A special thanks to all the engineers at NVIDIA for joining our community and thanks of course to our community for coming out and asking some great questions.
| Timestamp |
Nick |
|
Message |
| |
| 05/28/2009 2:58:31 PM |
EVGA_Marketing |
|
Welcome everyone, we are going to be put the chat into the moderated mode. Once i give you the go ahead please submit your questions. We have some EVGA shirts and also some great games to giveaway for good questions. |
| |
| 05/28/2009 2:58:56 PM |
EVGATech_JacobF |
|
Go ahead and post your questions now |
| |
| 05/28/2009 2:59:36 PM |
EVGATech_JacobF |
|
Remember |
| |
| 05/28/2009 3:00:14 PM |
NVIDIA_Philip |
|
Hello everyone - we'll kick off the chat in just a moment |
| |
| 05/28/2009 3:00:28 PM |
NVIDIA_Patrick |
|
Hi everyone! |
| |
| 05/28/2009 3:02:25 PM |
EVGA_Marketing |
|
Alot of great questions coming in everyone, the NVIDIA team is working to get answers now and you will see them posted shortly |
| |
| 05/28/2009 3:04:54 PM |
NVIDIA_Philip |
|
Q : Blazen22: Without CUDA would there be Physx?
A : PhysX works on both GPUs and CPUs. However, CUDA is an architecture (GeForce GPUs) that allows GPUs to process PhysX much faster than CPUs |
| |
| 05/28/2009 3:05:52 PM |
PoppaSquat |
|
Do your mobile graphics chips support CUDA? If so how much of an increase of performance is seen? |
| |
| 05/28/2009 3:06:58 PM |
NVIDIA_Philip |
|
Mobile GPUs do support CUDA - GeForce 8 series and newer. |
| |
| 05/28/2009 3:07:01 PM |
NVIDIA_Manju |
|
Q : bomberboysk: Do you think that (ugh)ati might become cuda compatible in the future?
A : ATI's compute architecture is not as versatile as the NVIDIA's GPU computing architecture and cannot handle C code and so it will not be able to keep up with CUDA |
| |
| 05/28/2009 3:08:13 PM |
NVIDIA_Patrick |
|
Q : bomberboysk Since nvidia cuda helps offload much of the current cpu intensive tasks, how hard is it for developers to code existing programs to function with cuda
A : CUDA has a C API so it is actually quite easy for developpers to move their C algorithms to CUDA. This is the beauty of CUDA. Other interfaces like OpenCL and DirectX Compute are also being developped. |
| |
| 05/28/2009 3:08:52 PM |
NVIDIA_Manju |
|
Q : bomberboysk: With CUDA, on average how many times more effecient is say, a GTX285 than the latest cpu offerings from say, intel.
A : The GTX 285 can be a couple orders of magnitude greater in performance tahn the latest quad core CPUs. There are hundreds of third party performance application posted on cuda zone at nvidia.com/cuda |
| |
| 05/28/2009 3:08:55 PM |
NVIDIA_Philip |
|
Q : How much of performance increase is CUDA to a game?
A : CUDA is an architecture of GeForce, which allows the GPU to utilize parallel processing apis - like PhysX |
| |
| 05/28/2009 3:09:03 PM |
NVIDIA_Nadeem |
|
Q : Atlas Folder: When can we expect the CUDA GPU limit to be raised from 8?
A : CUDA already supports more that 8 GPU - we have TESLA Clusters with 100's of GPU's |
| |
| 05/28/2009 3:09:17 PM |
NVIDIA_Michael |
|
Q : spitfire5792 Do you think CUDA has been successful in achieving what it was made for?
A : Absolutely. CUDA was created to provide a scalable platform for ISVs to enable great GPU computing apps as simply as possible. In the years that its been available, it has grown to support a variety of development approaches, such as C, OpenCL and DirectX Compute. Its enabled by a strong ecosystem of proven development tools, academic research and the largest number of shipping GPU computing applications and is now supported by over 100 million NVIDIA GPUs. |
| |
| 05/28/2009 3:10:29 PM |
NVIDIA_Philip |
|
Q : Blazen22: Where is CUDA located within the card?
A : CUDA is an overall architecture of a GPU that allows for parallel processing |
| |
| 05/28/2009 3:10:56 PM |
NVIDIA_Nadeem |
|
Q : Blazen22: How does CUDA use real-time ray tracing for graphics?
A : There are raytracing solutions which use CUDA to accelerate many parts of the ray tracing pipline - if that doesn't answer your question please ask again in a different way |
| |
| 05/28/2009 3:11:36 PM |
NVIDIA_Philip |
|
Q : Blazen22: How much time and research went into developing Cuda?
A : NVIDIA spends aobut 2 years developing a new GPU architecture from start to market - so quite a while! |
| |
| 05/28/2009 3:11:43 PM |
NVIDIA_Nadeem |
|
Q : Blazen22: Without CUDA would Tesla cards be able to supercompute?
A : The Tesla brand was launched to leverage CUDA. |
| |
| 05/28/2009 3:12:25 PM |
NVIDIA_Patrick |
|
Q : bomberboysk: Since the development/release of CUDA, how many apps are now out that are cuda supported?
A : There are currently hundreds of CUDA-enabled applications available in the market. You can refer to http://www.nvidia.com/object/cuda_home.html# for more details. |
| |
| 05/28/2009 3:12:48 PM |
NVIDIA_Manju |
|
Q : wakkoguy: Where did the idea for CUDA come from?
A : a few years ago many university research to ers started using shader programming tp use the massive number of GFLOPS on a GPU to do scientfic problems. This was called GPGPU the programmign model was diffiult. So NVIDIA developed the CUDA archietcture to greatly ease the programming model for general compute applications |
| |
| 05/28/2009 3:13:33 PM |
NVIDIA_Philip |
|
Q : wakkoguy When will the other interfaces like OpenCL be available?
A : Open CL was developed on GeForce. GeForce GPUs take advantage of several parallel processing APIs like CUDA, PhysX, Direct3D, etc. |
| |
| 05/28/2009 3:13:52 PM |
NVIDIA_Nadeem |
|
Q : wakkoguy: Do you think CUDA will help the GPU become the dominant processor in the future?
A : We promote the best use of all the processors in the system - so the best programs will use all the processors. But since performance per watt is much better for GPU it is really possible that the GPU will represent a larger percentatge of the over computing power of a PC |
| |
| 05/28/2009 3:14:15 PM |
NVIDIA_Manju |
|
Q : shud_: Can CUDA help with in game water, lighting, textures, and more?
A : Yes, defeinitely, CUDA is already used in game water through the use of PhysX which is written on top of CUDA. Lighting is also acceleratable with CUDA |
| |
| 05/28/2009 3:15:24 PM |
NVIDIA_Philip |
|
Q : Ceej I use Badaboom and am absolutely blown away by how much faster and easier it has made video converting. What other applications are getting ready to be released?
A : Great! Other existing apps are Motion DSP's vReveal, which cleans up grainy and choppy video in real time, and ArcSoft which converts DVDs to near HD quality in real time...more on the way soon! |
| |
| 05/28/2009 3:15:33 PM |
NVIDIA_Patrick |
|
Q : spitfire5792: As newer cards come out, what should we be expecting from CUDA?
A : Yes. CUDA has been part of our GPU since the GeForce 8 series and continues to be part of our futur GPU plans. There are currently more than 120M CUDA-enabled GPUs on the market. |
| |
| 05/28/2009 3:16:13 PM |
NVIDIA_Michael |
|
Q : Skydive: How is NVIDIA working with Stanford to improve drivers and make folding more powerful?
A : Distributed computing is a great platform for using the GPU, obviously. Weve worked with the Pande Group at Stanford quite a bit to ensure their parallel processing environment is well tuned for GPU computing. In some cases weve provided technical direction for optimization on CUDA and weve seen huge performance gains vs both CPU and other GPUs. |
| |
| 05/28/2009 3:16:25 PM |
NVIDIA_Nadeem |
|
Q : wakkoguy: How does CUDA talk to other parts of the CPU, Memory, and Northbridge?
A : The CUDA runtime which runs on the host communicates to the GPU via the northbridge and uses shared memory on the host and DMA to exchange data and code between CPU and GPU |
| |
| 05/28/2009 3:16:27 PM |
NVIDIA_Manju |
|
Q : hosps: What are the advantages/disadvantages between using CUDA vs. OpenCL vs. DirectX Compute?
A : CUDa is NVIDIA's GPU computing architecture and it runs C with CUDA extensions, OpenCl and DX Compute. The differences btweent hese are stylisitc. Today the only one avalableto develop applications with is C with CUda extensions |
| |
| 05/28/2009 3:17:10 PM |
NVIDIA_Nadeem |
|
Q : Kryptogen: Will C++ be available in the future for CUDA?
A : yes in the works |
| |
| 05/28/2009 3:17:42 PM |
EVGATech_JacobF |
|
Good question, hosps and kyptogen email me at jacobf@evga.com you won a free copy of Baddaboom. Put Baddaboom in the subject header. |
| |
| 05/28/2009 3:18:38 PM |
NVIDIA_Nadeem |
|
Q : Kryptogen: Since the calculations done with the CUDA tech. are very fast how important is the other hardware in the computer so there won't be any bottlenecks?
A : depends on the application - hetrogenous computing depends on the balance. |
| |
| 05/28/2009 3:19:07 PM |
NVIDIA_Philip |
|
Q : spitfire5792 How does CUDA work with the new 3D technology?
A : Today, there is no relation. |
| |
| 05/28/2009 3:19:24 PM |
NVIDIA_Patrick |
|
Q : hosps: What are the advantages/disadvantages between using CUDA vs. OpenCL vs. DirectX Compute?
A : OpenCL and DirectX_Compute are NOT competing against CUDA they are different ways of accessing CUDA (in addition to C, Fortran, etc.) |
| |
| 05/28/2009 3:19:44 PM |
NVIDIA_Nadeem |
|
Q : wakkoguy When will the other interfaces like OpenCL be available?
A : We already have an OpenCL driver for signed up developers - its undergoing qualification as we speak. |
| |
| 05/28/2009 3:19:44 PM |
NVIDIA_Manju |
|
Q : bomberboysk: Since cuda is a highly paralell system, will there be/is there already support for multiple cuda enabled devices(like sli), and if so, how well will the scaling be(near 100% i would imagine..)
A : yes there is already support for multiple CUDA devices and in fact games using PhysX can use multiple CUDA enabled devices already |
| |
| 05/28/2009 3:20:41 PM |
NVIDIA_Nadeem |
|
Q : spitfire5792: As newer cards come out, what should we be expecting from CUDA?
A : CUDA is going to get better and better :) |
| |
| 05/28/2009 3:21:38 PM |
NVIDIA_Manju |
|
Q : Blazen22: How much Algorithm is put into CUDA?
A : CUDA is a higly programmable architecture and you can program any algorithm into it. It gives great performance speedups for those algorithms where there is a lot of parallelism. There are a hundreds of example you can find at nvidia.com/cuda |
| |
| 05/28/2009 3:22:03 PM |
NVIDIA_Nadeem |
|
Q : shud_: Can CUDA help with in game water, lighting, textures, and more?
A : CUDA kernels can interop DirectX and OpenGL, and can help with graphical operations |
| |
| 05/28/2009 3:22:06 PM |
NVIDIA_Philip |
|
Q : spitfire5792: Is it possible to use CUDA to overclock your computer, and if so will it increase the amount of heat given off, less than other ways of overclocking would?
A : no you cannot use CUDA for overclocking |
| |
| 05/28/2009 3:23:45 PM |
NVIDIA_Philip |
|
Q : FreeJet: What is more important, CPU or GPU?
A : As more and more applicaitons gain more relative performance using a GPU versus a CPU, the GPU will become more and more important |
| |
| 05/28/2009 3:24:20 PM |
NVIDIA_Michael |
|
Q : spitfire5792: Ati has been a long time competitor of yours, do you think they will develop something similar?
A : Its hard to speak specifically for what theyre doing, but I believe the goal for all of us to ensure that as many ISVs as possible can develop scaleable high quality GPU computing apps. To do that, you need to ensure there is a well supported ecosystem of development approaches. Thats the reason for CUDA. Some have called CUDA proprietary, but as an architecture native to NVIDIA GPUs, it supports many different approaches for GPU computing... C, OpenCL, DirectX Compute, even Fortran. The reason CUDA has been so successful is that its well established, well understood, and supportive of many development approaches. |
| |
| 05/28/2009 3:24:43 PM |
NVIDIA_Philip |
|
Q : alynd: Thank you for giving us the opportunity to grill you. Having said that, my 10 year old son is interested addicted to video games (in a good way) and wants to learn how to code graphics so that he can mod existing games or start learning to code his own. The box I built him has my old EVGA 8800 GTX so he is (in theory) able to start learning to code on that. What would be some good resources to start him with, particularly for his age and newbiness at coding.
A : nvidia.com/cuda |
| |
| 05/28/2009 3:25:10 PM |
NVIDIA_Patrick |
|
Q : hosps: Which part of the cards architecture is more critical to overall CUDA performance (memory speed, shader speed, Core speed or all three)?
A : Good question! The number of GPU cores comes first (by far), then clock speed, then memory speed. For example, the Badaboom and vReveal performance scale almost directly proportionally with the number of GPU cores. |
| |
| 05/28/2009 3:25:28 PM |
NVIDIA_Nadeem |
|
Q : Kyle_Bennett: Are we going to see games any time soon (this year) that heavily utilize PhysX in a fashion other than effects physics? If so, would the demands on Cuda and PhysX be so much that the game would not be playable with gameplay physics enabled without a GeForce GPU?
A : There are games already which use "PhysX" for game play - but generally GPU accelerated PhysX has been limited to effects, too allow for multi-player games to be fair. There are some developers which are considering using accelerated PhysX on servers to help to handle the large game play impacting loads. There is nothing in the tech which prevents using really heavy GPU physX for game play - it would boil down to a developer decision. |
| |
| 05/28/2009 3:25:43 PM |
NVIDIA_Manju |
|
Q : micah_675: Are there any plans to use CUDA in servers as the primary means of processing power?
A : CUDA is already on several supercomputers and in fact powers the Tsubame Computer at the Tokyo Institiute of Technology whcih is 29th inthe world |
| |
| 05/28/2009 3:26:09 PM |
NVIDIA_Philip |
|
Q : Skydive: Does NVIDIA have plans to use CUDA and it's GPU processors outside of the graphics industry?
A : we already do! Tesla is a great example of that |
| |
| 05/28/2009 3:28:25 PM |
NVIDIA_Philip |
|
Q : Parelem: The games that I have seen with PhysX have been absolutely amazing, can we expect more developers to take advantage of CUDA and PhysX in the near future as there isn't a large selection of games using it now?
A : CUDA is an architecture - a way of designing a GPU to process things in a parellel manner. PhysX is an API that takes advantage of a CUDA-enabled GPU |
| |
| 05/28/2009 3:29:16 PM |
NVIDIA_Nadeem |
|
Q : bomberboysk: Do you have any plans of releasing a dedicated CUDA processing unit, say, that would fit into an expresscard slot of a laptop?
A : CUDA runs on NVIDIA GPU . Mobile GPU can be on MXM modules or express cards, which could be a plugin i'm not aware of any commercially available units |
| |
| 05/28/2009 3:29:43 PM |
NVIDIA_Philip |
|
Q : PoppaSquat How long has Cuda been around and did Nvidia expect it to get at the level it is at now?
A : CUDA has been a part of NVIDIA GPU design since we launched the GeForce 8 series |
| |
| 05/28/2009 3:29:54 PM |
NVIDIA_Patrick |
|
Q : FreeJet: What is more important, CPU or GPU?
A : The GPU, of course!! ;) So much processing power in these transistors! The beauty of the CUDA SDK is that it exposes this humongous processing power to every programmer! |
| |
| 05/28/2009 3:30:25 PM |
NVIDIA_Manju |
|
Q : spitfire5792: What new things does CUDA offer that was never offered before?
A : CUDA allows developers to apply the massive parallel processing of the GPU (an order of magnitude more GFLOPS) to general pcompute problems with the same ease of programming as writing a simple C program |
| |
| 05/28/2009 3:31:51 PM |
NVIDIA_Michael |
|
Q : hosps: With Microsofts upcoming .NET 4.0 has added Parallel Extensions to the .NET environment, any chance of a CUDA interface to add the same type of parallelism ease?
A : The CUDA programming environment and SDK already provides fairly extensive support of, and approaches to parallel computing. The great thing about this aspect of .NET is that it further demonstrates momentum in parallel processing. |
| |
| 05/28/2009 3:31:53 PM |
NVIDIA_Nadeem |
|
Q : clamatowas: Could you develop an OS on CUDA. Removing the Intell or AMD CPU all together and make a computer that is NVIDA SuperCharged?
A : The current GPU architecture require a host. The telsa supercomputers still use host CPU's to launch Kernels. Heterogenous computing depends on a good balance between serial and parallel processing. |
| |
| 05/28/2009 3:32:27 PM |
NVIDIA_Manju |
|
Q : Kryptogen: What advantages does CUDA offer over other parallel processing technologies?
A : There is an installed base of over 110M CUDA enabled processors and that is something that the world has never experienced and is trransformational. There is no other parallel processing technlogy that has been close to that |
| |
| 05/28/2009 3:32:43 PM |
NVIDIA_Patrick |
|
Q : Nomak02: does CUDA improve my cinematic viewing?
A : Definitely! The SimHD plug-in from ArcSoft is a very good example of that! |
| |
| 05/28/2009 3:33:05 PM |
NVIDIA_Philip |
|
Q : Yin~Yang Does CUDA and PhysX work as one?
A : CUDA is part of the architecture of our GPUs. PhysX is the API that works using either a CPU or our GPU (for a much, much better experience) |
| |
| 05/28/2009 3:33:26 PM |
NVIDIA_Nadeem |
|
Q : Atlas Folder: Under Vista/Windows 7 is nVidia working with Microsoft to remove the "must extend the desktop" to get CUDA functioning on multiple GPUs?
A : We already have a solution for VISTA - check out th SLI/PhysX control panel. Things are slightly different for Win7, but we'll have a solution for you. |
| |
| 05/28/2009 3:34:12 PM |
NVIDIA_Nadeem |
|
Q : hosps: Which part of the cards architecture is more critical to overall CUDA performance (memory speed, shader speed, Core speed or all three)?
A : Its depends on the application but in general the number of processors, core clock and memory bandwidth all have a major impact |
| |
| 05/28/2009 3:34:30 PM |
NVIDIA_Philip |
|
Q : tylerritchie: Is OpenCL going to replace CUDA, or co-exist with CUDA?
A : They will co exist. GeForce GPUs accelerate PhsyX, Direct3D and Open CL applicaitons. We were the first to market with an OpenCL driver earlier this month. |
| |
| 05/28/2009 3:35:01 PM |
NVIDIA_Michael |
|
Q : spitfire5792: How will CUDA change when Directx11 enabled cards come out?
A : It doesnt really need to change. Thats one of the great things about CUDA. Its consistent. Thats why developers are able to work effectively with it. The CUDA architecture is resident on all our DX10 GPUs and will also be on all new GPUs, including DX11. |
| |
| 05/28/2009 3:35:49 PM |
NVIDIA_Nadeem |
|
Q : Ceej: What percentage of the GPU does CUDA use for its application?
A : There are some applications which use 100% of the GPU for CUDA. The telsa cards for example do not have graphics output and are 100% CUDA |
| |
| 05/28/2009 3:36:27 PM |
NVIDIA_Manju |
|
Q : Blazen22: Can you lay out how Physx,CUDA,and the gpu itself work together?
A : PhysX is a an API that uses the CUDA architecture on the GPU to accelerate Algortithms |
| |
| 05/28/2009 3:36:40 PM |
NVIDIA_Philip |
|
Q : will160628: Are there any new CUDA applications or API features either recently released or soon to be released that you are truly excited about?
A : Motion DSP' |
| |
| 05/28/2009 3:36:55 PM |
NVIDIA_Philip |
|
Motion DSP's vReveal is what I'm most excited about |
| |
| 05/28/2009 3:37:11 PM |
NVIDIA_Nadeem |
|
Q : PoppaSquat Will texture rendering be supported in future versions of CUDA?
A : Its in the pipeline |
| |
| 05/28/2009 3:37:54 PM |
NVIDIA_Manju |
|
Q : chizow: How can ATI cards run OpenCL if they cannnot handle C code?
A : In their latest architecture ATI has put in support for those aspects of C that are supported in OpenCL. There is no release of ATI of openCL implementations to date |
| |
| 05/28/2009 3:38:06 PM |
NVIDIA_Philip |
|
Q : Yin~Yang Can older cards become CUDA enabled?
A : CUDA is an architecture INSIDE the GPU - with the 8 series and up - you cannot enable it with software if it's not there in the hardware already. |
| |
| 05/28/2009 3:38:19 PM |
NVIDIA_Nadeem |
|
Q : bomberboysk I remember reading somewhere where a hospital used cuda to process images from some kind of scan(ct or mri, cant remember), is this a common application for the cuda platform?
A : Yes this a great application for CUDA - check out CUDAZone application gallery - there are some great examples of 3D Tomography etc. its saving lives and saving money |
| |
| 05/28/2009 3:39:26 PM |
EVGATech_JacobF |
|
Good question bomberboy send me an email at jacobf@evga.com for a free copy of Baddaboom. |
| |
| 05/28/2009 3:39:39 PM |
NVIDIA_Patrick |
|
Q : will160628: Are there any new CUDA applications or API features either recently released or soon to be released that you are truly excited about?
A : NVIDIA recently added a video encoding acceleration library to CUDA, which makes it easier for third-party software vendors to take advantage of the GPU for encode and transcoding tasks in their applications. |
| |
| 05/28/2009 3:40:03 PM |
NVIDIA_Nadeem |
|
Q : unvjarhead: Does CUDA perform differently on professional GPUs (Quadro) when compared to a similarly spec'd GeForce card?
A : Quadro cards have been tested to much higher levels for professional levels of reliable operation and have large framebuffers for CUDA applications |
| |
| 05/28/2009 3:40:07 PM |
NVIDIA_Manju |
|
Q : onrul: Does CUDA mean less CPU emphasis, and more GPU emphasis in the future PC manufacturer market?
A : As more and more critical compute hungry apps start using CUDA and GPU computing the emphasis on GPUs inthe PC market will increase |
| |
| 05/28/2009 3:40:10 PM |
NVIDIA_Michael |
|
Q : micah_675: Are there any plans to use CUDA in servers as the primary means of processing power?
A : Sure. NVIDIA GPUs and CUDA are perfect for many server-based apps, which are often naturally parallel processing environments. In fact, we have a product called Tesla built specific for high performance computing. http://www.nvidia.com/page/servers.html |
| |
| 05/28/2009 3:42:14 PM |
EVGA_Marketing |
|
Hang on guys, had a small connection issue. NVIDIA Team will be back shortly |
| |
| 05/28/2009 3:42:22 PM |
NVIDIA_Philip |
|
Q : spitfire5792 What are your recommendations for the best cards to use for CUDA?
A : CUDA performance increases based oof cn the number oresvvGPU.....fvv |
| |
| 05/28/2009 3:42:39 PM |
NVIDIA_Nadeem |
|
Q : chizow Do you think GPGPU will ever progress to the point traditional x86 CPUs are no longer necessary? What major change would allow for this? Simply a non-x86 based OS?
A : We are promotion hetrogenous computing using both CPU and GPUs for best use of the PC |
| |
| 05/28/2009 3:42:50 PM |
zerran2001 |
|
joe broke it already!! [Emotion=emwink.gif] |
| |
| 05/28/2009 3:43:00 PM |
NVIDIA_Philip |
|
yes but he fixed it - we're back! |
| |
| 05/28/2009 3:43:27 PM |
NVIDIA_Manju |
|
Q : shud_ shud_: you said that it takes 2 years to come up wt CUDA, does this mean that only every 2 years there is a big update to CUDA or are we going to see updates monthly?
A : New versions o the CUDA toolkit, SDK and Drivers are added regularly, CUDA 2.1 was releases in Decemeber 2008 and CUDA 2.2 was released in May |
| |
| 05/28/2009 3:43:57 PM |
NVIDIA_Nadeem |
|
Q : FreeJet: Tell us something about that cool Tesla C1060 computing processor I see advertised on your site.
A : Its great it has 4GB of memory which is great for big datasets |
| |
| 05/28/2009 3:45:03 PM |
NVIDIA_Philip |
|
Q : Ceej Do you think that it is possible that in the future "too many" applications will use the GPU, making the benefits lesser?
A : No! the more the better! |
| |
| 05/28/2009 3:45:03 PM |
NVIDIA_Nadeem |
|
Q : mberboysk: Are there any current plans to be able to run a virtual machine accelerated(emulating an x86/x86-64) by cuda?
A : no plans i'm aware off. |
| |
| 05/28/2009 3:45:51 PM |
NVIDIA_Nadeem |
|
Q : TaRp1960 Will there ever be adjustment in the control panel for CUDA
A : We may have options to select which GPU to select for CUDA and stuff like that |
| |
| 05/28/2009 3:45:55 PM |
NVIDIA_Manju |
|
Q : bomberboysk: How large is the team of developers that worked on CUDA?
A : All the major teams in NVIDIA are working on CUDa including the Architecture team,the software team, the compiler team, the chip team, the application team... It's hard to estimate since it isan inseparable part of the GPU but I would sayover 2000 people are working on developing CUDA technology |
| |
| 05/28/2009 3:46:21 PM |
NVIDIA_Nadeem |
|
Q : RP520 Is there any possibility in the future of CUDA be used with XNA?
A : XNA runs on XBOX360 and that does not support CUDA |
| |
| 05/28/2009 3:47:04 PM |
NVIDIA_Patrick |
|
Q : LouisRoss: As a developer (not games) I would like to propose to my company that we accelerate some of our code using CUDA. However, it would fly better if I could show it scaling automatically from CPU to multi-core CPU to GPU, without code changes. Is this available?
A : If your code is in C, then the transition will be very easy. |
| |
| 05/28/2009 3:47:07 PM |
NVIDIA_Philip |
|
Q : Nomak02 reading the nvidia website about CUDA always really confused me and gave me a headache, is there a simpler answer to what it does for my gaming?
A : CUDA is the architecutre of the GPU - how it's designed to utilize parallel processing APIs....like PhysX. Using a CUDA designed GPU will allow you to use GPU-powered PhysX instead of default CPU physX |
| |
| 05/28/2009 3:47:19 PM |
NVIDIA_Michael |
|
Q : Yin~Yang: And what is Motion DSP?
A : They make vReveal, a great new app that enables one-click enhancement of your personal videos. Its amazing, actually! You can take flawed video and easily clean it up with auto-contrast, resolution, stabilization, lighting, etc With NVIDIA GPUs, this can be done as much as 5x faster than a typical CPU. Check it out http://www.vreveal.com/ |
| |
| 05/28/2009 3:47:32 PM |
NVIDIA_Philip |
|
Q : nke12: we can trust CUDA technology?
A : If you trust Joe, you can trust CUDA technology. It's not going anywhere |
| |
| 05/28/2009 3:47:32 PM |
NVIDIA_Nadeem |
|
Q : Nomak02 reading the nvidia website about CUDA always really confused me and gave me a headache, is there a simpler answer to what it does for my gaming?
A : check out the PhysX sections - hopefully they don't gove you a headache (but sorry if it does ) |
| |
| 05/28/2009 3:47:39 PM |
NVIDIA_Manju |
|
Q : Ceej In the future, will it be possible to dedicate a GPU specifically for CUDA applications?
A : This is already possible today. Using the NVIDIa control panel you can dedicate a GPU to a CUDA application |
| |
| 05/28/2009 3:48:32 PM |
NVIDIA_Philip |
|
Q : spitfire5792 What kind of(if any) detrimental effects can CUDA have on a computer, should we be worried about anything?
A : Absolutely none! |
| |
| 05/28/2009 3:48:58 PM |
NVIDIA_Manju |
|
Q : RP520 Is there any possibility in the future of CUDA be used with XNA?
A : Yes we are planning on adding support for other high level languages including FORTRAN |
| |
| 05/28/2009 3:49:50 PM |
NVIDIA_Philip |
|
Q : Nomak02: does CUDA help a lot with folding home?
A : It helps a lot! Have you checked out "team whoopass" |
| |
| 05/28/2009 3:50:01 PM |
NVIDIA_Patrick |
|
Q : PoppaSquat What are future plans to introduce CUDA supported programs to Mainstream consumers?
A : CUDA is available to consumers applications developers and many of them have been releasing applications in the consumers market (Ex. MotionDSP, CyberLink, Elemental technologies, Nero, ArcSoft, CoreCodec, etc.) with real performance benefits to the users. |
| |
| 05/28/2009 3:50:13 PM |
NVIDIA_Nadeem |
|
Q : chizow From what I've read CUDA is the high-level API and runtime tools. Some comments here suggest CUDA also encompasses DX Compute and OpenCL as low-level driver API along with Nvidia's native C for CUDA. Is this correct? There just seems to be much confusion with regards to Nvidia's stance on OpenCL and DX Compute support and even many of the answers here seem to dodge the question.
A : Sorry for the confusion. CUDA is our architecture for parallel computing - whcih includes the processing cores the languages which run on the procesors and we have SDK to program it directly. As such our OpenCL and DirectX compute solutions leverage our CUDA architecture. I hope that helps. |
| |
| 05/28/2009 3:50:30 PM |
EVGATech_JacobF |
|
RP520, Nomak, chizow and Poppasquat, email me at jacobf@evga.com with a subject of baddaboom for a free copy. |
| |
| 05/28/2009 3:51:26 PM |
NVIDIA_Nadeem |
|
Q : RP520: Do you foresee integration of CUDA with future Apple, Windows or Linux operating systems?
A : CUDA is already available for Mac and Linux |
| |
| 05/28/2009 3:51:47 PM |
NVIDIA_Philip |
|
Q : wakkoguy: What future game would you like to see use CUDA to the fullest (Bioshock 2?)
A : we're working with top developers to build games that take advantage of GPU PhysX (using CUDA) - several top publishers have signed licenses to develop for PhysX (EA, 2K, and more!)....the more games that take advantage of PhysX on a GPU, the better for you guys! |
| |
| 05/28/2009 3:51:55 PM |
NVIDIA_Michael |
|
Q : Nomak02: does CUDA help a lot with folding home?
A : At one point, NVIDIA GPUs represented 5% of total systems used on Folding@Home, but over 50% of total processing power! |
| |
| 05/28/2009 3:52:03 PM |
NVIDIA_Nadeem |
|
Q : wakkoguy: What future game would you like to see use CUDA to the fullest (Bioshock 2?)
A : I would like to see all future games to use CUDA :) |
| |
| 05/28/2009 3:52:42 PM |
NVIDIA_Manju |
|
Q : Ceej: You said earlier that GPU processing more efficient than CPU processing, why is that?
A : GPU processing is very efficient at accelrating parallel problems since it has architectural features such as shared memeory and hardware scheduling of thousnds of threads that makes it very efficient for these problems |
| |
| 05/28/2009 3:53:16 PM |
NVIDIA_Patrick |
|
Q : Atlas Folder: Do all CUDA algorithms scale linearly with cores?
A : This depends on the how the application is programmed. In the case of Badaboom and vRevel, the scaling is very close to linear. |
| |
| 05/28/2009 3:53:30 PM |
NVIDIA_Philip |
|
Q : TaRp1960 Are CUDA drivers updated with gpu drivers
A : Yes |
| |
| 05/28/2009 3:53:37 PM |
NVIDIA_Nadeem |
|
Q : tylerritchie: How does CUDA compare to AMD FireStream?
A : CUDA has a complete development system and has been used in shipping professional applications. :) |
| |
| 05/28/2009 3:54:02 PM |
EVGATech_JacobF |
|
Good questy tylerritchie email me at jacobf@evga.com for a free copy of VReveal |
| |
| 05/28/2009 3:54:13 PM |
EVGATech_JacobF |
|
Put VReveal in the subject header |
| |
| 05/28/2009 3:55:33 PM |
NVIDIA_Nadeem |
|
Q : crosstip: What do you think the future of GPU as a CPU initiatives like CUDA are? Do you think they are going to become mainstream and be the next adopted fad in the industry?
A : Data parellel computing will be part of mainstream software developement |
| |
| 05/28/2009 3:56:36 PM |
NVIDIA_Philip |
|
Q : clamatowas: Is Cuda a programing platfourm? Or a architecture hardwired into the GPU?
A : CUDA is a design archiecture of GeForce GPUs (8 series and higher); hardwired into the GPU. PhysX is an example of an API that takes advantage of CUDA. Developers can design apps to take advantage of CUDA GPUs using C - the most prevelant language in the world. |
| |
| 05/28/2009 3:56:55 PM |
NVIDIA_Nadeem |
|
Q : mikel_s: I am going into a Game Design & Development course at the Rochester Institute of Technology. How useful do you suppose a rudimentary knowledge of CUDA workings would benefit me?
A : It would help you alot. Game programming have many elements which map to data parallel processing. |
| |
| 05/28/2009 3:57:21 PM |
NVIDIA_Philip |
|
Q : spitfire5792: Does overclocking a GPU cause instability with CUDA other then the norm that overclocking gives?
A : OCing does not affect CUDA performance....unless of course you fry your GPU to death and it no longer functions :) |
| |
| 05/28/2009 3:58:25 PM |
NVIDIA_Nadeem |
|
Q : shud_ Shud_: Is CUDA used for the new Hologram machine for doctors to take a 3d image of the person inner organs and are able to get a precise look of whats in them without having to open up the body? saw it on Science channel!
A : Depends on the machine but CUDA is being used in some shipping medical imaging equipment and others in development |
| |
| 05/28/2009 3:58:49 PM |
NVIDIA_Nadeem |
|
Q : clamatowas When you say "The current GPU architecture require a host" are there any plans to change that? Or is it even possible?
A : Its technically possible |
| |
| 05/28/2009 3:59:23 PM |
NVIDIA_Nadeem |
|
Q : TaRp1960 Will cuda technology ever be used in TV's
A : Its certainly possible - and that usually means it will happen |
| |
| 05/28/2009 3:59:25 PM |
NVIDIA_Patrick |
|
Q : Ceej Can CUDA accelerate audio applications as well?
A : Yes. |
| |
| 05/28/2009 3:59:58 PM |
NVIDIA_Michael |
|
Q : crosstip: What do you think the future of GPU as a CPU initiatives like CUDA are? Do you think they are going to become mainstream and be the next adopted fad in the industry?
A : Also, keep in mind that the GPU and CPU will work together to enable more optimized PCs. The sheer mathematical power and huge number of parallel cores in a GPU will make it essential for the best visual computing apps. GPU will not really be a fad; it will be a key part of your PC or all types of visual computing apps. |
| |
| 05/28/2009 4:00:02 PM |
EVGA_Marketing |
|
We are taking the last few questions and getting answers from the NVIDIA Team, we will be closing out the chat here shortly. |
| |
| 05/28/2009 4:00:18 PM |
NVIDIA_Manju |
|
Q : shud_ shud_: Can CUDA replace RISC & CISC in the cpu in the future?
A : CUDA is part of a heterogeneous programming environment and is not intended to replace the CPU. There will always be serial aspects of a problem forwhich CPUS may be appropriate |
| |
| 05/28/2009 4:00:45 PM |
NVIDIA_Nadeem |
|
Q : bomberboysk: Are there any current ways that CUDA acceleration could help say, a file or web server?
A : Yes - I know of some developers using CUDA to implement security protocols on network data packages as they are flying realtime in the server |
| |
| 05/28/2009 4:01:47 PM |
NVIDIA_Nadeem |
|
Q : nke12 we will have better perfomance from cuda if we have an 64 bit version of os?
A : CUDA performance per se is not affected - but an application using CUDA could be affected on the host side |
| |
| 05/28/2009 4:02:10 PM |
NVIDIA_Patrick |
|
Q : Yin~Yang Does CUDA generate more heat when it is being used?
A : Any processor will generate heat when used. |
| |
| 05/28/2009 4:02:51 PM |
NVIDIA_Nadeem |
|
Q : Yin~Yang Does CUDA generate more heat when it is being used?
A : Also please note that performance per watt for GPU is much better than CPU's so using CUDA could save power |
| |
| 05/28/2009 4:04:34 PM |
EVGATech_JacobF |
|
Everyone who won a prize make sure to send me your shipping address when you send the email. |
| |
| 05/28/2009 4:05:13 PM |
NVIDIA_Nadeem |
|
Q : hosps GASS has created a CUDA.NET wrapper for supporting CUDA within the .NET Framework. Have you worked with this company to support it and would you consider this a risky or safe framework to develop with?
A : we have worked with GASS - i'm not personally involved - you should evaluate their solution for your application - GASS is a good company. |
| |
| 05/28/2009 4:05:26 PM |
NVIDIA_Michael |
|
Q : Yin~Yang Why isnt CUDA being used on ATI cards?
A : Both ATI and NVIDIA have been working to enable GPU computing for a while now. But NVIDIA chose to make it easier for software developers by creating CUDA, which is an architecture that is part of every currently shipping NVIDIA GPU. There are over 100 million NVIDIA GPUs with CUDA today. Developers like CUDA because its enabled a consistent environment for them and also supports many standard programming approaches to GPU computing, including OpenCL, DirectX Compute, C, and others. |
| |
| 05/28/2009 4:05:31 PM |
EVGA_Marketing |
|
Ok, everyone, we are finalizing the last questions, then we will open for everyone to say thanks to NVIDIA Team for joining the CUDA Chat today |
| |
| 05/28/2009 4:05:35 PM |
NVIDIA_Manju |
|
Q : theclaww Windows 7 is supposed to sport a feature that allows any GPU, even Intel GMA for parallel processing. How does this compare to CUDA and if CUDA is better, is it because it is built into the GPU architecture?
A : Windows 7 has Direct X compute which will make GPU computing part of the OS. DirectX Compute is supported through NVIDIA's CUDA architecture Windows 7 will get GPU computing support thru the CUDA architecture |
| |
| 05/28/2009 4:05:59 PM |
zerran2001 |
|
thanks you guys go green! |
| |
| 05/28/2009 4:05:59 PM |
EVGA_Marketing |
|
The chat is now in free chat, please say thanks to the NVIDIA TEAM! |
| |
| 05/28/2009 4:06:02 PM |
x1xtomx1x |
|
Are all the prizes given away? |
| |
| 05/28/2009 4:06:03 PM |
NVIDIA_Philip |
|
thanks guys for all the great questions - sorry we couldn't answer them all! |
| |
| 05/28/2009 4:06:12 PM |
mack4033 |
|
thank you team |
| |
| 05/28/2009 4:06:14 PM |
Ceej |
|
Thanks Nvidia and EVGA! :) |
| |
| 05/28/2009 4:06:15 PM |
sigvoror |
|
Thanks for your time |
| |
| 05/28/2009 4:06:16 PM |
clamatowas |
|
Thanks! |
| |
| 05/28/2009 4:06:17 PM |
Kryptogen |
|
thanks dudes |
| |
| 05/28/2009 4:06:19 PM |
Yin~Yang |
|
7 of my questions answered and got nuttin -_- |
| |
| 05/28/2009 4:06:20 PM |
ScrumHalf |
|
Thanks Nvidia! |
| |
| 05/28/2009 4:06:23 PM |
mack4033 |
|
youy rock |
| |
| 05/28/2009 4:06:24 PM |
NVIDIA_Patrick |
|
Thank you everyone for you questions! I hope you enjoyed this chat! |
| |
| 05/28/2009 4:06:24 PM |
nke12 |
|
thank you!! |
| |
| 05/28/2009 4:06:25 PM |
clamatowas |
|
LOL |
| |
| 05/28/2009 4:06:26 PM |
clamatowas |
|
Sucka |
| |
| 05/28/2009 4:06:27 PM |
Skydive |
|
Thanks for joining us again, guys! |
| |
| 05/28/2009 4:06:28 PM |
x1xtomx1x |
|
Thanks everyone! |
| |
| 05/28/2009 4:06:28 PM |
mack4033 |
|
you rock |
| |
| 05/28/2009 4:06:29 PM |
theclaww |
|
Thanks!!!! |
| |
| 05/28/2009 4:06:29 PM |
PoppaSquat |
|
Thank you! |
| |
| 05/28/2009 4:06:29 PM |
EVGA_Marketing |
|
Thank you everyone for joining todays chat. we have a few final questions then we will open it up for everyone to thank the NVIDIA Team for joining the CUDA Chat today. |
| |
| 05/28/2009 4:06:30 PM |
chizow |
|
thanks Nvidia and EVGA, great chat |
| |
| 05/28/2009 4:06:31 PM |
Nomak02 |
|
again thanks NVIDIA you guys never let us gamers down |
| |
| 05/28/2009 4:06:32 PM |
NVIDIA_Nadeem |
|
Thanks everyone - great questions - pleased to see everyone is so interested in CUDA - look out for our next CUDA webinar series |
| |
| 05/28/2009 4:06:32 PM |
TaRp1960 |
|
Thanks for your time |
| |
| 05/28/2009 4:06:33 PM |
humanavatar |
|
Thank you Nvidia team!!! |
| |
| 05/28/2009 4:06:35 PM |
hosps |
|
Thank you for your answers... |
| |
| 05/28/2009 4:06:38 PM |
RP520 |
|
thanks |
| |
| 05/28/2009 4:06:41 PM |
theclaww |
|
Thanks for yesterday too!!!!!! |
| |
| 05/28/2009 4:06:45 PM |
XcaliburFX |
|
thanks guys |
| |
| 05/28/2009 4:06:46 PM |
Nomak02 |
|
will continue to buy your products especially when they're EVGA |
| |
| 05/28/2009 4:06:47 PM |
tylerritchie |
|
Thanks NVIDIA |
| |
| 05/28/2009 4:06:48 PM |
NVIDIA_Michael |
|
Thanks a lot! |
| |
| 05/28/2009 4:06:48 PM |
NVIDIA_Manju |
|
Thank you everybody |
| |
| 05/28/2009 4:06:48 PM |
clamatowas |
|
Yeah |
| |
| 05/28/2009 4:06:52 PM |
boredgunner |
|
thanks all |
| |
| 05/28/2009 4:06:53 PM |
NVIDIA_Philip |
|
keep buying EVGA stuff ;) |
| |
Products
