www.my10sen.com

ATI’s new 256MB Radeon HD 3850 graphics card is the best-performing 3D card in its price class. For $179, it will let you play most current 3D games at reasonable resolutions and detail settings. The Radeon 3850 now includes hardware support for DirectX 10.1. This means that these cards will be able to play any games that take advantage of the next iteration of Microsoft’s DirectX programming interface. This is fastest 3D card under $200, solid home-theater capabilities and supports multiple card configurations with up to four cards in the same PC. The Radeon HD 3850’s other new feature is its support for PCI Express 2.0. You can still use the card on current PCI Express motherboards, but when the PCI-E 2.0 motherboards hit, you’d gain added graphics data bandwidth.


Radeon HD 3850 also retains all of the highlights of the Radeon 2900’s core technology, which makes sense, as the core design of the 3850 is a derivation of that of the higher-end chips’. Mostly that refers to its suitability as a home-theater card. Like the Radeon 2000 cards, the Radeon 3850 is HDCP compliant, which means it can display protected HD DVD and Blu-ray content at resolutions up to 2,560×1,600 from your PC, if you have such an optical drive and a supporting monitor or TV.

It also comes with an integrated audio chip, which means via ATI’s specialized DVI-to-HDMI adapter, you can pump both video and audio over an HDMI cable to an HDTV. That greatly simplifies home theater PC installations. Like most modern graphics cards, the Radeon HD 3850 requires a direct connection to your PC’s power supply to run. All you need is a free six-pin power line and you’ll be set. This model comes in 256MB of 900MHz DDR3 RAM with a 667MHz core GPU clock. The faster Radeon 3870 comes with 512MB of DDR3 running at 1.2GHz, and with a 775MHz core clock.

As they’re modern graphics cards, each uses the unified processing pipeline, which means that shaders, geometry, and all other calculations flow through the same path, which can adjust dynamically depending on the workload for each process.