Nvidia Corp. has quietly introduced two new professional graphics cards. The new Quadro M4000 and Quadro M5000 adapters are powered by the highly-successful “Maxwell” architecture. The novelties promise to deliver strong performance in graphics applications, but when it comes to computing, the new GPUs will be slower than some of their predecessors because of architectural peculiarities.
The new Quadro M4000 and M5000 graphics cards are powered by the GM204 graphics processing units with 1664 and 2048 stream processors, respectively. Both graphics adapters feature 256-bit memory bus and carry 8GB of GDDR5 memory. The more advanced Quadro M5000 features ECC memory and dual-slot cooling system. The more energy-efficient Quadro M4000 sports single-slot cooler.
The new Quadro graphics cards fully support all modern technologies and application programming interfaces, including OpenGL 4.5, OpenCL 1.2, DirectX 12, 3D Vision Pro, Quadro Sync, SLI and so on.
Since Nvidia’s “Maxwell” architecture does not include FP64 arithmetic logic units (ALUs), performance of GPUs based on Nvidia’s latest technology in applications that require double precision computing is very low. Typically, professional graphics applications do not need FP64, but scientific applications used by creative professionals usually do. As a result, for many people previous-generation “Kepler” based Quadro graphics adapters may be a better choice.
Discuss on our Facebook page, HERE.
KitGuru Says: It is rather surprising that it took Nvidia almost a year to develop two professional graphics cards based on the GM204 graphics processing unit.