  6. Is it possible to load the memory and activate it, but not use it ? So for example if I have 2 GB of memory, then basically 2 GB will be initialized, but 1 GB will be used. When there is need of extra memory, the buffer memory can be used, something like SMART Memory.

Any other thoughts would be most welcome.

Thank You and appreciate the much help and support needed with memory estimation.

Regards,

Rohan

Outcomes