ATI Stream SDK V2.0 | OpenCL Full CPU & GPU Support | Introduction

[64NOMIS]

After looking at the documentation it seems that I was correct. You shouldn't have to recompile for other architectures because OpenCL enumerates and utilizes resources (GPU/CPU/Cell/Etc) at runtime. A compiled OpenCL Kernel should be able to run on either company's hardware. There is a way to include what they call Native Kernels in your program. They are hardware specific - but they also aren't standard and being able to execute native kernels apparently isn't even a requirement in the standard.

Brace yourselves. The Kernel is not compiled until just prior to execution (hardware-specific native Kernels seem unlikely in practice since the whole idea is to be hardware independent but the HPC guys might do this) when it is directed to be built/compiled by the application. The compilation step is hardware dependent; if the application is executing on a platform with an x86 processor and an ATI GPU the Kernel will have either path available for compilation, potentially even determined at runtime! So if the GPU is busy, your applications could complile the Kernel for the CPU, execute it, and at a later time, compile it for the GPU, and execute it! You could hold an iner-application race, compiling separate instances for both, one for x86 and one for the GPU, and running them concurrently! I describe some of this in the interview: http://links.amd.com/openinterview

[fellix_bg]

No HD3000 series support?

Why is that? they support Compute Shaders, just like the HD4000 series cards...

R600 architecture does not support LDS memory.

[Marios]

I can't see why a developer might use opencl to accelerate a program for the CPU and not program directly in a multi core multi thread way.
This way there is no need to use opencl runtime environment to generate binaries on the fly. I suppose this is a time consuming task.
There is no word from Intel about opencl support for their cpus. Why?
Is it because Larrabee as a "cpu", can be programed in a straightforward way to support GPGPU (physx etc) and doesn't really need opencl?
Intel owns Havok, Nvidia owns cuda Physx and promotes it with game developers, AMD is coming out with opencl.
Nice standards we have for our PCs.

[To(V)bo Co(V)bo]

Brace yourselves. The Kernel is not compiled until just prior to execution (hardware-specific native Kernels seem unlikely in practice since the whole idea is to be hardware independent but the HPC guys might do this) when it is directed to be built/compiled by the application. The compilation step is hardware dependent; if the application is executing on a platform with an x86 processor and an ATI GPU the Kernel will have either path available for compilation, potentially even determined at runtime! So if the GPU is busy, your applications could complile the Kernel for the CPU, execute it, and at a later time, compile it for the GPU, and execute it! You could hold an iner-application race, compiling separate instances for both, one for x86 and one for the GPU, and running them concurrently! I describe some of this in the interview: http://links.amd.com/openinterview

Think if they wrote AI "artificial intelligence" in parallel. Where you could have a program answer its own questions. If you think about life in general it is all reasoning. If you were able to compute enough possible outcomes to a given problem (or moment) and choose the best answer. You would have a form of intelligence. The main function of life is self-preservetion, apply this together and you got yourself a terminator. I dont think its to far off if your using the compute power these days. If you think about it, it isn't to far fetched. You got to think, modern inventions were all just imagination at one time or another.

[Hell Hound]

Think if they wrote AI "artificial intelligence" in parallel. Where you could have a program answer its own questions.You would have a form of intelligence. The main function of life is self-preservetion, apply this together and you got yourself a terminator.

[To(V)bo Co(V)bo]

I bet you skynet can run on opencl, using parallel instructions. A terminator could run on as little as 30 tflops. It could crunch the physics of gun trajectory, distances, and time. Finally a real use for physics.

[EniGmA1987]

Well hopefully if/when "skynet" takes over, it will use x86 architecture. Because then either it will make x86 into the best computer architecture ever, or it will have so much junk we have a good change of defeating it

[To(V)bo Co(V)bo]

No way possible, a terminator would definately need way more than 4gb of system memory to effectively kill people. I bet they also need Gddr5 bandwith to render complex infra-red video, and not forget last, ray tracing.

[Xoulz]

No way possible, a terminator would definately need way more than 4gb of memory to effectively kill people. I bet they also will need Gddr5 bandwith to render complex infra-red video, and not forget last, ray tracing.

Dude, it's widely known that the early terminator models used RAMBUS.




()

[To(V)bo Co(V)bo]

Your right forgot about those ones with rambus, they didnt use gddr5 till after the patent infringement lawsuit.

Skynet became self-aware after a new malicious folding at home client was released for the gt300 architecture. The new client was able to give skynet access to thousands of computers, and allowed for true real time physics processing without taking a fps hit.

[Lightman]

Your right forgot about those ones with rambus, they didnt use gddr5 till after the patent infringement lawsuit.

Skynet became self-aware after a new malicious folding at home client was released for the gt300 architecture. The new client was able to give skynet access to thousands of computers, and allowed for true real time physics processing without taking a fps hit.

So now we know why 12/2012 will be end of the world ....

[Angry]

So now we know why 12/2012 will be end of the world ....

Oh come on, we all know the world will end when Fermi is released...

[zalbard]

AMD's ATI Stream SDK 2.0 no longer beta!

AMD has decided to give developers an early Christmas gift by releasing the final version of its ATI Stream Software Development Kit 2.0 which brings OpenCL 1.0 support as well as the hardware acceleration for applications.

The final version of the ATI Stream SDK 2.0 features:

* First production release of ATI Stream SDK with OpenCL™ 1.0 support.
* New: Support for OpenCL™ ICD (Installable Client Driver).
* New: Support for atomic functions for 32-bit integers.
* New: Microsoft® Visual Studio® 2008-integrated ATI Stream Profiler performance analysis tool.
* Preview: Support for OpenCL™ / OpenGL® interoperability.
* Preview: Support for OpenCL™ / Microsoft® DirectX® 10 interoperability.
* Preview: Support for double-precision floating point basic arithmetic in OpenCL™ C kernels.
* Updated OpenCL™ runtime to conditionally load ATI CAL runtime libraries to allow execution on compatible CPUs without ATI Catalyst™ installed.
* Updated OpenCL™ runtime to allow simultaneous use of OpenCL™ and ATI CAL APIs in a single user application.
* Updated cl.hpp from the Khronos OpenCL working group release.
* Various OpenCL™ compiler and runtime fixes and enhancements

It supports Windows XP, Vista, 7 as well as openSuse 11.0 and Ubuntu 9.04 operating systems and supports all the graphics cards ranging from Radeon HD 4000, HD 5000, Mobility HD 4000 as well as ATI FireStream 9270 and 9250 and FirePro cards.

You can find it here.

Hm, time to see if there's a performance improvement for benches.

[zanzabar]

AMD's ATI Stream SDK 2.0 no longer beta!

Hm, time to see if there's a performance improvement for benches.

it benches the dc benchmark 10-15% faster

[Boogerlad]

still doesn't work with dolphin. *sigh*

[Calmatory]

... NOW I wish I had Radeon GPU on my laptop.

Is this compatible with mobile HD3470?

[Nedjo]

... NOW I wish I had Radeon GPU on my laptop.

Is this compatible with mobile HD3470?

supports all the graphics cards ranging from Radeon HD 4000, HD 5000, Mobility HD 4000 as well as ATI FireStream 9270 and 9250 and FirePro cards

[Calmatory]

Ah, damn. Thanks.

Guess I'll build a dedicated box for OpenCL stuff then. Easier and cheaper than replacing the current laptop + OS.