Again you Scoff at me Sid, to bad you never really read what you Post ...

Crunching on video cards is an intermediate step. . .

. . . the next generation of parallel computing will be on true co-processors:

The NVIDIA Tesla can't come any where near what the ATI HD 58xx Series of Cards can do. The Tesla has only 1/3 the GFLOP's at best that the 5850 has & costs 4 Times the Price, or at least that's what I've read & seen. The Tesla Cards are basically just older Nvidia gaming cards with more memory, I think its just the 200 series chips their using & why it only has 1/3 the GFLOP's the HD 5850 has.

I read the whole Berkeley Article and no where do they talk about Co-Processors, they talk about somehow adding 1000 Processors to the CPU Core only. I haven't heard anything anywhere either about the Tesla being anything remotely designed to be used with BOINC Platform ... At $1200 to $1400 each & 1/3 the Processing Power I doubt there are going to be many takers for the Tesla unless that all changes to the Teslas advantage Price wise & Performance wise ...

We're bringing you another behind the scenes look here at NVIDIA. This week we met with Danny Shapiro, director of marketing for Quadro, who spoke with us about NVIDIAs work with the film industry, and the hand GPUs played in helping Avatar snag the Oscar for best visual effects.

More GPU Routines for Scientific Research - Now Available from Numerical Algorithms Group


Scientists in a wide array of disciplines --- biochemistry, physics, geology, genomics, oceanography, etc.--- who are interested in achieving top performance from GPUs in diverse applications using Monte Carlo simulations can now obtain an updated version of NAG numeric routines for GPUs from the Numerical Algorithms Group (NAG).


General Purpose GPUs (graphical processing units) were originally used for 3D gaming acceleration on personal computers but have recently been at the forefront of numerical and scientific computation. Monte Carlo simulations are used in a wide array of technical computing applications in diverse areas such as finance, engineering simulations, drug discovery, scientific research, oil and gas exploration, and more.

Speaking for NVIDIA, a leader in GPU computing, Andrew Cresci, GM Vertical Marketing comments,

The ecosystem around GPU computing is growing rapidly and NAG’s additions to their routines for GPU computing could not be more timely. NAG’s numerical libraries are renowned for delivering top performance while maintaining the highest standards of accuracy. There are now some 60,000 active CUDA developers, and providing access to trusted algorithms from NAG is a major milestone that enhances the maturity of NVIDIA’s GPU computing architecture.

More . . .

On the new Top500 list, there are three GPU-accelerated supercomputers. Two of them use Nvidia’s new Tesla card C2050 with the Fermi chip while the third is equipped with the dual-GPU card AMD Radeon HD 4870 X2. It’s possible, with a little effort, to estimate the CPU's contribution to the results, which allows a rough comparison of the Tesla C2050 and the Radeon HD 4870 X2: around 140 to 160 teraflops per card with the highly optimised Linpack benchmark in double precision. That’s only about 30 per cent of the theoretical maximum performance.


More . . .

Nvidia: It's Time to Kill CUDA


Why are we talking about this today? Because Nvidia stands at a crossroads, with two closed, proprietary APIs that have mainstream potential: the general-purpose computing CUDA API, and the PhysX physics-acceleration API, which sits on top of CUDA. These are both promising technologies, but only owners of Nvidia hardware can harness their power. Meanwhile, there are two emerging open standards that mirror what Nvidia is doing with its proprietary development. One is OpenCL 1.0, and the other is a general-purpose GPU computing API, which Microsoft will include in DirectX 11.

There are a relatively small number of consumer applications that use CUDA, PhysX, or OpenCL right now, but the possible applications for the tech are endless—grossly simplified, these APIs let graphics chips perform CPU-like functions. The question Nvidia needs to be asking is simple: Will developers write their general-purpose GPU computing apps using a proprietary API that works on only a subset of PCs—those stuffed with Nvidia hardware—or will they use an open API that will work on every PC on the market?

Nvidia’s path is clear: It needs to stop trying to convince us that closed APIs are good, and instead embrace OpenCL and Microsoft’s yet-to-be-named solution. It needs to port PhysX to run on one of the open APIs, then use PhysX as a platform to advertise the kind of power that Nvidia delivers (with the recipients of all this messaging being ATI diehards and anyone considering the forthcoming Larrabee GPU from Intel).

By focusing on what its always done well—building kick-ass hardware—instead of force-feeding us closed APIs, Nvidia will thrive. As for CUDA? It’s served its purpose, but its time has passed. It’s time to kill CUDA.


More. . .

Nvidia Launches Fermi-based Quadro GPUs


Tuesday and Nvidia demonstrated at its booth its latest Quadro Fermi graphics cards. The cards are based on the same technology as the Nvidia GTX 400 series of consumer cards, but designed for workstation applications.

The Quadro 6000, 5000 and 4000 have 448, 352 and 256 processing cores respectively, with graphics memory sizes of 6 GB, 2.5 GB and 2 GB. While having fewer processing cores than the highest-end GTX 400 series consumer card, they also have up to 4x more graphics memory available. On the mobile front, the Quadro 5000M has 320 processing cores and 2 GB of graphics memory. For truly high-performance GPU-based computing, there is the Quadro Plex 7000 with 896 processing cores and 12 GB of graphics memory. All of the cards are compliant with OpenGL 4.0/4.1, DirectX 11, and shader model 5.0.

The Quadro 4000 and 5000 are available immediately, while the Quadro 6000 and Quadro Plex 7000 are expected to be available this fall. Mobile workstations from Dell and HP based on the Quadro 5000M are expected to be available in 3Q 2010.


More . . .

Gainward GeForce GTX 460 2GB “Golden Sample”


Today Gainward proudly introduce you its world first own designed Gainward GeForce GTX 460 2GB “Golden Sample”. Don’t be fooled by your old creed about 2GB myth! For the first time, these beefy 2GB frame buffers are to be filled quickly owing to the most powerful and efficient GF104 architecture in graphics cards history. Even better, Gainward never does things by half, these 2GB memory are to be overclocked! It wears the badge of “Golden Sample” represents its invincible performance of this untamed beast. It’s not only huge, it’s fast and it’s full of stamina “not merely for show”.

Gainward factory over-clocked GTX 460 2GB “Golden Sample” from 675MHz to 700MHz for all 336 CUDA Cores and 1,800MHz for 2 GB GDDR5 256 bit memory bank, equivalent bandwidth is up to 115.2 GB/sec, which is more than 33% boost from standard 192 bit GTX 460.


More . . .

ARM Fellow Jem Davies at Fusion Developer Summit


Jem Davies speaks to the developers gathered in Bellevue about the future of GPU computing, Moore's Law, and the power of low power.

Nvidia's Kepler to Arrive in 2012


Kepler will be succeeding the Fermi architecture and use a 28 nm production process. There have been rumors that Nvidia is struggling with the 28 nm production and that manufacturing is causing the delay. Of course, Nvidia isn't commenting on the delay and no chip manufacturer would ever give a statement on yield issues anyway. However, Nvidia is under pressure of getting Kepler out on time and especially get to Kepler right. (The current Fermi architecture was delayed several times and had a less than perfect launch.)

According to information released by Nvidia so far, Kepler cards will triple the dual-precision floating point performance of Fermi and hit up to 6 dp GFlops, while its successor Maxwell (scheduled for a 2013 release) is expected to with almost 16 dp GFlops. These are big promises and Nvidia wouldn't want to miss them.


More. . .

Noted technology researcher, Jon Peddie recently published an analysis of the GPU market, noting that the “discrete GPU is alive and thriving.” Using the famous quote from Mark Twain, "reports of my death are greatly exaggerated" he goes on to explain how misconceptions about market evolution are leading to some false ideas about the future of discrete GPUs.

He claims that “although the popular press and some analysts would like to believe otherwise, there’s a simple truth in the [GPU] business: You can’t get the same level of performance from integrated graphics as you can from a discrete GPU. It’s a matter of basic physics.”

"Discrete GPUs have, and will continue to hold, a multi-year lead over integrated graphics, and software applications and operating systems aren't standing still. Therefore, the notion that the evolution of integration spells the end of discrete GPUs just isn't founded."



More . . .