It has been a long, hard road for Cyrix. Competing with Intel is a tough business, and Cyrix's execution just wasn't good enough. After the company became part of National, progress slowed dramatically while strategies were rethought and turfs established. Now, National has thrown in the towel on the PC processor business, leaving Cyrix with an uncertain future.
Ironically, Cyrix unveiled a revamped processor road map just weeks after National's announcement. Cyrix had once planned to differentiate its PC processors through integration. Now, National will develop integrated processors for information appliances, but Cyrix-focused on PCs-will stick with the integration levels and bus interfaces defined by Intel.
Thus, the MXi, an integrated processor based on Cyrix's Cayenne core, and the M3, a next-generation version that was to include Direct Rambus memory interfaces and Cyrix's Jalapeno core, have both been canned (or at least indefinitely delayed). These chips are too pricey for information appliances, and Cyrix didn't find much receptivity in PCs.
Instead, Cyrix's road map now calls for a chip code-named Gobi, which integrates the Cayenne core with a 256k on-chip L2 cache and a Socket 370 bus interface. If it ships as planned in the third quarter, it will be the first Celeron pin-compatible chip on the market and could mark the beginning of Socket 370 as a multivendor standard, following in the footsteps of Socket 7.
The M3 has been replaced on the road map with Mojave, due in the second quarter of 2000, which will put the Jalapeno core (Cyrix's first ground-up new design since the 6×86) into a Gobi-like Socket 370 product.
The new road map shows that Cyrix's plan calls for continuing to compete head-to-head with Intel, rather than differentiating through integration. This can be a successful strategy-especially with AMD heading off in a divergent direction-if sufficient clock speeds can be achieved. If Gobi is limited to Intel's lowest Celeron speed and below, it will be relegated to bargain-basement prices where it is tough to make money. Mojave should be much faster, but Intel will have changed the landscape by then; the question is how Mojave will compare to whatever Intel is shipping at the time.
Quick, nearly flawless execution is necessary to succeed against Intel. Intel's competitors have had a hard time clearing this bar, and it will be even harder during another ownership transition.
-Michael Slater () is principal analyst at cahners Micro Design Resources ().
Leave a Reply
You must Register or Login to post a comment.