NVIDIA Quadro FX 1700

Last, but certainly not least, in terms of Quadros being announced today is the FX 1700.

You could play spot the difference again between the FX 1700 and FX 370 and FX 570; the PCB is identical for all three. The differences lie in terms of the output configuration -- the FX 1700 adds an analogue HD output -- and the board memory, where the FX 1700 has double that of the other boards at 512MiB. That represents a doubling of the memory available on the Quadros it replaces, and that includes Quadro FX 3500. Indeed NVIDIA claim FX 1700 performance in vertex-heavy situations is higher than an $800 FX 3500, and at 42W maximum board power, it does so with less board power than the 3500 and no additional power connector.

The single slot cooler is the same as on the FX 370 and FX 570 (as is the PCB), with the fan pushing air over the sink towards the backplane.

NVIDIA pitch the Quadro FX 1700 at $699, undercutting current prices for Quadro FX 3500, and if the rest of the board's performance is similar to the vertex-heavy performance NVIDIA are quoting, it should easily displace the FX 3500 as the Quadro of choice at the sub $1000 price point.

The G84-powered Quadro FX 1700 ships in October, after being announced today. It's a full CUDA target like the rest of the Quadro's talked about today, and supports the common features we talked about with the FX 570.

We'll test its performance when we get a sample, and whispers have Apple offering the Quadro FX 1700 as an option for the Mac Pro sooner rather than later. None of the Quadros announced today support SLI, which is a little surprising but not unexpected, but they all support the orthogonal filtering ability we've discussed at length in our architecture analysis of G8-series hardware, 8Kx8K textures and other improvements brought by G8-series silicon, that G7-series doesn't provide.