CHIP DESIGNER Nvidia has updated its Quadro range of workstation graphics cards to use its Kepler architecture.
Nvidia has rolled out its Kepler architecture to most of its consumer desktop and laptop Geforce branded chips and its Tesla accelerators but its Quadro workstation cards have largely been left with GPUs from the Fermi era. Now the firm has updated the main part of its Quadro range, with the Quadro K600, Quadro K2000 and Quadro K4000 all sporting Kepler GPUs.
Although Nvidia's high-end Quadro K5000 was updated with a Kepler chip a while back, the majority of the firm's workstation business is in the more affordable $600 Quadro K2000 and $1,270 Quadro K4000.
The firm has given the two cards 384 and 768 CUDA cores, respectively, with the Quadro K2000 having 2GB of memory and the Quadro K4000 having 3GB with memory bandwidths of 64GB/s and 134GB/s, respectively, though both cards lack ECC memory.
Nvidia's low-end Quadro K600, priced at $200, offers 192 CUDA cores and 1GB memory with memory bandwidth set down to 29GB/s. All of Nvidia's updated Quadro cards sport active cooling despite the Quadro K4000 drawing a very respectable 80W, while the Quadro K2000 and Quadro K600 draw 51W and 41W, respectively.
Nvidia quoted single precision floating point performance for all of its Quadro cards, saying that the chips are optimised for single precision performance, as workstation workloads rarely use double precision floating point operations. It should be noted that Nvidia's Geforce Titan lacks ISV certification but provides significantly higher single precision and double precision computing performance.
Nvidia said Kepler Quadro graphics cards will be available from today from PNY, its sole workstation graphics card partner in Europe, with Dell, HP, Lenovo and Supermicro selling systems with its Kepler Quadro cards. µ
Sign up for INQbot – a weekly roundup of the best from the INQ