To: Jim McMannis who wrote (32988 ) 5/28/1998 6:05:00 PM From: Matt Webster Read Replies (2) | Respond to of 1570320
I haven't weighed in on the K6 and AMD in a while, but it seems like now would be a good time. K6-2 looks like a good product in that it brings gaming performance up to par with Intel Pentium II. However, to get the nice performance increase, it requires the support of graphics card drivers or DirectX 6. I don't think this is a real problem because 3DFX and Nvidia (the two dominant suppliers for gamers) have 3DNow support. As for DirectX 6, it doesn't matter because there are no DX6 games; when the games come, they will install the drivers seamlessly (hopefully). As for DX5 games, the window is only about 3 months, and if I know AMD the K6-2 won't be shipping in volume until then anyway... As much as I would like to see AMD provide competition to Intel, I still fail to see how the K6-2 will ever make serious inroads into the Intel franchise. I'm sure that it will sell in some low-end boxes (never Gateway's or Dell's), but the Intel roadmap is such that it will never reach into the high-end profitable part of the line. There is no equivalent of Xeon or 1MB Pentium Pros that are the purest cash cows for Intel. As for the 100 MHz Super7 project, I think this is an unmitigated failure. As a corporate buyer, there is absolutely no way I am going to buy an oddball chipset for my corporate desktops. It is simply too much risk for virtually no gain. The whole 100 MHz bus speed issue of the new PII's and the K6-2 is a crock and a real disservice to consumers, who would be better served by either more RAM, a 3DFX card, a dual modem setup or some other gadget. The DRAM industry should have killed the 100 MHz SDRAM standard and saved itself a ton of money. The bottom line is that with the K6-2 AMD has created a competitor to the Celeron, no more, no less. It will be able to boast strong gaming and basic productivity support, and it will be cheap. I imagine it will be in the $100-150 range. It will not be competing with the high-end of the Intel PII line. When the process ramps up to more than 333 MHz, Intel's PII line will have moved on, pushing the PII 350-400 into the sweet spot from the high end and moving on to 0.18 micron. We've seen this happen time and time again. Factor in AMD's historical yield problems, and it's worse. I think the K5 and K6 demonstrate that microprocessors are a natural monopoly. If the incumbent chipmaker has any business savvy and enough cash to maintain a lead in fabrication equipment and process, it will be next to impossible to catch up. It is time for AMD to exit the mainstream processor business and concentrate on a profitable niche. I would suggest they try to compete at the high-end only, perhaps with some Alpha bus derivative, parallel processing or what have you. They could then farm out the fabrication to TSMC or some other offshore producer and not have to worry about the huge fixed costs of equipment and fab maintenance. If AMD would move to the high end, and NSM and IDT work at the low end, Intel would still have competition, which would keep prices low for the consumer. If AMD did have a viable competitor to Xeon, I bet it could work out a favorable fabrication deal with a variety of players. So here's the question of the day: If ordered to do so, would it be profitable for AMD to concentrate its efforts on the high-end niche? Would anyone else recommend this strategy for AMD? I am neither long nor short AMD or Intel. Matt