Nvidia overhauls Quadro workstation graphics line-up with Kepler GPU
Workstation graphics line gets Kepler GPU from top to bottom, boasting greater performance
Nvidia has introduced new Quadro graphics cards for workstations, extending its next-generation Kepler GPU across its entire mainstream line.
Available from Tuesday, through distributor PNY and in workstation systems from HP, Dell, Lenovo and others, the new Kepler-based cards offer a two times increase in performance with more on-board memory and double the number of supported displays compared with earlier products.
"We've been shipping the Quadro K5000 since last August, now we're adding more cards targeted at the volume part of the market, making them the workhorse cards in our workstation line-up," said Sandeep Gupta, senior director of Nvidia's Professional Solutions Group.
Nvidia's new products comprise the entry-level Quadro K600, the mid-range Quadro K2000 and Quadro K2000D, with the K4000 the high-end card, boasting a performance increase of up to 30 percent over previous Nvidia generations.
The two mid-range cards are identical except for their display options; the K2000 has twin DisplayPort outputs and a DVI connector, while the K2000D has a mini DisplayPort and twin DVI connectors.
In comparison, the K600 has a single DisplayPort and DVI, while the K4000 has twin DisplayPort outputs and a single DVI connector.
All of these products can fit into a single PCI Express card slot, Nvidia said, unlike the existing K5000 which takes up the space of two slots inside a PC chassis.
Nvidia's new line-up offers customers more features and greater performance at essentially the same price point, according to Gupta, with 4K display capability, support for up to four HD displays from a single board, and a dedicated H.264 video encoder in hardware that frees up the GPU cores to do other tasks instead.
The products scale from 192 GPU cores and 1GB of memory in the K600, through 384 cores and 2GB in the K2000, up to 768 cores and 3GB memory in the K4000 model.
Nvidia quotes the K600 as being capable of 336 gigaFlops, scaling to 1.246 teraFlops for the K4000, with benchmark scores under the SPECviewperf 11 OpenGL benchmark of 26 for the K600, rising to 53 for the K4000.
However, Gupta said professional users were looking for industrial-grade reliability as much as performance, and that Nvidia has long emphasised this in its Quadro products
"With this generation we've continued to put in a lot of energy in this regard," he said, adding that Nvidia also has special tools for enterprise-wide deployment.
Nvidia also said that workstation trends were driving the demand for even greater performance, as engineers and graphics professionals are starting to use suites of tools rather than just a single application.
"An engineer might want to use SolidWorks for design, but then he might also want to do some simulation work to test the structural integrity of the design before he hands it off to the CAE team using COSMOS or Abaqus," said Gupta.
Fortunately, extra power can be added through Nvidia's Maximus technology, which pairs its Tesla general-purpose GPU with a Quadro card inside a single workstation. The new Kepler-based Quadro cards all support this capability.
Pricing for the new Quadro graphics cards starts at £149 for the Quadro K600, with the Quadro K2000 and K2000D at £389 and the Quadro K4000 costing £829.
The existing Quadro K5000 and Quadro K5000 for Mac products are priced at £1,559.