| ||Thread Tools|
|31st October 2005, 20:00||#1|
Join Date: May 2002
NVIDIA Eyes Four-Way Multi-GPU Shortly
As hardcore gamers constantly want more and more power out of their hardware, designers of appropriate products are working to fulfill the demand. According to a web-site, NVIDIA Corp. has plans to enable four-way graphics solutions as early as in several month time. Currently, some makers of mainboards are already testing their products with four slots for graphics cards.
NVIDIA ForceWare Release 90 drivers are expected to allow up to four graphics processors (GPUs) to work in collaboration, The Inquirer web-site has reported. It is anticipated that the driver will be available sometime late this year or early in 2006. Theoretically, this may allow installation of four graphics cards or two graphics cards each of which utilize two GPUs.
While the cost of 4-way multi-GPU solution is unlikely to be lower than $1800 - $2000 or more only for graphics cards, provided that the drivers work fine, end users who agree to pay that amount of money are likely to get unbelievable performance. Even ATI Technologies, who first doubted about the success of multi-GPU technologies on the desktop market, recently said that 4-way multi-GPU was possible on the consumer market.
Gigabyte Technology, a leading maker of mainboards, has developed GA-8N-SLI-Quad-Royal mainboard that has four physical PCI Express x16 slots for graphics cards. The platform is designed for Intel Pentium 4 processors and incorporates a pretty unusual set of NVIDIA’s logic: nForce4 SLI x16 Intel Edition SPP as well as nForce4 SLI for AMD processors. The chips are connected using 16-/16-bit 800MHz HyperTransport link which provides 3.20GB/s bandwidth in total, whereas every PCI Express x16 slot provides 5GB/s bandwidth to hosted graphics cards. Theoretically, such mainboard may make four available PCI Express slots for graphics cards, allowing create four-way SLI configurations provided that there is either appropriate MIO connector, or NVIDIA manages to allow SLI without connection between all the graphics cards.
NVIDIA currently neither confirms, nor denies possibilities of four-way multi-GPU applications. “It looks like a great technology demo,” a spokesman for NVIDIA Corp. has said about Gigabyte’s GA-8N-SLI-Quad-Royal. The technology may become very interesting for professionals seeking for the highest performance possible.
|Thread||Thread Starter||Forum||Replies||Last Post|
|NVIDIA GF100 GPU Fermi Graphics Architecture||jmke||WebNews||0||18th January 2010 08:34|
|Driverheaven Nvidia Mobility Modder Released V18.104.22.168||jmke||WebNews||0||6th August 2009 14:12|
|NVIDIA GT216 first 40nm GPU from NVIDIA?||jmke||WebNews||1||20th October 2008 23:59|
|NVIDIA Fights Two Class Action Suits over GPU Failures||jmke||WebNews||0||15th September 2008 20:57|
|Futuremark: NVIDIA GPU PhysX not allowed||jmke||WebNews||0||21st July 2008 10:42|
|Nvidia Readies Mobile GeForce 8-Series GPU Shortly||jmke||WebNews||0||14th November 2006 12:39|
|ATI and Nvidia to battle for 90nm PCIe GPU market by 1Q 2006||jmke||WebNews||0||2nd December 2005 19:00|
|Nvidia narrows ATI’s lead in desktop GPU market share||jmke||WebNews||0||1st November 2005 17:33|
|NVIDIA Unleashes World's Fastest Mobile GPU||Sidney||WebNews||0||25th February 2005 01:45|
|NVIDIA Is the Only GPU Manufacturer||Sidney||WebNews||0||24th August 2004 06:44|