NVIDIA has just announced its entry-level or cheapest card offering in the 700 series, the GeForce GT 740. This card is a simple rebrand of the GT 640 so that means it is not based on the Maxwell micro-architecture, but rather uses the Kepler-based GK107 chip.
The card will feature 384 CUDA cores, 32 TMUs, and 16 ROPs. The 128-bit wide memory interface will hold 1 GB, but there will be 2 GB variants available. The card has a very low TDP of only 64W so not external power is required, although some variants with overclocks and 2 GB of memory will have a 6pin PCI-Express power connection.
Compared to its predecessor (GeForce GT 640) the GT 740 will have a higher GPU clock of 933 MHz (compared to 902 MHz on the GT 640). There is no trace of GPU Boost technology, that is what differentiates mid-range GTX cards from entry-level GT cards.
NVIDIA is looking to take on the sub-$100 market with this card and to compete against integrated solutions like AMD’s A-Series APUs and Intel Iris Pro 5000 series.
Many different AIB partners have released their versions of the GeForce GT 740 today, we have pictured them below.