PARIS As mobile phones with gaming features grow hotter, dedicated mobile game handheld devices are loosing their luster among SoC designers, industry executives said.
The latest victim is STMicroelectronics' development project called Pocket Multimedia Platform (PMM), which targeted handheld entertainment applications. The company told EE Times that it cancelled the project about six months ago.
"Under the current economic conditions, we found the [PMM] platform a bit of a risk," an ST spokesman said. ST stopped the project in mid-2002 and decided to have the team "refocused on the imaging area," the spokesman said.
ST originally intended to include on the PMM platform, based on SuperH, Inc.'s lower-power RISC core, flexible combinations of several cores including VLIW cores and hardware blocks for audio, video and graphics acceleration. The cores were to be designed for deeply embedded multimedia functions.
In a Novermber 2000 interview with EE Times, Philippe Geyres, ST's coporate vice president, said, "Customers don't care how many cores we use. What matters in a power-conscious platform is the best system architecture that can minimize bus traffic and memory access."
As recently as 12 months ago, ST and Imagination Technologies were saying that ST's PMM platform would integrate a SH RISC core with Imagination Technologies' PowerVR MBX 3D/2D and video acceleration mobile graphics technology, along with a range of IP from ST, including connectivity, video and display, storage and security and related software.
While dumping the SH RISC-based PMM platform, ST has solidified a path to pursue the ARM-based Nomadik processor, a multimedia application processor recently introduced by the company for the cell phone market. Although the ST spokesman stressed that the PMM platform-based SoC was never intended for wireless mobile handsets, it is unclear which customers ST had in mind when it launched the project three years ago.
ST is not the first company to be burned by the unproven market for dedicated mobile gaming devices.
NeoMagic Corp. was also forced to cancel a SoC design integrating 32-bit MIPS 4Kc RISC CPU, dedicated hardware and 2D/3D rendering engine. It as originally developed for a handheld gaming system.
Mark Singer, vice president of corporate marketing at NeoMagic, said the gaming device was under development by "a Silicon Valley-based skunk work, funded by a large system company in the cellular industry." As funding dried up and the parent company pulled the plug on the start-up in 2001, "our SoC project also went away," said Singer.
Although Singer declined at the time to identify NeoMagic's customer, it was reportedly a mobile gaming company called Red Jade, which was launched in 2000 and funded by Ericsson in Silicon Valley.
Red Jade "was cancelled before it went to production due to the downturn in Ericsson's fortunes, but they did seem to come up with some good ideas along the way," said Randall Fahey, an independent consultant focuing on DSP and wireless communications.
Despite missteps, ST and NeoMagic said they are focused on the development of multimedia application processors for mobile handsets. ST is betting big on the Nomadik processor platform, while NeoMagic, moving away from its failed MIPS-based SoC project, is refocused on an ARM-based SoC, based on a new architecture called Associative Processing Array (APA).
APA uses circuit structures that perform bit-level processing within an array of memory cells. Fahey said, "APA seems to make some sense, because lots of processing occurs in the memory structure without moving things around very much." He added that it could solve "a huge problem of power consumption in a processor."