As NVIDIA’s last Quadro rollout took place over two-years-ago, we’ve been a little overdue on some new models. Well, those new models are here, and aside from the top-of-the-line K6000, the entire lineup is covered.
The Quadro K5200, seen below, might look almost identical to the model it replaces, the K5000, but it’s what’s under the hood that matters here. While the K5000 boasted single-precision floating-point performance of about 2.1 TFLOPs, the K5200 bumps that to almost 3.1. Likewise, the other models increase performance nicely over the previous generation (which is what we’d expect given it’s been two years), and contrary to what some might think, the K420 is not going to have the “highest” pricetag. Apologies – I’ve been trying my damnedest to weed out my poor jokes.
As the K6000 was released just last summer, and is also Kepler-based, NVIDIA is leaving it intact as the beefiest offering. The K5200, which I’d suspect will retail for around $2,000 (if K5000 pricing is anything to go by), offers double the GDDR5 of the K5000, and as mentioned above, a 40% performance boost. That boost does come with a higher TDP, however, of 150W, from 122W.
|
Cores |
Memory |
Mem. Bw |
Mem. Bus |
Perf. ** |
Power |
K6000 * |
2880 |
12GB |
288 GBps |
384-bit |
5196 GFLOPs |
225W |
K5200 |
2304 |
8GB |
192 GBps |
256-bit |
3074 GFLOPs |
150W |
K4200 |
1344 |
4GB |
173 GBps |
256-bit |
2072 GFLOPs |
122W |
K2200 |
640 |
4GB |
80 GBps |
128-bit |
N/A |
68W |
K620 |
384 |
2GB |
29 GBps |
128-bit |
N/A |
51W |
K420 |
192 |
1GB |
29 GBps |
128-bit |
N/A |
41W |
Both the K6000 and K5200 are based on the GK180GL core, while the K4200 is based on GK104GL. The K2200 is a bit interesting, because its GM107GL is based on Maxwell, an architecture NVIDIA debuted with the release of the GeForce GTX 750 Ti earlier this year.
For a full rundown of the ports that each model offers, and other smaller details, you can hit-up NVIDIA’s Quadro product page.
To give users an idea of what to expect from the latest Quadros, NVIDIA’s prepared a bullet-list:
- Interact with data sets or designs up to twice the size handled by previous generations.
- Remotely interact with graphics applications from a Quadro-based workstation from essentially any device, including PCs, Macs and tablets.
- Run major applications – such as Adobe® CC, Autodesk Design Suite and Dassault Systèmes SOLIDWORKS 2014 – on average 40 percent faster than with previous Quadro cards.
- Switch effortlessly from local GPU rendering to cloud-based offerings using NVIDIA Iray® rendering.
Admittedly, I had hoped to see a little more Maxwell infusion with this latest Quadro line-up, but given the target market, NVIDIA’s exceptionally careful about how it rolls-out new architectures. Overall, the new models are a big improvement over the last ones, so I’d chalk this up as a great launch.