NVIDIA Quadro DDR

The release of NVIDIA’s Quadro and GeForce based cards will impact professional users that depend on powerful graphics accelerators for their technical applications.  NVIDIA’s Quadro and GeForce GPU (Graphics Processor Unit) are capable of 50 billion floating point operations per second (50GFLOPS) which is considerably higher than most mainstream professional cards with on-board transforming and lighting engines that boast performance of around 5 billion floating point operations per second.  The Quadro DDR is the top of NVIDIA food chain; it is their top gun and most expensive NVIDIA based card.

NVIDIA's Quadro GPU ( NV10GL)  is set apart from the consumer grade GeForce (aside from it’s price tag) by enabling certain features on the Quadro that were left disabled on the GeForce since the latter was not intended to be a high end solution.  Among these features is enhanced support for anti-aliased points and lines, which although isn’t a commonly used feature with most users, was an extremely poor performance point for the GeForce.  While these settings might conceptually be “enabled” on the GeForce through driver tweaks/registry hacks, it is unclear exactly what methods NVIDIA went to in order to make sure that they were disabled on the GeForce, to this date no one has discovered the magic hack. 

On paper, NVIDIA’s Quadro very closely resembles the GeForce, with a few exceptions.  The core clock of the Quadro is 135MHz, a 15MHz increase from the 120MHz core clock of the GeForce.  The higher core clock allows the Quadro to achieve a 13% higher fill rate of 540 Mpixels/s.   The Quadro also features a peak triangle rate of 17 million triangles per second up from the GeForce’s 10 – 15 million triangles per second. The Quadro designs ( SDR and DDR ) typically are configured with 64 MB of memory to further differentiate the Quadro cards from the consumer grade GeForce cards which has 32 MB of memory.

The initial releases of the GeForce and Quadro card designs used 128 bit SDR ( single data rate - 1 bit of data per tick per memory line ) SDRAM memory system running at 166 MHz providing 2.656 GB/s of memory bandwidth.  The memory bandwidth provided by this design was not adequate when the cards were used in certain render intensive applications.   NVIDIA's solution was to provide support for DDR ( double data rate - 2 bits of data per tick per memory line ) SDRAM running at 150 MHz, this provides 4.8 GB/s of memory bandwidth (an increase of 80.7%). 

The addition of the DDR configuration to the Quadro design mix further enhances NVIDIA's ability to compete with the established players, 3Dlabs, Intergraph, and E&S, in the professional graphics card market. When this reference Quadro DDR design reaches the marketplace soon it will likely be in the form of a ELSA GLoria II DDR graphics card, since ELSA has been selected by NVIDIA to be exclusive vendor for Quadro based professional graphics cards. 

The cost for the DDR version of the Quadro will be more than for the SDR version, which costs about $650. US,  and given the  current cost differential between the SDR and DDR versions of the GeForce ( ~ $90. US ) and allowing for twice the memory,  a realistic cost projection for the Quadro DDR would be about $830. US. Is the added cost for the DDR version worth the projected cost difference?

Another problem NVIDIA has is timing: the production version of the Quadro DDR card is not yet on the market; at the same time, the consumer cards using the next generation NV15 GPU's are due to arrive on the scene this spring. The window of opportunity for cards using the NV10GL chips ( the Quadro DDR and SDR ) is not very large.