LabVIEW GPU Computing

[jl8987]

MathGuy - Any updates on the GPU front?  What are the plans for this package?

[pierpamar]

Sorry, I realized that I probably I posted the following comment in the wrong spot... I post it here again:

MathGuy -

I recently started working on the integration of my LabVIEW code with CUDA. I have a quick question for you: I have a Tesla C1060, which comes with 4GB global memory, and for now I am using WinXP 32-bit, LabVIEW 2009 32-bit, NVIDIA 32-bit drivers+toolkit (v2.3) for the GPU and the LabVIEW GPU Computing 32-bit package which you posted. However, using a 64-bit platforms would dramatically improve my performance in terms of memory usage and increase performances. Is there any plan in the near future to update the package also for 64-bit applications? I saw in the forum many people pointing out the same issue and all of them had to downgrade to 32-bit in order to get the GPU working. Thanks.

[gaisi]

Hi, MathGuy:

I am learning to use this LVCUDA thing for my image processing right now, and I just have some questions.

Though you have provide VIs like "Copy 1D Array to CUDA Memory(SGL)", it seems that the only way to perform functions( like FFT )is still calling the "Call Library Function Node". The only difference is that we have to create a compute context beforehand and link the CLN to the DLL we have made.

Am I right?

Besides, as the CUDA toolkit has already provide us a CUFFT.dll, could I just link my CLN directly to that file?

Lot Thanks.

[gaisi]

Hi, MathGuy:

I have met some problem which is really frustrating.

I tried to copy your example and realize a simple addition between two number and call it from labview. However, after I run the program in the labview, the labview shut itself autometically.

Hope you could help me.

It seems that I cannot attach my programs here.

[MathGuy]

You may want to post your problem to the GPU Computing group's discussion board. I'm pretty sure you can attach files there.

The first items to verify are the versions of CUDA and Windows you are using. The module supports CUDA 2.x (although 2.3 has not been thoroughly checked) and 32-bit Windows versions (XP & & have been tested fairly well - Vista has not been).

[gaisi]

Hi, Have you meet such problem that the FFT results got from Labview and GPU are different?

My input is 1,2,3,2,1,2,3,2,1,2, and the results from GPU are 19,1.23607,3.23607,3.23607,1.23607,1,3.23607,3.23607,1.23607;

And the ones from GPU computing is 19,2.80079,3.74502,3.74502,2.80079,3.29691,2.80079,3.74502,3.74502,2.80079;

[MathGuy]

Glad to hear you've had success getting the module working. I moved this comment to a discussion thread (http://decibel.ni.com/content/thread/8767) under GPU Computing so that it will be more visible.

[Carrera]

Hi, MathGuy!

I am very excited to find the LabVIEW GPU Computing library.

I have installed the following versions of cuda driver, toolkit, sdk in this order to my 32bit Windows Vista machine equippied with a NVIDIA GeForce 9500GT.

     devdriver_3.2_winvista-win7_32_263.06_general

     cudatoolkit_3.2.16_win_32

     gpucomputingsdk_3.1_win_32

Both deviceQuery.exe and bandwidthTest.exe in the SDK worked perfect, but "European Call Option.vi" didn't.

I found that "Get LVCUDA DLL Path.vi" is returning the right path for the dll (C:\LVCUDA\bin\lvcuda.dll), but "Get CUDA Context.vi" is returning the error code 13.

My question is:

Do the above mentioned CUDA drivers work fine with "LabVIEW GPU Computing - LV86_Win32"?

If yes, what are possible reasons for the error?

If no, which version should I install?

Thank you in advance for your support.

[MathGuy]

You've installed a version of CUDA that is not supported by the current LVCUDA library. We built the module using CUDA v2.1 and have tested it against v2.2. Others have mentioned that it also works w/ v2.3 but I have not personally validated that. Unfortunately, CUDA runtime libraries have yet to be compatible across major revisions (e.g. v2.x to v3.x).

[Carrera]

Thank you for your reply.

I look forward to a new version of LVCUDA library.