IBM lead architect: Cell CPU could take PS3 beyond gaming, into Linux

Hopewell Junction (NY) - In his first in-depth interview since releasing the version 1.0 specification for IBM's Cell microprocessor last week, the principal developer of the CPU's innovative new synergistic processor elements (SPE) tells Tom's Hardware Guide he sees Sony's PlayStation 3 - the first major platform to utilize Cell - as the driver for a new general-purpose programming environment, using Linux but bypassing the PC.

Responding to our story last Thursday on the release by IBM that morning of the 1.0 specs, Cell's chief architect, Dr. H. Peter Hofstee, advised us not to characterize the SPEs as specialized co-processors, dedicated to occasional tasks such as graphics or arithmetic. In that story, we compared SPEs to the co-processors of old, and characterized them as subordinate to the principal processing element of the Cell system, the Power Processing Element (PPE), based on the existing PowerPC architecture. But in doing so, Dr. Hofstee warned, we tended toward a trap into which others have fallen, in which the role of the SPEs appears to be reduced in importance. More than just co-processors, Dr. Hofstee said, the SPEs are fully-capable processing units that are capable not only of running threads spawned off from a main program, but also running "single-core," scalar programs in their entirety - not only multithreading, but multitasking.

But also, in making that distinction, Dr. Hofstee wanted to make certain we recognized the Cell as a powerful general-purpose processor. "[Cell] is already fairly general-purpose, even today," he said, "but of course, over time, we expect it to go even further. Over time, [whether] it is going to become the new general-purpose standard, that is to be determined." In characterizing the general-purpose nature of Cell, he told us that development systems used by IBM today are running Linux, and that general-purpose applications are being developed using a suite of Linux-based tools.

Hofstee told us that PlayStation 3 plans to use a high-end graphics processor to complement the Cell CPU. "So that concept was there from the very beginning," he said, "that we would complement the more general-purpose processor. Quite clearly, even though Cell can be pretty darned good at pushing polygons, that is not its reason for being in the game console."

What Dr. Hofstee could neither confirm nor deny, and what IBM can neither confirm nor deny, is whether any of these Linux-based development systems could be considered PlayStation 3 prototypes. An IBM spokesperson told us that Dr. Hofstee could not speak on behalf of Sony, nor for the Sony/Toshiba/IBM alliance (STI) which is responsible, at least on paper, for Cell's eventual marketing. But as Dr. Hofstee told us, "I think that this game space is a good economic engine behind this kind of innovation. There's enough momentum there that people will, in fact, take a lot of software to this platform."

Dr. Hofstee finds himself today in the type of quandary that the engineers at Atari Computer, 20 years ago, could only dream of finding themselves in. Back then, Atari's engineers had conceived an elegant, useful, and relatively fast general-purpose platform design for the 68000-based ST computer. But Atari had extreme difficulty convincing the general public that it was a competitive general-purpose platform, at the same time it was obligated to maintaining its sales commitments with retail outlets that pushed it as a game machine. Consumers rejected the ST as unsuitable for general use, especially compared with the IBM PC, even though some general-purpose ST applications at the time were more powerful.

Today, Hofstee's best option for introducing the Cell processor to the market at large, is as the driver for a gaming platform. But this is what the market wants, as PlayStation 3 is likely to be a tremendous seller, and what Sony apparently hopes will be the driver for a major new interactive media standard. Twenty years later, the market wants a game machine. Yet Hofstee sees PS3 as an opportunity to introduce consumers to a new and powerful architecture whose purpose extends beyond gaming.

The evolutionary platform Dr. Hofstee has in mind, however, is not a Linux-based PC. "I don't know if the conventional desktop is a system that I'm most interested in," he told Tom's Hardware Guide. What users are demanding is performance improvements, he said, but not necessarily in the context of the traditional personal computer. A new kind of platform, he envisioned, could make a significant difference in people's lives. At the same time, he said, there's an "insatiable demand" for more performance in the workstation and supercomputer spaces, and finally a need for greater game performance. Some of the brightest minds to emerge in science and technology, he pointed out, entered the game programming business when there were other choices available to them. As a result, a future platform of some sort, empowered by Cell, could fuse all three of these groups together. "These are all going after sort of the very similar set of problems ... so those are the spaces that I'm interested in. I'm really not interested in something that can do conventional PC apps."

The platform that would drive these emerging markets at this time appear like something other than a PlayStation 3, so we asked it could be. "I hope it's going to be pretty optimal for a PlayStation," he responded. "If it's not, then we have missed something."