Tuesday, August 12th 2014

NVIDIA Announces the Entry-level GeForce GT 720

NVIDIA announced its entry-level graphics card for this generation, the GeForce GT 720. Based on the 28 nm GK208 silicon, the chip features 192 CUDA cores, and a 64-bit wide DDR3 memory interface, holding 1 GB of memory. It features core clock speeds of 797 MHz, and memory clock speeds of 1600 MHz DDR. Most cards based on this chip are expected to be single-slot, half-height, and passively cooled. NVIDIA is capturing the sub-$50 market with this chip.
Add your own comment

9 Comments on NVIDIA Announces the Entry-level GeForce GT 720

#1
micropage7
so newer stuff from nvidia now is entry level
Posted on Reply
#2
RCoon
Gaming Moderator
At last, real, actual, new chips for the low end, not just rebadged GT 2xx's!
Nevermind, GK208 is used in the 640 chips...
Posted on Reply
#3
zsolt_93
It is still GK, GK208 was released long ago for notebook use now they made a desktop card out of it.
Posted on Reply
#4
Jorge
Nvidia needs to sell all the crap they can because they are in a bind going forward as we shall publicly see soon.
Posted on Reply
#5
1c3d0g
by: Jorge
Nvidia needs to sell all the crap they can because they are in a bind going forward as we shall publicly see soon.
Source?
Posted on Reply
#6
Prima.Vera
How does this compare to the integrated GPU on AMD/Intel's processors?
Posted on Reply
#7
john_
At Nvidia's product page it also mentions the option for GDDR5. And what is also supported and their is a little irony in that, is Adaptive VSync.
http://www.geforce.com/hardware/desktop-gpus/geforce-gt-720/specifications

by: Prima.Vera
How does this compare to the integrated GPU on AMD/Intel's processors?
With DDR3 RAM it only has 14.4MB/sec bandwidth, the same that GT210 DDR3, or GT620 have. In other words it's slow. Really slow, because it doesn't have the bandwidth necessary to run anything. Compared to iGPUs? Probably slower than an iGPU on a quad core AMD, or compared to Iris Pro. But that's just a guess. With 64bit data bus and DDR3 it's useless even with 19200 CUDA cores, not 192. Now that I mention it, it could be a good cheap option for CUDA or PhysX.
Posted on Reply
#9
1c3d0g
by: Fluffmeister
Jorges bottom hole.
I thought so, too! :shadedshu:
Posted on Reply