NVIDIA has updated its Quadro family with three new models based on the Fermi architecture. The flagship is called Quadro 6000 and ships with 448 CUDA cors and 6GB memory buffer.
NVIDIA focuses extra on Quadro 6000, but also Quadro 5000 where the latter sports 352 CUDA cores with 2.5GB memory. The family is completed by the entry model Quadro 4000 and QuadroPlex 7000, a system with dual Quadro 6000 graphics cards for a total of 896 CUDA cores and 12GB GDDR5 memory.
|Specifications||Quadro 4000||Quadro 5000||Quadro 6000||QuadroPlex 7000|
|CUDA-kärnor||256 st.||352 st.||448 st.||896 st.|
|Memory (GDDR5)||2,0 GB||2,5 GB||6,0 GB||12 GB|
||142 watt||152 watt||225 watt||System|
|MSRP||1 199 USD||2 249 USD||4 999 USD||14 500 USD|
Quadro 6000 builds on the same GF100 GPU used by GeForce GTX 470 and GTX 480 and with the same number of CUDA cores as GTX 470 it is the enormous memory buffer and the ten times higher prices that keeps the two cards apart.
The first reviews of NVIDIA’s new Quadro cards indicates that NVIDIA is having no problems with keeping AMD’s fastest FirePro card behind, even if the functionality and not the least prices are quite different.
With the Fermi architecture NVIDIA set out for performance and just like the desktop equivalencies the new Quadro families shows this.
Reviews of NVIDIA Quadro 6000/5000;