Nvidia has pulled out the world's smallest chainsaw again. Last time we heard of this mythical tool, Nvidia split the GM204 inside the GeForce GTX 980 to produce the GM206 of GeForce GTX 960 fame. This time, the workstation-oriented GM204GL found inside the Quadro M4000 and M5000 is on the chopping block. When the silicon sawdust clears, we're left with the GM206GL and the shiny new Quadro M2000.
|Product||Memory interface||Memory||Memory bandwidth||CUDA cores||TDP|
|Quadro M4000||256-bit||8GB GDDR5||192GT/s||1664||150W|
|Quadro M2000||128-bit||4GB GDDR5||106GT/s||768||75W|
|Quadro K1200||128-bit||4GB GDDR5||80GT/s||512||45W|
AIDA64's patch notes effectively spilled the beans on the lineage of the GPU inside the newest Quadro a little over a week ago. The Maxwell GM206GL GPU in the M2000 has 768 CUDA cores, roughly half of the M4000's GM204GL GPU with 1664. The M2000 ships with 4GB of GDDR5, half of the M4000's 8GB of GDDR5. About the only thing Nvidia didn't cut on the M2000 is the display outputs —both the M4000 and the M2000 have 4 standard DisplayPort 1.2 outputs on board.
While these specs are a big step down from the M4000, they should still put it well ahead of the next-cheapest Quadro, the K1200. The 75W Quadro M2000 also doesn't require a six-pin PCIe power input, which could make it attractive to buyers who need more performance than the K1200 but don't have a power supply compatible with the M4000. Nvidia didn't say how much the M2000 would sell for. A listing for a PNY Quadro M2000 is on Newegg now, but at the moment the listing only shows an April 21 release date.