Realtek AC97's Hardware Acceleration Mechanism for AC97

The traditional hardware acceleration of sound: IN-LINE

The following is a brief explanation of the traditional hardware acceleration of sound processing, using DVD playback as an example.

As we all know, when a computer plays a DVD-ROM, the CPU processor must first separate the encoded Dolby AC-3 sound from the MPEG-2 movie and place it in the DRAM buffer, and then through the AC-3 hardware decompression device from the DRAM buffer to put forward the sound data, decode it, and finally mix it with the 5.1-channel as a

With this in mind, let's take a look at the AC97 standard hardware acceleration method: multi-trip

The AC97 compliant chipset is different from the traditional audio output method in that the audio data can be redirected to USB or IEEE 1394, which can be achieved in the following steps: The first step first by the AC-3 hardware acceleration device from the DRAM to extract the CPU processor has been separated from the data beforehand; the second step by the AC-3 to perform decompression and mixing operations, the synthesized data re-feed into the DRAM in another cache area. At this point, an interrupt signal will be generated to tell the operating system that the synthesized sound data has been prepared; the third step is that the operating system will work with the CPU processor to transfer the processed sound data to the new buffer, and then feed the data into the USB pipeline and wait for the output; the last step is that the USB controller will obtain the relevant sound data and send it to the corresponding digital speakers.

How to implement AC97 in your system

In order for manufacturers to port AC97-compliant chipsets to motherboards, INTEL suggests that designers use the following three approaches:

One is to integrate both the controller and the sound decoder chip onto the motherboard, and have the two communicate via AC-LINK. The advantage of this is that the integration between the chipset and the motherboard can be optimized;

The second is to put the controller on the motherboard and the sound decoder chip on the interface card, so that the two still communicate via AC-LINK. This would give the user more flexibility in choosing the decoder chip. In addition, the interface card can also optionally add modem voice part of the transmission function.

Thirdly, the controller and decoder chip are all made on the card, which communicates with other peripherals via a 32- or 64-bit PCI bus. Of course, using IEEE 1394 or USB to communicate with the motherboard is also feasible, as long as the controller can support the IEEE 1394 or USB interface. Of course, in addition to a SIDEBAND HEADER signal line to complete the connection between the controller and the motherboard. It should be noted that this approach is the least difficult to design, and motherboard designers only need to reserve the SIDEBAND HEADER socket and related circuitry on their motherboards.