Microprocessors About Microprocessors: Myths and Technology. By Dwight Moreland CIS 221/L Instructor, Mr. John Sharifi 6 June, 1999 Index About Microprocessors Page Myths about Sub-$1000.00 PCs 1 Industry Trends.5 Shift to On-Chip Cache Pays Off.6 How Far Can Intel Go?.9 Intel Seeks Salvation on Internet.11 Socket Strategy Challenging for AMD 14 Review.17 Work Cited18 About Microprocessors Myths About the Sub-$1,000 PC The emergence of the sub-$1,000 PC market reflects an inexorable trend toward less-expensive systems that will surely doom Intel in the long run. You could easily be convinced of this argument by the recent stories on low-cost PCs. These stories, however, are based on myths, not facts, and the emergence of this new market is not likely to upset the status quo.
Myth: Sub-$1,000 PCs represent up to 40% of the PC market. This gem is based on misinterpretations of a survey by Computer Intelligence that shows an increasing percentage of PC sales below $1,000 during the course of 1997. Sales in this category peaked at nearly 40% in August, but this survey measures only the U.S. retail desktop PC market, which represents less than 10% of the worldwide PC market and has specific characteristics that make it unusually receptive to sub-$1,000 PCs. In particular, it is mainly a consumer channel; two-thirds of all PCs are still bought by businesses, where the sub-$1,000 concept is far less popular.
Many business users seek faster systems to increase productivity. Others shun low-end systems to avoid rapid obsolescence. Although low-cost PCs run today’s software adequately, firms looking to upgrade to Windows NT and other advanced software in the future might not be able to afford the limitations of a sub-$1,000 PC. These systems also lack expansion slots in some cases, giving new meaning to the term fixed asset. The survey also doesn’t include worldwide demand. While the sub-$1,000 PC is popular in emerging markets such as China, the large markets such as Europe and Japan have been slow to change their focus.
Finally, the survey excludes mobile systems, which make up more than 20% of the total PC market but are not available for less than $1,000. We believe total sales of sub-$1,000 PCs were perhaps 5 million units in 1997 and will approach 10 million units, roughly 10% of the worldwide PC market, in 1998. Myth: Sub-$1,000 PCs have been enabled by low-cost CPUs from Cyrix and AMD. Sub-$1,000 PCs have been available for years; the problem has been that their capabilities were well below the needs of contemporary software. The new popularity of low-cost systems has been driven mainly by two factors: a drop in DRAM and storage prices along with a lack of performance-hungry software.
DRAM prices have dropped nearly 90% in the past two years. The price of a 1G hard drive also fell. In years past, PC makers would simply increase the DRAM and hard-disk capacity in their entry-level systems, keeping prices constant. Component prices fell so fast in 1997, however, that many vendors chose to discount their entry-level configurations rather than beef them up. This decision was aided by a relative stagnation in software demands. Using a low-end processor from one of Intel’s competitors can shave $20 to $40 in cost, a significant amount in a sub-$1,000 PC.
But similar savings can be achieved by adjusting the memory, storage, or other features of the system, which account for 80% of the total system cost. Many PC makers have decided the Intel brand name is worth the extra expense. Both HP and Micron, for example, have chosen a Pentium/MMX-200 for their $799 PCs. Myth: The trend toward lower prices is irreversible. The factors that made low-cost PCs popular can easily be halted or even reversed. Since December, DRAM prices have firmed and even started to rise, as bleeding chip makers finally began to reign in production.
This situation, exacerbated by Asian financial turmoil, could cause shortages as early as next year. In addition, the unusually fast progress in hard-disk density over the past two years is likely to return to historical rates. More important, new software will emerge that demands faster processors, more memory, and larger hard drives. Windows 98 won’t do the trick, but digital photography, video conferencing, voice recognition, and other muscular applications might. Both Intel and Microsoft, companies I avoid betting against, are investing heavily to make sure that today’s software doesn’t remain good enough. Demand for low-cost PCs will dry up if those systems can’t run the most popular new programs.
These factors won’t change immediately, and sub-$1,000 PCs are likely to be popular in some channels throughout this year and probably beyond. They will remain a relatively small part of the total PC market, however, particularly when measured in dollars. This price segment will provide a foothold for Intel’s competitors, but to achieve their market-share goals, AMD and Cyrix must do more than simply sell cheap processors. Industry Trends In February 1999, Intel introduced the Pentium III CPU (code name Katmai). It is a Pentium II with the new instructions for multimedia, known as Streaming SIMD extensions. On May 7, 1997 Intel introduced the Pentium II CPU (code named Klamath).
It was heralded as the fastest microprocessor for the server environment. Major system manufacturers also unveiled their offerings which use the Pentium II processor with MMX technology. Intel shifted motherboard manufacturers over to this slot 1 configuration from the previous Zero Insertion Force (ZIF) socket. This Pentium II processor is packaged in a SEC (Single Edge Cartridge), and this cartridge fits into a slot on the motherboard. The Pentium III processor is packaged in SECC2 cartridge. The socket 7 (Pentium) and socket 8 (Pentium Pro) use Zero Insertion Force (ZIF) socket.
Slot 2 configuration is Intel’s workstation and server design, based on their Pentium II Xeon processor. A Slot M configurations are also being developed by Intel. AMD and Cyrix built 6th generation CPUs with the equivalent packaging as the Intel Pentium family CPUs. Even though they are targeted to compete with the newer Intel processors with MMX technology, they work in many of the Pentium based motherboard, ie. socket 7 solutions. AMD, VIA, and SiS are producing or have plans for producing chipsets capable of handling a 100 MHz external bus for socket 7 motherboards.
AMD calls this Super7. According to Intel, the SEC was designed to overcome many of the inherent limitations of the ZIF socket solution. Many industry observers state that Intel’s Pentium Pro fabrication processes was not efficient. However, Intel has recently returned to socket processors with the release in Fall ’98 of their Celeron processor in PPGA (Plastic Pin Grid Array) packaging. Again, yield and cost were reasons for the change in product.
Shift to On-Chip Cache Pays Off Intel’s Slot 1 and Slot 2 Will Give Way to Sockets by 2000 Intel’s decision to release its Mendocino processor without a module in 1Q99 is just the tip of the iceberg. By the end of next year, Intel will expect to be shipping moduleless processors into all of its market segments, and by the end of 2000, virtually all of its chips will plug into sockets instead of slots. This trend will be enabled by a shift to on-die level-two (L2) cache, which makes today’s module structure superfluous. The initial purpose of the Slot 1 module was to hold the external L2 cache chips required by the Klamath and Deschutes CPUs. Mendocino doesn’t need external cache chips, as it is Intel’s first processor to incorporate the entire cache subsystem. To maintain compatibility with these earlier processors, Mendocino is currently shipping in a Slot 1 module, despite the fact that, other than the CPU, the module contains no active components and is nearly empty.
Mendocino will discard this vestigial module early next year, instead appearing in a 370-pin plastic PGA that plugs into a standard socket. Intel will continue to offer the Slot 1 version as well, but we expect PC makers to convert to the so-called Socket 370 over time, as it is smaller and less expensive than Slot 1. By the end of 1999, the Celeron segment will be largely converted to Socket 370, and future Celeron parts are likely to appear only in a socketable form. Intel will benefit from this transition as well. According to the MDR Cost Model, the PPGA version of Mendocino will cost about $10 less to manufacture than the current module.
According to our shipment forecast for 1999, that decrease could save Intel up to $300 million next year. Not one to let a good idea go to waste, Intel will eliminate modules from its other product lines as well. On-die L2 cache is very attractive in the mobile market because it reduces power consumption and physical size as well as cost. Sources indicate that in early 1999 Intel will deploy a mobile product code-named Dixon that has 256K of L2 cache on the chip. Given its power and size advantages, Dixon is likely to displace the current Mobile Deschutes fairly quickly, eliminating the comparatively bulky minicartridge.
Process technology is holding back the other segments. In the current 0.25-micron technology, even Mendocino’s piddling 128K cache takes up about 35 mm2 of die area, adding $10 to the manufacturing cost of the chip. Dixon’s cache is likely to double that cost overhead, eliminating any cost advantage over a module. In a 0.18-micron process, however, a 256K cache will fit into that same 35 mm2, making it less expensive to integrate that amount of cache than to add an external cache. For this reason, we expect Intel’s Coppermine processor, a 0.18-micron version of Katmai, to incorporate an on-die 256K cache.
With on-die cache, Coppermine could plug into the same Socket 370 as Mendocino, although Intel will probably also offer a Slot 1 version as a transition vehicle. We expect Coppermine to appear in 2H99, and by mid-2000, it could largely eliminate Slot 1 from the PC market. Even the Slot 2 segment could move away from modules. Intel says that future high-end processors Cascades, Foster, and McKinley will include on-die L2 caches as large as 2M (see MPR 10/26/98, p. 16).
Like Mendocino, Cascades is likely to appear in both slot and socket versions, but Foster will probably use only a socket. Because servers and workstations are less sensitive to the cost and size penalties of the module, the transition to sockets will take longer in these segments than in the PC segments. The 0.18-micron process is required in this segment as well: regardless of cost, the 0.25-micron process is simply incapable of building a processor with 2M of on-die cache, which is required to match the current Xeon line. Again, the cost savings of on-die cache are considerable: we estimate a 2M Cascades, even at a projected die size of 375 mm2, will cost 50% less to build than today’s 2M Xeon module, given the high cost of the Xeon’s four custom cache chips. Other vendors are also putting large caches onto their processors. AMD’s next K6, known as Sharptooth, will include a 256K L2 cache. IDT will instead use 128K of primary cache.
These chips will plug into Socket 7 but, like the Socket 370 parts, will not need external L2 cache, thus reducing system costs. The big exception to this trend will be AMD’s K7, which will introduce the Slot A module just as Intel is shifting its products to Socket 370. Once the K7 moves to 0.18-micron technology, it too is likely to adopt on-die L2 cache, turning Slot A into Socket A. The big losers in this transition are SRAM vendors, which won’t have the PC market to sell into much longer. But shifting SRAM production to the processor vendors will require them to add fab capacity.
Building this capacity will take time, slowing the transition to on-die cache. But build it they will. In two years, modules will be a fading memory. How Far Can Intel Go? When the U.S. Federal Trade Commission (FTC) launched a new antitrust investigation of Intel a few months ago, the usual tales made the rounds: stories of Intel threatening to cut off supplies of its processors to system makers that purchased from Intel competitors or otherwise transgressed. Intel denies all of these tales, of course, forcing the FTC to sift through a mountain of paperwork seeking the proverbial smoking gun.
Intel freely admits, however, to making a different kind of threat, one that may ultimately catch the FTC’s attention. Intel maintains, for the most part, good relations with all of its major customers. For system makers that design their own system-logic ASICs or motherboards, a key benefit of this relationship is early access to technical specifications of future Intel processors. Boards and particularly ASICs can take a year or more to design, test, and debug before they are ready for production. To deliver products based on the newest Intel processors when they are released, system makers must have detailed technical specifications for those processors a year or so ahead of time. To obtain early access to this technical information, Intel’s customers must sign a nondisclosure agreement (NDA) with Intel.
Like most NDAs, Intel’s agreements contain a claus …