Radeon HD 7970

  For quite a long time, since the time of the Radeon HD 2000 series, the AMD GPU architecture has not undergone fundamental changes, until the HD 6000 "Cayman" series, AMD, for its graphic solutions, used the VLIW (Very Long Instruction Words) architecture, which is more degree is very efficient in processing the graphics instructions that underlie 3D graphics calculations, but proved to be ineffective for general-purpose computing, when the GPU must queue jobs with scalar instructions in general-purpose applications. And now, starting with the HD 7000 series, the GPU uses a new microarchitecture - GCN (Graphics Core Next), designed primarily to improve the efficiency of computing not directly related to 3D.

For example, to execute a certain queue of instructions, a GPU on the VLIW4 architecture will spend six cycles, and a GPU on the GCN architecture will execute the same instructions in just four cycles. Based on this, the theoretical peak performance of the Radeon HD 7970 can reach 7.5 times higher than the Radeon HD 6970 due to more efficient use and distribution of computing resources of the new GCN architecture.

GPU architecture

The Tahiti GPU is manufactured using the advanced 28nm process technology at  TSMC . The structural structure of the GPU is shown in the image below, as you can see, "Tahiti" uses 32 calculation units (CU) with 64 ALUs (arithmetic and logic unit) per CU, which is a total of 2048 ALUs. operating clock frequency of 925 MHz, the GPU throughput for 32-bit mathematical calculations reaches 3.8 TFLOPS and 947 GFLOPS for double-precision GFLOPS operations.

The L1 cache provides about 2 TB/s throughput at this clock speed, using a 768 KB L2 cache to support it.
With six 64-bit memory controllers forming a 384-bit memory bus, using fast GDDR5 memory operating at 1375 MHz, the Radeon HD 7970 is capable of delivering an impressive 264 Gb/s bandwidth, which is 100 Gb/s higher than the Radeon HD 6970.

The Tahiti GPU has two updated geometry processing engines optimized for tessellation with an increased buffer cache for vertices, which has improved the performance of calculations from 1.7 to 4 times, depending on the number of operations that are applied to the original vertex.  
 
AMD PowerTune Technology

One of the interesting innovations in the Tahiti GPU is the PowerTune advanced power management technology, which provides direct control over the power consumption of the GPU. This technology helps to set high GPU frequencies, get high performance in games, and not be afraid that consumption may go beyond safe limits. Similar technologies were already used in the Radeon HD 5000 family, but they were rather primitive and changed the frequency and voltage of the GPU in steps, in the Radeon HD 6900 the system moved to a qualitatively different level.

The GPU constantly, at intervals of milliseconds, measures the load and power consumption of all constituent units and, based on the analysis of the task being performed, automatically adjusts the frequency and voltage, remaining within the specified TPD, and if possible, the GPU clock frequency increases.

Also, the new GPU has a new technology - ZeroCore, which allows to reduce the power consumption of the GPU in idle mode. ZeroCore puts the GPU into "sleep" mode when the monitor is not receiving a signal from the PC.

According to AMD, the graphics card consumes only 15 watts when processing a static Windows environment and does not consume power when there is no signal to the monitor. 
 
 
PCI Express 3.0

The Radeon HD 7970 introduces support for the third generation of the PCI Express standard (PCI-E 3.0), which should increase the data transfer rate to 16 GB / s in both directions and help eliminate weaknesses when working with some heavy applications.

Eyefinity 2.0

The updated technology supports configurations with five displays in a row in landscape or portrait orientation. Now the central monitor may be larger than the others vertically.

Simultaneous operation of Eyefinity, AMD HD3D and CrossFire technologies is supported. The maximum resolution of the "prefabricated" screen has been increased to 16384 × 16384 pixels.

Radeon HD 7970

The table shows the technical specifications of the Radeon HD 7970 compared to the previous generation model, the Radeon HD 6970. The most important innovation in the Tahiti GPU is its production using 28nm technology. "Tahiti" contains 4.31 billion transistors, which is 63% more than its predecessor "Cayman", while its die area is somewhat more compact - 365 versus 389 mm2. The clock frequency of the GPU has increased to 925 MHz. Computing power has increased by about 40% and is 3.79 TFLOPS. The number of texture units increased from 96 to 128, which made it possible to obtain a texturing speed of 118.4 GT/s.

Like its predecessor, the Radeon HD 7970 uses 1375 MHz (5500 MHz) GDDR5 memory, but with a 384-bit bus, thanks to which the memory bandwidth is increased from 176 to 264 GB / s. 

Radeon HD 7970 Specifications 

Appearance

The cooling system and general design of the Radeon HD 7970 at first glance is similar to the Radeon HD 6970, even the dimensions (26.25 cm long and 11.25 mm wide) almost exactly repeat the dimensions of its predecessor, but still the HD 7970 has some differences.

The front side of the video card is covered with a glossy plastic casing with a bright red insert and a more "refined" design than its predecessor.
There is no protective heat-dissipating metal plate on the back side of the board, unlike the Radeon HD 6970.

The cooling system itself has not changed cardinally, but it has some improvements aimed at improving the cooling of the board. The cooling system fan now has wider blades and a larger diameter of the impeller, the developers removed the second DVI port from the rear bar to expand the grille for blowing hot air.