AMD Radeon HD 7970 Evaluation

Over the past 12 months, AMD has made the Radeon HD 6970 their premier GPU graphics card, competing with the GeForce GTX 570 and sometimes challenging the mighty GTX 580. For about $ 369 you can still get a hell of a graphics card today that consists of 2.6 billion transistors, a chip that is 389 mm2 in size and has a TDP of 250 W, but above all is capable of Run all modern PC games smoothly with the exception of a few extreme scenarios.

Compared to its previous generation board (the Radeon HD 5870), the Radeon HD 6970 was an average of 24% faster. Now, a year later, AMD is launching its Radeon HD 7000 series, codenamed "Southern Islands".

The Radeon HD 7970 is the first in a series of upcoming graphics cards that will make the leap into the 28 nm manufacturing process. The new HD 7970 will be AMD's new flagship GPU graphics card from January, when the board is expected to ship.

The chip shrinkage means AMD can cram more transistors into the same space, a lot more. Although the chip size is only slightly smaller, there are about 1.7 billion more transistors at 365 mm2, making a total of a whopping 4.3 billion. That number corresponds to that of the GeForce GTX 580, which has 3 billion transistors in its massive 520 mm2 chip.

Nvidia will update its graphics palette in early 2012. The GeForce 600 series is expected to use the 28nm process as well. In order to beat Nvidia, AMD has postponed the (soft) start date to the end of December.

Just two weeks before it'll be possible to get your hands on a new Radeon HD 7970 graphics card, it's definitely nice to take a look at its performance now. Before we dive into our gaming benchmarks, though, let's take a moment to dig deeper into the capabilities and features of the new card.

AMD Graphics Core Next

The Radeon HD 7000 series is a big leap for AMD and represents the most significant architecture overhaul in the last decade. It was then that AMD adopted the Graphics Parallel Core architecture, which uses groups of scalar processors that process very long instruction words, commonly called VLIW be abbreviated.

The Radeon HD 5000 series used the VLIW5 architecture, while the 6000 series switched to a more advanced VLIW4 architecture last year. The Radeon HD 7970 and all other graphics cards based on the high-end Tahiti core, however, replace the VLIW stream processor clusters with the processing units designated by AMD as GCN (Graphics Core Next).

A GCN is basically a GPU that can perform both graphical and computing tasks with a high level of efficiency. This is AMD's answer to Nvidia's Fermi architecture and makes the Radeon HD 7970 a powerful computer workhorse. It's worth noting that only high-end HD 7000 series cards are based on the new Graphics Core Next architecture, while all other models continue to use VLIW.

In essence, a shader cluster is now referred to as a GCN computing unit, and each unit is a superscalar processor with scalar and vector elements that follow a new non-VLIW instruction set architecture. This is a more efficient architecture that delivers more performance per millimeter square of GPU chip area.

The HD 7970 features a new 9th generation tessellation geometry engine with optimizations such as increased reuse of vertices, improvements to off-chip buffering and larger parameter caches. This helps improve performance for all tessellation factors with up to 4x the throughput of the HD 6900 series (Gen 8).

Leave a Reply

Your email address will not be published. Required fields are marked *