Last week, I did a short piece about the way how Nvidia is trying to bridge the 32-/64-bit divide, and today, the company officially unveiled Quadro FX 4800 and FX 5800.
Quadro FX 4800 shares a lot of similarities with Adobe-oriented CX, but features 216 shaders (yes, GTX-260 brother here, if my sources were correct) and 1.5 GB of GDDR3 memory. But the star of the today’s launch is FX 5800, the new flagship of Quadro fleet.
In a way, we already know everything about FX 5800, since Nvidia demonstrated the product back in August at Siggraph 2008, followed by Nvision 08 – so, specs come as of no surprise to anybody.
FX5800 features GT200 chip with all 240 shader processors (with 30 “invisible” FP64-capable Double Precision register units – 1 per group of 8 shaders), coupled with massive 4GB of GDDR3 memory clocked at 816 MHz in DDR mode (1.63 GT/s). As it usually happens with Quadro products, Nvidia did not reveal the clocks, but fill-rate of 52 billion texels and 300 milion triangles/sec can lead to calculator, and calculator leads us to believe that for the first time in ages, Quadro product is actually clocked higher than the reference GeForce card it was built upon. As far as I know, GeForce GTX 280 comes with fill-rate of 48.2 Gtexels, while this Quadro card features 4 GTexel/s more, or roughly 40 MHz higher clockspeed. Of course, I could be wrong…
The card supports Quadro G-Sync II, but I am much more interested in the fact that you can just take the old Quadro card out, put the FX5800 in and hook it up to an SDI daughterboard, which is something that broadcast industry will appreciate. The price of the board is $3400, and seeing how the prices raged in the previous era, I have to say that having competition is really, really good. ATi Radeon forced Nvidia to reshuffle its GeForce pricing segment and now FirePro did the same with Quadro FX series.
Thanks heavens, we (finally) have a war in workstation world. Now it is time to see RV770 chip with 4GB of memory as well… even though, ATI could technically came up with 4GB board as well, if it would use 4850X2 as a base for it. There is no way in hell we could see 4GB GDDR5 boards on the market, the higher density chips are not available atm.
P.S. with Nvidia continuing its funny usage of “FX” moniker, isn’t it funny that currently, the most powerful 3D card on face of the planet shares the same name as the most powerful dustbuster card that ever existed… yes, the mighty loud GeForce FX 5800. Yes, please do click on a link, some Monday laugh guaranteed. 😉
P.P.S. Where’s the OpenGL 3.0 support guys?