Workstation GPUs are not our main focus at Phoronix, but with the increasing use of Linux on workstation systems, we will be starting to look at professional graphics products this month and likely more of them in the future. We are beginning this expedition by looking at the Quadro FX1700, which is one of NVIDIA's mid-range workstation graphics cards. This Quadro graphics card boasts 512MB of video memory, support for CUDA, and OpenGL 2.1 support. According to NVIDIA's product literature, the Quadro FX1700 is engineered to deliver exceptional performance, quality, and price for professionals.

Features:

- 512MB GDDR2 Video Memory
- 128-bit Memory Interface
- 12.8GB/s Memory Bandwidth
- PCI Express x16 Interface
- CUDA (Compute Unified Device Architecture) Support
- PureVideo / PureVideo HD Technology
- Dual Dual-Link DVI Connectors
- 42W Maximum Power Consumption
- DirectX 10.0 / OpenGL 2.1

Contents:

PNY Technologies is NVIDIA's exclusive AIB partner for the Quadro series, and this FX1700 comes with a part number of PNY VCQFX1700-PCIE-PB. The graphics on the outside of the product packaging are rather austere, considering this is a workstation product that needs not any goblins or females lavishing the product box. Inside the box were the PCI-E Quadro graphics card, a Windows driver CD, quick start guide, two DVI to VGA dongles, and a component output video adapter. All of these components were bundled together inside a plastic structure that was shrink-wrapped.

Examination:

The other Quadro FX models accompany the FX1700 as a mid-range professional graphics solution is the FX1500 and FX3450. Total there are currently 16 parts amongst NVIDIA's entry, mid-range, high-end, ultra high-end, and specialty Quadro graphics cards. Both the Quadro FX1500 and FX3450 only have 256MB of video RAM, compared to the 512MB found on this graphics card. While the vRAM is halved, their memory interface is doubled to 256-bit, which provides much more memory bandwidth. The FX1700 is the only Quadro mid-range component right now that supports OpenGL 2.1 (and DirectX 10). The NVIDIA GPU core used by the Quadro FX1700 is the G84GL, which is derived from the G84 that is used on the consumer GeForce 8600GT and 8600GTS graphics cards.

The heatsink found on the NVIDIA Quadro FX1700 is rather small and only dissipates heat from the GPU die, but is sized appropriately seeing as this PCI-E graphics card has a maximum power consumption of 42 Watts. During operation, this single-slot solution was near silent and we had run into no thermal issues. The 512MB of 128-bit GDDR2 video memory is made up of Hynix HY5PS121621C ICs. No external PCI Express power source is required for the FX1700.

On the opposite side of the PCB are just four screw heads, which secure the FX1700 cooler in place. Neither the FX1500 nor FX1700 support SLI (Scalable Link Interface).

At the rear of the graphics card are the two dual-link DVI ports as well as a TV output port, which can connect to the PNY component output adapter. This PNY adapter provides three component outputs, one S-Video output, and one composite output.

NVIDIA's latest binary display drivers under Linux fully support the NVIDIA Quadro FX1700. We happen to be using the most recent NVIDIA 169.12 driver.

During our testing, the NVIDIA G84GL core had operated comfortably between 50° ~ 60°C. The Quadro FX1700 does have two PowerMizer states.

The Linux benchmarks of the NVIDIA Quadro FX1700 aren't available yet, but will be published later this week. This article is just to serve as a preview of this mid-range product and to inform our readers of the forthcoming workstation graphics articles. For the Quadro FX1700 benchmarks, we will not only be benchmarking this card under Ubuntu Linux, but we will be directly comparing its performance to Microsoft Windows and Solaris Express. In addition, we will also be looking at an ATI FireGL component in just a few days. Stay tuned!