ATI Radeon HD 4770: ATI's First 40nm GPU

Graphics Cards
Viewing page 1 of 8 pages. 1 2 3 4 5 6 7 8 Next
ATI Radeon HD 4770: ATI's First 40nm GPU

Apr. 28, 2009 by Lawrence Lee

ATI Radeon HD 4770 512MB
PCI-E Graphics Card

As the underdog in the chip wars, AMD has traditionally targeted users with products that offer good relative value, rather than maximum performance. A prime current example is their fastest desktop processor, the Phenom II. Their current flagship CPU, the Phenom II 955 Black Edition falls a bit short of Intel's fastest and and more costly Core 2 or Core i7 chips, but it was well received by the most of the tech web press due to AMD's aggressive pricing. Their graphics division, ATI, has been following a similar path, offering great value cards like the energy efficient Radeon HD 4670 and HD 4830 for gamers with tighter pockets. In these troubled financial times, it's could well turn out to be a very effective strategy.

Currently, consumers looking for a budget gaming card for about $100 have ATI's Radeon HD 4830 and nVidia's GeForce 9800 GT to choose between. They are neck-in-neck in terms of performance, but as of date, 9800 GT's are retailing for about $10 more. To hammer this position further, ATI has released its first GPU built with a 40 nanometer manufacturing process, the Radeon HD 4770. ATI says it is a better performing, more energy efficient replacement for the HD 4830.

The Radeon HD 4770.

ATI rated the HD 4830 as a 110W card and employed a single-slot cooler. Strangely, the new HD 4770, despite having a 80W thermal envelope, has a dual-slot reference heatsink. Both cards require a 6-pin PCI-E power cable.

Tale Of The Tape: HD 4770 vs. HD 4830
Model HD 4770 HD 4830
GPU Core RV740 RV770LE
Manufacturing Process 40nm 55nm
Transistor Count 826 million 956 million
Stream Processors 640 640
Core Clock 750 MHz 575 MHz
Memory Clock 800 MHz 900 MHz
Memory 512MB GDDR5 512MB GDDR3
Memory Bandwidth 128-bit
51.2 GB/s
57.6 GB/s
Average board power 80W 110W

Compared to the HD 4830, the 4770 has an equal number of shaders units, a higher clock speed, but smaller transistor count, slower memory and less memory bandwidth. The die-shrink from 55nm to 40nm is no doubt the reason for lower power requirement. With the smaller die, it's surely cheaper to build.

Technical specifications according to GPU-Z.

1 2 3 4 5 6 7 8 Next

Graphics Cards - Article Index
Help support this site, buy from one of our affiliate retailers!