What's next for Via and the C3?

THE CESSATION OF overt hostilities between VIA and Intel this week ranks as one of the more surprising developments in the IT industry so far in 2003. With the threat of Intel litigation gone, VIA may finally be able to gain some ground in the P4 chipset market, though it's still not clear if Intel has granted VIA an 800 MHz FSB licence or a Hyperthreading licence.

In the long run, however, the point is debatable, as Via has only three years before it must stop manufacturing Intel pin-compatible processors, and four years to continue making Intel pin-compatible chipsets. This type of restriction, though years down the road, will force Via to make substantial changes to their business model. It has has several options, however, including:

Attempted renegotiation of the compatibility clauses, with the goal of winning/licensing the right to make CPUs/chipsets compatible with Intel for a longer period of time.

Developing its own, independent line of chipsets and processors separate from any Intel or AMD product.

Moving to pin-compatibility with another processor manufacturer — with AMD being the most logical choice.

Although there are many factors which will affect Via's strategy, C3 performance and positioning will be a critical component of its chipset plans. Choosing to build the initial versions of the C3 to be pin-compatible with the Pentium III made sense at the time. The P3 still holds a large chunk of the total computing market, Socket 370 boards are cheap and easy to manufacture, and Via has extensive experience in designing and building chipsets for them. By leveraging the huge installed base of the P3 and providing replacement CPUs, Via hoped to gain market share at relatively little cost, similar to how AMD continued to extend the Socket 7 motherboard platform long after Intel had left it. It even went out of its way to provide remapping tricks to allow 400MHz K6-2 CPU's to be used in older Pentium-class Socket 7 boards.

Unfortunately, the C3 hasn't yet been able to capitalize on the Socket 370 platform in the same manner as the K6-2 did. Although the K6-2 lacked the FPU performance of the Pentium II processor, the chip offered acceptable floating-point muscle and highly competitive integer performance. The K6-2 also maintained rough speed parity with the Pentium II, which made a higher-speed AMD processor an excellent replacement for a lower-end Pentium chip. Certain motherboards running P200 or P233 processors could jump all the way to a K6-2 500 or 550.

The C3, however, cannot replace the P3 in this manner as the chip in its current configuration is incapable of maintaining even a rough level of total performance parity with the Intel CPU. Via's Nehemiah core will offer much-improved performance over Ezra-T, but the low cost of current P4 and AthlonXP systems make it unlikely that the C3 will ever see much use as a drop-in P3 replacement, even when the chip inevitably exceeds the P3 in performance.

Faced with this fact, Via has developed the EPIA line. EPIA is a multimedia-focused mini-ITX form factor system series designed for users who want stylized, unique computers primarily designed for multimedia systems. By selling the C3 in an embedded form and capitalizing on the CPU's extremely low power and heat characteristics, VIA has created a profitable (and expanding) niche market for its processor division.

Via's C3 development is now at a crossroads. The EPIA line provides the processor with its own market, but there are rumours of a P4 mini-ITX system soon to be entering the market and if that happens the C3 will once again find itself playing second-fiddle. Via can choose to continue using and advancing the Socket 370 platform, but in three to four years any Intel-compatible platform Via is currently using will have to be scrapped. Via has two choices: Either the C3 must be redesigned to make it clock-for-clock competitive with the other x86 CPUs on the market -- with either a unique chipset of its own or pin-compatible with an AMD design -- or it must shift the chipset that the EPIA platform is designed on. Rather than designing a new one of its own, however, it might make more sense for Via to use its own KT800 Athlon 64 chipset.

The entire Hammer platform is designed to offer high amounts of internal bandwidth and eliminate I/O bottlenecks — exactly the type of thing i that a multimedia system would need. Using an Athlon-64 compatible design would allow Via to capitalize on its already completed chipset development and the platform is very young, giving it a great deal of headroom. Maintaining the EPIA line as an embedded system, moreover, would protect the C3 from being consumed by the Athlon 64. While Athlon 64 will offer thermal improvements over AthlonXP, it will not be aimed into the same type of ultra-low power/passive cooled market the C3 is, so direct competition could be avoided in this way as well.

This is the development route for Via that seems to make the most sense. There's little point in continuing to cling to the Intel platform now that the company's ability to use it has been restricted — better to put the intervening years into finding other solutions. Keeping the C3 an embedded solution protects it from having to directly compete with higher-performing solutions from Intel and AMD, and the Hammer platform offers performance characteristics that could be very attractive for any multimedia platform. We'll just have to see how things turn out. µ