Geforce GTX 750 Ti GigaBite's attitude shows Performence For unseen electricity consumption
After two years of video cards in the process 28 nanometer, in terms of anode This is still not the time to switch to 20 nanometers. It would probably have to wait for the end of this year.
First of all, a bit of history. Two years ago, Anodia launched the Kepler architecture, with its first series of graphics cards GeForce GTX 600. These were GTX 680 and-GTX 670. graphic cards These improved the ratio between Performence For electricity consumption.
The power consumption was so improved that if a GTX 580 video card based on a production process of 40 nanometers received 100 percent Performence For 100 percent of power,GTX 680 We received 130 performance percentages for 70 percent of power. As a result, even simpler models enjoyed lower power consumption and higher performance.
Throughout the duration of the chip manufacturing process, it is common for a manufacturer of graphics cores to generate one series of similar efficiency data (the ratio of power consumption to performance). In other words, a more advanced product from that series will necessarily consume more electricity. So, core GK104 Who,GTX 680 Consumes far less power thanGK110 You'll find in780 GTX Ti, Based on the same 28 nanometer manufacturing process and contains more internal components than its little sister.
At the end of last year,TSMC Take the time, and the production of 20 nanometers will have to wait a few more months. Anodia Deviated from its custom and custom of companies like it and began the process of designing and manufacturing the next generation, in small pieces, already on the existing 28 nanometer.
The name of the new architecture is Maxwell (Maxwell). Anodia Designed the architecture with two main characteristics in mind - low power consumption and high parallel performance. After the prototype was ready and functioning properly, a process of mass production of a video card began in the entry market based on one complete Maxwell processing unit, which would be part of the processing power potential you can find in richer core components. This product is GeForce GTX 750 Ti.
In this new architecture, Anuidia decided that she would stick as many controllers and caches as possible to fewer processors. As a result, the processors receive higher internal traffic control than we have seen in the previous Kepler architecture. This can be compared to a class of students in which there is a teacher for each 20 student versus a teacher for each 30 student. There are 128 processor groups, each of which has five processors. That is, the total 640 internal processors in Maxwell's core which is also called GM107 (MT GeForce Maxwell 107).
On the video card GeForce GTX 750 Ti Lies the core of theGM107 Which includes 640 Processors With five of them in the group. Each set of 32 Processors Tight to visit זיכרון 32 bit width, so that the total memory controller bandwidth is 128 bit. The type of memory is GDDR5 Four-channel 2 gigabyte.
The version we received for review is the "Windforce OC של Giga Bite. It comes at a very high core frequency of 1215 MHz (compared to 1020 in default versions) and frequency זיכרון Standard of 1350 MHz. This results in a volume of traffic זיכרון graphic Of 86.4 gigabytes per second, less than the amount of graphics traffic seen in other modern video cards. For comparison, this is the amount of memory traffic you get on the video card GeForce GTX 8800.
GeForce GTX 750 Ti Launched at the price of 150 dollars as a solution parallel to theGTX 650 Ti Boost. In this critique you will see how such a tiny and modest processing unit copes with graphic cards Others in the market.