Nvidia Admits Showing Dummy Fermi Card at GTC, Claims First Graphics Cards on Track for Q4 2009

Nvidia: We Have Working GeForce Fermi Graphics Cards

by Anton Shilov
10/02/2009 | 01:08 PM

Nvidia Corp. earlier this week unveiled impressive details about its next-generation Fermi architecture for graphics processing units (GPUs) and even demonstrated an add-in card powered by the new chip. But while the new GPU architecture impresses, the board that was shown-off does not seem to be a working sample. Nvidia itself claims that the card was an “engineering sample” and that the first commercial Fermi products will be released in Q4 2009.


“[The Fermi chip was] not on the prototype that was held up [at the keynote and press conference]. Modifications may be made in this before the product is released. There was a working Fermi chip used in the Fermi demo,” explained Luciano Alibrandi, a spokesman for Nvidia.

Nvidia graphics card powered by Fermi-G300 graphics processor. Image by PC Watch

The card that was shown by Jen-Hsun Huang, chief executive officer of Nvidia, at the GPU Technology Conference did not seem to be the final product from the start, as we mentioned in the news-story on the matter. Besides, many press outlets as well as technology forums (1, 2) over the Internet expressed doubts that the demonstrated Fermi-G300 card was real.

Nvidia graphics card powered by Fermi-G300 graphics processor. Image by PC Watch

There are numerous questionable facts about the card that Nvidia demonstrated during its GPU trade-show:

While Nvidia admits that the board on the public display was an “engineering sample” and did not feature its new Fermi-G300 graphics chip (also known as GT300, NV60), the company stresses that it did show actual performance numbers that the new GPU can achieve without using simulation as well as showcased a demo that was run by its new graphics processor.

“The Fermi demo that we showed was driven by a Fermi-based Tesla chip. The performance numbers that were displayed, which showed five-times the performance of the current generation Tesla chip, are proof of that. […] We are actually showing working sample to some of the press attending GTC,” said Mr. Alibrandi.

But while Nvidia does not want to demonstrate its new Fermi-G300 graphics card openly in early October, it does state that it will be able to ship the product commercially in late 2009 and denies the claim that the first Fermi products will only show up in Q1 2010.

“The first Fermi GPUs are expected to launch by year’s end,” stressed Mr. Alibrandi.

Perhaps, Nvidia is somewhat too optimistic. At the GTC, Nvidia demonstrated the A1 revision of the Fermi-G300 graphics chip made in late August, whereas usually the company uses only A2 or even A3 revisions on commercial products. It usually takes months to create a new revision.

For example, ATI, graphics business unit of Advanced Micro Devices and the arch-rival of Nvidia, demonstrated its latest RV870 “Cypress” processor in action back in early June ’09, but was only able to release the chip commercially in late September ’09, about four months later.