Do Intel, AMD, and Nvidia really support OpenCL?

  • Thread starter Thread starter Fusiontron
  • Start date Start date
  • Tags Tags
    Nvidia Support
Click For Summary

Discussion Overview

The discussion revolves around the legitimacy and applicability of OpenCL for GPU programming, particularly in the context of simulating heat transfer across different architectures. Participants explore the compatibility of OpenCL with various graphics card vendors and share insights on programming challenges and resources for learning OpenCL.

Discussion Character

  • Exploratory
  • Technical explanation
  • Debate/contested
  • Mathematical reasoning

Main Points Raised

  • One participant questions the legitimacy of OpenCL for their project, seeking insights on its effectiveness across different GPU architectures.
  • Another participant explains the relationship between OpenGL and vendor-specific implementations, suggesting that similar principles may apply to OpenCL, though they are unsure about OpenCL specifically.
  • A different participant supports OpenCL's legitimacy, noting its dependence on the graphics card and its capability to work with heterogeneous systems of CPUs and GPUs from different vendors.
  • Concerns are raised about the difficulty of programming in OpenCL, with a suggestion that prior exposure to parallel computing may be necessary for effective use.
  • Resources for learning OpenCL are shared, including books and specifications that may aid in understanding parallel processing concepts.

Areas of Agreement / Disagreement

Participants express varying levels of confidence in OpenCL's suitability for the project, with some supporting its use while others remain uncertain about its programming challenges and effectiveness. No consensus is reached regarding its overall legitimacy or ease of use.

Contextual Notes

Participants highlight the dependence of OpenCL's effectiveness on specific hardware and the potential need for vendor-specific extensions. There is also mention of the learning curve associated with programming in OpenCL.

Who May Find This Useful

Individuals interested in GPU programming, parallel computing, or those exploring cross-vendor compatibility in computational tasks may find this discussion relevant.

Fusiontron
Messages
108
Reaction score
2
I'm working on a project using GPUs to simulate heat transfer. We are trying to create a superset which will execute on all the main architectures. How legit is OpenCL for this purpose?
 
Technology news on Phys.org
Fusiontron said:
I'm working on a project using GPUs to simulate heat transfer. We are trying to create a superset which will execute on all the main architectures. How legit is OpenCL for this purpose?

Hey Fusiontron.

The way OpenGL works is that the graphics card vendor (i.e. ATI/nVidia/Intel etc) provide a code library (DLL file) that implements the actual routines that are created by the OpenGL standards board, of which many of the members are from graphics cards companies.

What you have to do when you code the routines is query the DLL for the OpenGL version and any extensions that you wish to use.

If the OpenGL is set to a specific version or greater than a specific version, you are gauranteed to have all the standardized features of that particular version.

Basically you have to do two things:

1) Query the supported version. If the version supports everything you need and the system has a version greater than or equal to your version, then you are done and you can guarantee that the code that relies on having that version will run ok.

2) If you do not have a high enough implementation then you have to query for extensions and use them.

Extensions are basically non-standard features that vendors implement. These features can be experimental or they can be highly optimized routines that make use of new features in the architecture of the graphics device itself.

The features are vendor specific in most cases and in some cases you can use these extensions to do the same things that you can do in higher OpenGL version standards when the device has a lower standard. This is actually the kind of thing that was done in the video game Doom 3 to support the older video cards: custom extensions had to be used.

In terms of computing with regard to graphics, there are OpenGL standards that gaurantee specific computational platforms which will be useful for you, since once the standard for your graphics card is guaranteed, you just run the same code for every card that adheres to the standard.

I'm not sure about OpenCL specifically, but for OpenGL it is not hard to program GPU's and get results back if you use a high enough implementation.
 
I can vouch for OpenCL. It's dependent on your graphics card. Some of the newer ones support OpenCL, from both ATI and Nvidia. I can also vouch for the fact that it is pretty difficult to program. Whether or not it's suited to your purpose <shrugs>, depending on the computational algorithm you use it may be. If your problem domain is already highly parallelizable, then GPU computing with OpenCL will work great for you. (You may need to designate a month or three to learning it if you've never had exposure to parallel computing before).

OpenCL is about as legitimate as DirectX or CUDA, but unlike DirectX or CUDA, OpenCL was designed to be capable of working with a heterogeneous system of CPU's and GPUs. This means that it's capable of working with CPU's and GPUs from different vendors equally well.

Which GPUs are you working with?
 
Thanks for the responses. It looks like OpenCL will probably work for our project. I'm going to visit the head of the project tomorrow to report to him these findings. Now, if anyone could hit me up with some good resources for learning it, then that'd be real helpful.
 
GPU Computing Gems Emerald Edition:

https://www.amazon.com/dp/0123849888/?tag=pfamazon01-20

--- Should be a fantastic resource.

Programming Massively Parallel Processors:

https://www.amazon.com/dp/0123814723/?tag=pfamazon01-20

--- This introduces you to the concepts of parallel processing and terminology for the GPU. It spends a lot of time (the first several chapters) explaining concepts in CUDA, but the way of thinking is similar to OpenCL.


OpenCL Khronos specification:

http://www.khronos.org/registry/cl/

--- This thing is your bible. Don't leave home without it.
 

Similar threads

  • · Replies 7 ·
Replies
7
Views
3K
Replies
5
Views
3K
  • · Replies 12 ·
Replies
12
Views
2K
Replies
4
Views
2K
Replies
14
Views
2K
  • · Replies 4 ·
Replies
4
Views
8K
Replies
10
Views
5K
Replies
4
Views
1K
Replies
5
Views
3K