Nividia’s Maxwell Architecture after so many rumors and news now finally revealed.
Maxwell is divided into two Version, One is GM107-400 which will be the Gtx 750Ti, and another is GM107-300 the thin client will equip with the 768 Stream Processors, 16 Rops and 64 texture units, GM107-300 will be used in the Gtx 750.
Additionally the Die Size of Gm107 is 156mm2 which is around 32% larger than Gk107 and around 30% smaller than Gk106.
If we see the earlier rumors there are news that the new Maxwell will come on with the 20nm technology, but still sadly we will not se this technology from Nvidia at first, but likely we will be able to see some 20nm chips on the Q3 of this year. The Reason for naming the head of two products included GeForce 700, instead of using the new GeForce 800, is waiting for the new process (and yes the GeForce 800M series notebook will not mention).
Kepler VS Maxwell
GK107-450 | GK106-400 | GK107-300 | GM107-400 | |
GPU | 28nm | 28nm | 28nm | 28nm |
Die Size | 118mm2 | 221mm2 | 156mm2 | 156mm2 |
CUDAs | 384 | 960 | 768 | 960 |
TMUs | 32 | 80 | 64 | 80 |
ROPs | 16 | 24 | 16 | 16 |
Memory Bus | 128-Bit | 192-Bit | 128-Bit | 128-Bit |
TDP | 65W | 140W | ??? | ~75W |
There are no new technical features included, but deep optimization and improvements are there, or in other words, we can say that the Maxwell is Kepler Refresh.