News
 

Bookmark and Share

(4) 

Nvidia Corp. reportedly plans to release a new dual-chip flagship graphics card next month in a bid to compete against AMD Radeon HD 6990, the forthcoming flagship from the arch-rival. The novelty will, apparently, be based on the top-of-the-range code-named GF110 graphics processing unit (GPU).

Little is known about the graphics card that is said to be called GeForce GTX 590. The board is presumably set to be based on two GF110 graphics chips and will therefore have 1024 stream processing units, 128 texture units and 96 raster operation units in total. In order to simplify the design of the board, the GF110 chips will rely on two 384-bit memory buses to connect to 3072MB of GDDR5 memory (each chip will have 1536MB frame-buffer).

Several web-sites have already published images of the yet unreleased dual-GPU graphics boards powered by Nvidia's graphics processors, but it is not completely clear whether the solutions are actually set to be released by all Nvidia's partners or just by some of them. It is also uncertain whether the cards will be launched at all; back in the past several makers demonstrated dual-chip prototypes, but did not release them commercially afterwards.

Based on images published on NordicHardware and TechConnect web-sites, the dual-GPU GeForce GTX 590 will have two 8-pin PCI Express power connectors, thus, will consume between 300W and 375W.

Given the fact that Nvidia GeForce GTX 580 graphics card is significantly faster ATI Radeon HD 6870, the dual-chip GeForce GTX 590 will have all chances to outperform the Radeon HD 6990 code-named Antilles. Although the flagship graphics cards do not  represent a substantial share of AMD's or Nvidia's revenues, they attract attention to the "GeForce" and "Radeon" brands and therefore advertize mainstream products and prove technological leadership of their developers.

Nvidia did not comment on the news-story.

Tags: Nvidia, Geforce, Fermi, gf110, antilles, Radeon

Discussion

Comments currently: 4
Discussion started: 02/01/11 06:29:44 AM
Latest comment: 02/11/11 10:04:58 PM
Expand all threads | Collapse all threads

[1-3]

1. 
Ok...

1. Please realize the 6970 is not the 6870. Yes, it's confusing, but it makes the articles look bad.

2. This card has 10 chips per GPU. Let's do some math. Does 10 chips equal 384-bit? No it does not. It equals 320-bit.

Sooo....This product likely uses chips in a 512/40 ROP/320-bit configuration...Which is actually an optimal design, believe it or not. Doing a little bit of guesstimating, it's possible a design like this could fit under 300W under realistic load at around 650mhz or so. nVIDIA is not going to sanction a card above 300W. It just ain't gonna happen. It might happen from partners, but not from nVIDIA.

IMHO, AMD has a more-efficient arch, but the question comes down to clockspeeds. Obviously more units at lower clockspeed is going to prevail in this situation, as it's a power saver and there is a set limit. To beat what nVIDIA may put out, AMD may run out of TDP because their arch generally runs above the best mix for power/performance. If they could enable more units and run at a lower clockspeed (and perhaps voltage), that could be the difference. Either way, I don't expect them to be miles apart.
0 0 [Posted by: turtle  | Date: 02/01/11 06:29:44 AM]
Reply
- collapse thread

 
It uses TWELWE memory chips (2x6chips) look again.
0 0 [Posted by: OmegaHuman  | Date: 02/11/11 09:59:35 PM]
Reply

2. 
It was expected...
0 0 [Posted by: East17  | Date: 02/02/11 11:53:23 PM]
Reply

3. 
It uses TWELWE memory chips (2x6chips) look again.

And 2x192-bit doesn't make 384-bit memory architecture, but just that 2x192b ... and it probably aint GF110 (512SP/64ROPs) but G114 (rebranded full functioning G104 chips) which have only 384SIMD shaders (and cutdown, salvaged predecessor GF104 had only 336SPs). And because 192-bit GF104 chips are bandwidth limited because of direct ROP-MC connection these cards will more than probably have cut down shader units along with only 48 functioning ROPs instead 64 (3/4==75% of 256bit and 64 ROPs). So these are in fact salvaged GF114 chips so that nVidia could compete with salvaged RV970 aka. Caymans in HD6990 which should be released in Feb2011 and sport two of that power hungry Caymans on same pcb.

So it might be power effective GF104 as 192bit versions were usually below 150W while 256bit versions easily reached 180W (Core@700MHz). Somewhat redesigned GF114 will reduce power draw and probably enable some 2x750Mhz chips to draw less than 300W (TDP 260W and we all know how nVidia reduces TDP listing)
0 0 [Posted by: OmegaHuman  | Date: 02/11/11 10:04:58 PM]
Reply

[1-3]

Add your Comment




Related news

Latest News

Tuesday, July 15, 2014

6:11 am | Apple Teams Up with IBM to Make iPhone and iPad Ultimate Tools for Businesses and Enterprises. IBM to Sell Business-Optimized iPhone and iPad Devices

Monday, July 14, 2014

6:01 am | IBM to Invest $3 Billion In Research of Next-Gen Chips, Process Technologies. IBM to Fund Development of 7nm and Below Process Technologies, Help to Create Post-Silicon Future

5:58 am | Intel Postpones Launch of High-End “Broadwell-K” Processors to July – September, 2015. High-End Core i “Broadwell” Processors Scheduled to Arrive in Q3 2015

5:50 am | Intel Delays Introduction of Core M “Broadwell” Processors Further. Low-Power Broadwell Chips Due in Late 2014

Wednesday, July 9, 2014

4:04 pm | Intel Readies New Quark “Dublin Bay” Microprocessors. Intel’s “Dublin Bay” Chips Due in 2015