Articles: Graphics
 

Bookmark and Share

(46) 

Table of Contents

Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 ]

NVIDIA’s GeForce 7800 GTX was introduced more than three months before the RADEON X1800 XT 512MB graphics card made it to the hands of the first reviewers and it will have more than four months lead over the top-of-the-range part from ATI Technologies when the model X1800 XT becomes available in early November. But while ATI is only about to deliver its leading-edge product, some graphics cards makers seem to offer accelerators that match and even surpass the top part of ATI in a number of benchmarks. Perhaps, there is no point to wait?

Having run into loads of issues with low yields, low performance, extreme clock-speeds and high power consumption with the GeForce FX product lineup, NVIDIA Corp. decided to create processors with higher efficiency and, perhaps, greater amount of logic, but moderate clock-speeds. After the GeForce FX 5900-series that ran at up to 475MHz or higher, NVIDIA released its GeForce 6800-series, which clock-speed range was from 325MHz to 425MHz, but which delivered performance much higher than that of the predecessors. This cost the company a huge amount of transistors: the GeForce 6800 Ultra (NV40) chip consisted of 222 million of gates, whereas the predecessor featured only about 125 million. Basically speaking, going up from 8 to 16 pixel processors forced NVIDIA to increase the number of transistors by nearly 78%. By contrast, the same kind of transition cost ATI about 45% of extra gates: the RADEON 9800-series weigh in 110 million of transistors, whereas the RADEON X800-series consists of 160 million of gates.

But NVIDIA’s difficulties were paid in fully: its engineers learned all the peculiarities concerning large low-speed chips and how to make them efficient. When the company started to develop the 7th generation of the GeForce products (G70), it already knew that the chip would consume tremendous amount of logic, would run at moderate clock-speed, would be made using proven process technology and would be able to deliver excellent performance. The result was fascinating and the GeForce 7800 GTX surpassed all the expectations: it was large in terms of die size and transistor count, it was showing very good benchmark results, it ran at frequency in-line with the previous generation, but it was consuming virtually the same amount of power, did not require large cooling system, did not draw excessive amount of heat… and had excellent abilities for further overclocking!

What NVIDIA Corp.’s engineers have achieved with the G70 was higher amount of pixel and vertex processors compared to the NV40, some extra features, and better ability to scale in terms of clock-speeds. While the company itself decided not to clock the reference GeForce 7800 GTX higher than 430MHz, it allowed its graphics card partners to pre-overclock their boards and sell higher-performance accelerators. And it now looks that the strategy proved to be a success, several makers of expensive computer components already sell the GeForce 7800 GTX with speeds much beyond the official specification. One of such manufacturers is ASUS with its Extreme N7800GTX TOP/2DHTV that ups core frequency from 430MHz to 486MHz and memory speed from 1200MHz to 1350MHz. So, let’s have a look at the beast!

 
Pages: [ 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 ]

Discussion

Comments currently: 46
Discussion started: 10/24/05 03:07:33 PM
Latest comment: 08/25/06 12:24:12 PM

View comments

Add your Comment