To: wanna_bmw who wrote (149030 ) 11/21/2001 9:02:08 AM From: Dan3 Read Replies (1) | Respond to Re: Intel's pipeline for next year includes: - Northwood with headroom up to 3GHz - basically a shrink of a mediocre chip- Hyperthreaded Xeons - SMT will help IPC on some code, this is a matter of catching up for Intel, not pulling ahead- DDR memory support - yes, copying AMD helps- McKinley based Itanium - bringing its performance closer to that of an entry level Celeron, and maybe this stepping will work. What will it cost and what software will it run?- Plumas dual DDR server chipset and i870 McKinley chipset. - a nice enough sounding chipset, without a competitive processor to run- .13u and 300mm wafer fabs - they've had .13 for months, 300mm may or may not save a few dollars per chip, 300mm is irrelevant. Infineon has been running 300mm for years - are they dominating the RAM industry and making higher profits?- Pentium 4 for mobile - kind of says it all, doesn't it? P4 is a terrible core design for a mobile chip, and Intel's leaky .13 won't help much. Mobile is looking very good for AMD next year.- USB 2.0 - Yeah, cling to that one, it will save Intel for sure.AMD's pipeline for next year includes: - A .13u process that's already spent - in your dreams, perhaps. Remember that when independent labs analyzed P4 and Athlon using one metric, they found that both had .09 transistors. - An SOI process that will add huge costs - direct FAB costs in this business are a tiny part of total costs. The day that changes, Intel will be selling processors for $25. Look at the results IBM and Motorola are getting from SOI. Consider that SOI can dramatically reduce power consumption, and that Intel is already in something of a corner due to P4's power hungrey design.- An 8th generation product that has already been delayed 3 times - Compared to Itanium it's racing to market. It's a design that is performance compatible with existing code. It's a design that will be no more expensive to implement that 32 bit solutions, so it can seed the market for 64 bit developers without AMD pouring an extra $5 billion into technical marketing.- Potentially, a much stronger mobile chip - almost certainly, a family of much stronger mobile chips. Even without SOI, Athlon4 looks quite competitive with P4. AMD will have low end and high end, SOI and simple shrink mobile chips next year.- a recovery in the Flash market - which for AMD will be golden. Flash is a large enough part of AMD's business that a flash recovery will have a huge, positive impact. Intel is basically a one trick Pony, which lives or dies with the corporate CPU market.- Madison based Itanium chip - a shrink of a lost cause.- Springdale DDRII chipset - copying AMD's standard- Serial ATA, Integrated Wireless Networking - new case colors, too? You're looking a little deperate, here. What does any of this have to do with Intel falling behind in the processor business?- Prescott based Pentium 4 with Hyperthreading - IPC on P4 is so far behind, that SMT will have a tough time doing catch up, much less producing any improvement relative to P4's competition.- 3GIO - So what? Color coded cables, too? Why aren't you writing that AMD's Hypertransport (with actual products shipping, or near to shipping, from various companies) is a big deal for them? The answer is that neither 3GIO nor Hypertransport will have a material effect on the relative performance of the two companies. But, in Intel's case, there isn't much else to hope for.- .09u manufacturing - with the lack of progress shown by Intel's .13, this is not looking particularly important. Will they make it to SOI in 2003? If not, leakage may make their .09 unable to keep up with their .13.- a Hammer based server chip - setting the standard for the next (64-bit) generation of IBM PC compatible systems for workstations and servers.- a Hammer based mobile chip - setting the standard for the next (64-bit) generation of IBM PC compatible computers for mobile systems.- .09u manufacturing, only through foundry services - letting them extend the next (64-bit) generation of IBM PC compatible computers to entry level systems.
