Last week a coalition of high-tech industry leaders — including Apple, Nvidia, and Sun — announced the formation of the HyperTransport Technology Consortium, a nonprofit corporation that supports the future development and adoption of AMD’s HyperTransport I/O Link specification, formerly code-named Lightning Data Transport or LDT. So what does all this mean?
HyperTransport interconnect technology is a new high-speed, high-performance, point-to-point link for integrated circuits, developed to enable the chips inside of high-performance computer, networking and communications devices to communicate with each other faster than with existing technologies. HyperTransport technology’s bandwidth of 12.8GB/sec purportedly offers up to a 48-fold increase in data throughput, compared with existing system interconnects that typically provide bandwidth up to 266MB/sec. HyperTransport can also be built atop the existing PCI standard, which should be an interesting combination since the latter will offer a performance bottleneck.
Computer-bus performance approximately doubles in performance every three years. While this may sound fast, it’s slow compared to developments in other areas of computing technology and results in a computer system “bottleneck.” This is the main issue that HyperTransport is designed to combat, David Every explains in a Working Mac article.
“HyperTransport isn’t just a new bus implementation that’s a little faster than its predecessors,” explains Every. “It’s a whole new bus architecture, designed to grow over time. It offers speeds that are dozens of times faster than current and other proposed buses.”
The current main bus on a Mac runs at around 1GB per second. AGP runs at less than half that, and PCI usually runs at about 128MB to 256MB per second, though all three have real-world issues that keep their performance lower than their theoretical numbers, Every says
“HyperTransport can possibly replace all of them (internally), and it runs at up to 12.8GB per second,” he adds. “Its standardized design will make it easier for hardware designers to work with more buses, and to design better-balanced systems that they can adapt to their needs.”
HyperTransport allows bus technology to leap ahead, not just one or two times faster, but dozens of times faster, he says. He predicts that it will do for bus technology and internal components what FireWire is doing for external devices.
The previously mentioned companies and over 177 others are working with AMD on the HyperTransport technology. As a charter member, Apple gets access to all specs and can be a part of ongoing management of the technology.
HyperTransport complements externally visible bus standards like the Peripheral Component Interconnect (PCI), as well as emerging technologies like InfiniBand and 10GB Ethernet. Although initially developed for high-performance personal computer and server platforms, the technology is also gaining momentum in networking and communications devices, embedded applications and other non-PC devices. Multiple products integrating support for HyperTransport technology are in development to support desktop and notebook PCs, workstations and servers, and Internet communication devices.
The HyperTransport Technology Consortium is a nonprofit corporation controlled by its members. The consortium promotes the common business interests of providers to the networking, telecommunications, computer and high-performance embedded applications through the conduct of a forum for the future development and adoption of the HyperTransport specification. Product samples using HyperTransport are currently available, and other components are planned to be in volume production by the end of 2001.
“The consortium represents the commitment of its members to develop industry-wide adoption of HyperTransport technology, and to drive this state-of-the-art technology into the market,” said Gabriele Sartori, newly elected president of the HyperTransport Technology Consortium, in a statement. “HyperTransport technology is a popular new chip-to-chip communications technology that is endorsed by major industry players. It is designed to provide a smooth transition from existing technologies and to support the necessary scalability and bandwidth for future product innovations.”
Jon Rubinstein, Apple’s senior vice president of hardware engineering, said that Apple was “delighted” to participate as one of the leaders in the HyperTransport Technology Consortium,” as the technology “represents a great technology with exciting future potential.”
Last month Nvidia, Apple’s current favorite provider of graphics chips, incorporated AMD’s HyperTransport data bus technology into its new nForce platform processing architecture, becoming the first chipset to incorporate AMD’s data bus technology. In Nvidia’s case, the HyperTransport technology is designed to deliver up to a six-fold increase in bandwidth between the nForce Integrated Graphics Processor (IGP) and nForce Media and Communications Processor (MCP), according to Dan Vivoli, the company’s VP of marketing.
The Nvidia nForce platform processing architecture combines the company’s graphics with the industry’s first implementation of the new Dolby Digital Interactive Content Encoder, a technology built in the MCP that encodes multi-channel audio into Dolby Digital 5.1 in real-time.
“Nvidia’s incorporation of AMD’s HyperTransport technology is critical in our quest to deliver unmatched system performance,” Vivoli said. “Today’s personal computer applications are increasingly complex with advanced 3D graphics, high-speed networking, streaming video, and cinematic 3D audio. In fact, the bandwidth supplied by HyperTransport allowed us to implement full Dolby Digital 5.1 3D audio processing and broadband networking in the MCP, something no other current technology allowed us to do.”
Both AMD and Nvidia have supported Double Data Rate (DDR) memory for personal computers and video cards. The Nvidia nForce platform processing architecture includes a 128-bit memory interface that is designed to double the peak data throughput over today’s PC2100 DDR systems to 4.2GB per second, which is 33 percent more than dual channel RAMBUS Pentium 4 systems.
While it’s unclear exactly what Apple’s interest in HyperTransport means, the company is obviously keen on the new technology, which interestingly is much faster than even the most recent revisions to the FireWire standard (which maxes out at 1,600Mbps), though this certainly doesn’t imply that HyperTransport is being lined up as a FireWire replacement.
As Every explains, HyperTransport is a way to combine some network/shared device concepts, such as FireWire, with a faster serial port concept known as Low Voltage Differential Signaling (LVDS), and apply these design philosophies to a computer’s main bus. A standardized design for all the buses makes it less expensive to add more devices on a bus, and you don’t have different types of buses to figure out and debug, he says.
And since HyperTransport was originally invented by AMD (and API), this will certainly reignite speculation that AMD is a potential chip provider for Apple. Regardless, HyperTransport could appear in Wintel systems built on AMD Athlon or MIPS chips before the end of 2001 or early in 2002. Based on Apple’s release schedules and adoption-curve cycles for new technologies, Every doesn’t expect a HyperTransport-enabled Mac system before mid/late 2002. Even then, Apple will probably phase HyperTransport only into higher-end systems (like servers) first and gradually move it into its full lineup, he adds.
“I also expect the first implementations to be a little conservative, such as using HyperTransport only for memory and processor buses, and bridging to traditional PCI and AGP for most I/O,” Every says. “This will make a big difference by itself, but gradually I expect it will be used more aggressively in topologies and supporting multiple device links, and then we’ll see some mind-blowing systems.”