CULA and CUBLAS

Support for issues specific to the Windows operating systems.

CULA and CUBLAS

Postby tang » Sat Nov 28, 2009 2:56 am

Dear Administrator,

I am trying to learn the new powerful numerical calculation libraries accelerated by GPU. However, there are two different libraries: CULA and CUBLAS, the former is the collection of LAPACK functions and the latter is the collection of BLAS level functions. However, it seems that I can't find some key functions used by some sparse solvers, i.e., the following functions:
DZNRM2
DNRM2
ZGEMV
ZDOTC
ZSWAP
ZTRSM
ZGEMM
ZCOPY
ZTRSV
ZGERU
ZAXPY
ZSCAL

From the CUBLAS, I can only find two functions:
DNRM2
ZGEMM

Do you have any plan to develop all BLAS level functions, include double complex data type?

Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Re: CULA and CUBLAS

Postby dan » Mon Nov 30, 2009 11:48 am

Hi Zhanghong,

The functions you're looking for are BLAS level functions. The GPU-accelerated BLAS library, CUBLAS, is an NVIDIA library and is currently actively developed. In the next version of CUDA, 3.0 (currently in beta), I believe that all of these functions will be available.

Dan
dan
Administrator
 
Posts: 61
Joined: Thu Jul 23, 2009 2:29 pm

Re: CULA and CUBLAS

Postby tang » Tue Dec 01, 2009 5:25 am

Dear Dr. Dan,

Thank you very much for your kindly reply. Have you released the beta version of CUDA 3.0? From your website I only find the latest version 1.1. How can we get the beta version of CUDA 3.0?

Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Re: CULA and CUBLAS

Postby kyle » Tue Dec 01, 2009 6:19 am

The current beta version of NVIDIA's CUDA 3.0 is currently only available to NVIDIA Registered Developers. You can register for that program here:

http://developer.nvidia.com/page/regist ... ogram.html
kyle
Administrator
 
Posts: 301
Joined: Fri Jun 12, 2009 7:47 pm

Re: CULA and CUBLAS

Postby tang » Tue Dec 01, 2009 6:50 am

Dear Dr. Dan,

Thank you very much for your so quick reply.
However, I am still confused. So now there are 3 different libraries accelerated by GPU, i.e., CUDA, CULA and CUBLAS, could you please tell me the difference between these libraries?

Now it seems that the CUDA 3.0 is what I looked for. I have the G220 1GB Nvidia card, but from the download page
http://forums.nvidia.com/index.php?showtopic=149959
I found so many versions, which one should I download?

Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Re: CULA and CUBLAS

Postby john » Tue Dec 01, 2009 7:01 am

Hi - to summarize the differences for you:

CUDA is Nvidia's language for developing Nvidia-based GPU applications.
CUBLAS is Nvidia's CUDA-based implementation of BLAS, which are simple linear algebra routines (dot product, matrix-vector product, matrix-matrix product, etc)
CULA is our product, and is also a CUDA-based library similar to CUBLAS. CUDA accelerates high-level linear algebra concepts (system solve, QR, SVD, eigenvalues, inverse, least squares, etc.)

CUDA is where you will want to begin for developing your own GPU programs unless you need any of the concepts covered by CUBLAS/CULA (in which case I encourage you to try them). When you get CUDA you will need to select the appropriate: driver, toolkit, and SDK (SDK is optional but useful.)
john
Administrator
 
Posts: 587
Joined: Thu Jul 23, 2009 2:31 pm

Re: CULA and CUBLAS

Postby tang » Tue Dec 01, 2009 7:14 am

Dear Dr. John,

Thank you very much for your so detailed explain. Do you mean that you suggest me to use the CULA? But from your document I didn't find the functions I listed above. Could you please tell me which library provide these functions?

BTW: maybe latter we will use the functions provided in CULA, such as GETRF, GEEV, but from the document I only fund the functions like "culaGetrf" and "culaGeev", they still can't replace the LAPACK library directly, at least we have to write some interface code to call these functions, am I right?

Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Re: CULA and CUBLAS

Postby dan » Wed Dec 02, 2009 7:57 am

Hi tang,

The first product you want to explore is CUBLAS, which you can get from the NVIDIA CUDA toolkit. It will provide many of the functions you listed above. As far as which version to get in the link you posted above, the version you want to install depends on your operating system, so if for example you're running 32-bit windows, get "XP32 195.39" (driver) and CUDA Toolkit for Windows 32-bit.

After trying out CUBLAS and it's BLAS functions, you might then want to try CULA, which provides LAPACK-style functions. When you do, explore the "bridge" interface, which already provides the interface code to call these functions in a manner that matches to existing LAPACK packages. Alternatively, you can write code to target one of CULA's other interfaces, but it seems like the bridge interface is the one you want to try first.

Dan
dan
Administrator
 
Posts: 61
Joined: Thu Jul 23, 2009 2:29 pm

Re: CULA and CUBLAS

Postby tang » Wed Dec 02, 2009 10:43 am

Dear Dr. Dan,

Thank you very much for your kindly reply. Your explain makes me clear up. I will try to learn and use it and if I have any problem, I will ask for your help again.

Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Re: CULA and CUBLAS

Postby tang » Mon Dec 14, 2009 6:27 pm

Dear Dan,

I have another question about the hardware. I got from others that there are two series of graphic card: GTX series and C1060. What's the difference between?

From one forum I got the following information:


"Just FYI,
GTX295 is a graphic card which come with a display output. but Tesla is not a graphic card come with display.

Tesla C1060 only a card which you install in your system & need installation of CUDA driver. It mostly help on rendering process, sometimes maybe some other programs. But it doesn't come with Display out at all.

So there is no comparison between Tesla & GTX295 we can make, unless someone who have tried it out to together with benchmark given showed it really improve your gaming experience after install Tesla.

Just my 2 cents. Thanks."

the link is as follows:
http://www.tomshardware.com/forum/263294-33-tesla-c1060

Could you please suggest a hardware configuration for GPU programming to use the CUDA and CUBLAS (cost performance)?

Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Re: CULA and CUBLAS

Postby john » Tue Dec 15, 2009 4:01 pm

The answer to this question is that the GTX products are targeted at gaming/graphics and the C1060 (and other Tesla products) are meant for computing. You can use a GTX for compute, but you cannot use a Tesla for graphics (because it does not have a display output.)

I would recommend the C1060 because it has very large onboard memory - 4 GB vs 1 GB for a GTX285. The GTX295 is basically two GTX285 parts on one card.

It's hard to make a recommendation without knowing a budget, but for a little over $2000 USD you can assemble a system with a Core i7 processor, 8+ GB RAM, and a C1060.
john
Administrator
 
Posts: 587
Joined: Thu Jul 23, 2009 2:31 pm

Re: CULA and CUBLAS

Postby tang » Tue Dec 15, 2009 5:12 pm

Dear John,

Thank you very much for your kindly reply. I asked this problem is because I found that someone said his program can work on C1060 but can't work on GTX285, So I guess that the GTX285 doesn't support some functions that C1060 support.

So your conclusion is that the GTX 285 can support ALL functions that C1060 supported and the ONLY difference is that C1060 has a much larger onboard memory?

Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Re:CULA and CUBLAS

Postby tang » Thu Jan 07, 2010 8:17 am

Dear administrator,

I have another question: I noticed that the CUDA have XP64 version and Vista/Win7 64 version, they are both for 64 bit windows system and if I develop a program with CUDA XP64 version, can I run it on a machine which have the Vista/Win7 64 installed?


Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Re:CULA and CUBLAS

Postby john » Thu Jan 07, 2010 8:32 am

Yes, that should work.
john
Administrator
 
Posts: 587
Joined: Thu Jul 23, 2009 2:31 pm

Re:CULA and CUBLAS

Postby tang » Thu Jan 07, 2010 5:17 pm

Dear Dr. John,

Thank you very much for your reply. I forgot to ask that if I run the program in a different OS, should I copy the correspond runtime libraries of CUDA to that machine? I wish both the same program and runtime libraries of CUDA can run in different OS, XP 32/Vista 32/Win7 32 or XP 64/Vista 64/Win7 64.


Thanks,
Zhanghong Tang
tang
 
Posts: 12
Joined: Sat Nov 28, 2009 10:28 am

Next

Return to Windows Support

Who is online

Users browsing this forum: No registered users and 1 guest

cron