I do not see a HD 4xxx card supported at http://developer.amd.com/tools/heterogeneous-computing/amd-accelerated-parallel-processing-app-sdk/system-requirements-driver-compatibility/ .
But IIRC, these cards had emulated opencl features, probably opencl 1.0 long time back. So you may be able to run some opencl programs, by installing some old SDK. But opencl emu, is another open source project, and I can't comment if it would run on old SDK.
Probably you can try using your CPU for running opencl.
Low-end (and mobility) HD4xxx are not OpenCL-capable at all cause they are just renamed HD3xxx architecture.
Medium range and above HD4xxx GPU are OpenCL 1.0 devices w/o true local memory (that's what emulated there).
I am a beginner to the OpenCL programming, i was using OpenCL Emu to compensate for the absence of GPU device on my laptop....
What i want is to run some simple openCL codes on my laptop with GPU, but the problem with emulator is that it does not give the execution time as taken by GPU execution!! guys, if you can provide me link or something which help me solve this problem i will be greatful
http://www.amd.com/us/products/notebook/graphics/ati-mobility-hd-4000/hd-4300/Pages/hd-4300-specs.aspx - This is the spec page of ATI Mobility Radeon 4300 series.
http://www.amd.com/uk/products/notebook/graphics/ati-mobility-hd-5400/Pages/hd-5470-specs.aspx - Here is the spec page of ATI Mobility Radeon 5470
You can see that OpenCL Support is listed in 5470 and not in the 4300 series spec.
I guess there is no OpenCL support for your GPU series.
is der any way that is can simulate the kernels in d absence of a valid GPU???
along with the execution time if possible
Yes, it's possible to use CPU as OpenCL device in case you got SDK installed on that PC.
You could have difficulties with installation though. Don't know how SDK install behave in absence of valid GPU. For example, you can't install Intel OpenCL SDK on non-Intel PC at all. Though if you will copy just needed files you can build against intel SDK on AMD APU PC.
Regarding execution times I almost sure it's impossible. You only can get some runtime estimates via such tool as SKA or more recent CodeXL....
Unfortunately, you dont have a straight option here. If you laptop GPU does not support OpenCL, then there is no way to get the numbers.
I wish there was a portal where you could submit your OpenCL application and get resutls of execution on various GPU hardware.
However, there is no such place, as far as I know.
If you could post your application sources (in this thread) and instructions on how to run it -- we will try running on a few GPUs that we have here and let you know what numbers your code is hitting.
If sources are confidential, you can post your binary. This may or may not work - depending on what version of the compiler you used to compile etc..
Now, I leave this decision to you.