Pentium 4 upgrade is scrapped by Intel
Changing course again, Intel on Thursday canceled its much-promoted 4-gigahertz Pentium 4 microprocessor and said it would switch to a new chip design.
The scuttling of the personal computer processor was another embarrassing move for the world's biggest chip maker and the latest in a series of missteps. The change in strategy is the result of big bets that Intel made over the past several years in both chip designs and expanding its manufacturing capabilities. The full significance of the shift remains to be seen.
To replace the discarded chip, Intel will create a 3.8-gigahertz chip with a double dose of memory in the first half of next year. This memory, known as cache, helps a processor run more efficiently by storing frequently used data close to the processor so that it doesn't have to retrieve data from the slower main memory chips.
If Intel is switching to a new design to make better use of its manufacturing plants, then the strategy seems more like a knee-jerk reaction to falling demand for its chips. But if the company has planned more deliberately, the new design could pay off in the long run.
``It's embarrassing to Intel because they talked about this 4-gigahertz goal publicly for so long,'' said Dean McCarron, an analyst at Mercury Research in Scottsdale, Ariz. ``On the other hand, it could be a non-event financially.''
Intel said it decided it could create chips with more built-in memory that didn't run at 4-gigahertz speeds but still delivered performance that would satisfy customers. The company also said it is designing chips that exploit manufacturing capacity that has become available more quickly than anticipated.
For Intel, the scuttling of the chip raises the risk that it might fall behind rival Advanced Micro Devices in the quest to provide the fastest processors. The move also means consumers won't get the fastest chips Intel had promised to make available for the holiday season.
Intel's detractors viewed the move as a stop-gap emergency measure to deal with the fact that its designs aren't as competitive as those of Advanced Micro Devices. AMD has begun to gain market share with chips that deliver more efficient performance than those made by Intel. Critics also contend Intel is switching gears to deal with a looming factory capacity problem.
Spin
``Everybody is putting their spin on this, like a presidential debate,'' said Kevin Krewell, editor in chief of the Microprocessor Report.
Intel said its moves were tactical, not desperate.
``We have planned for this,'' said Bill Kirby, Intel's director of platform marketing. ``If we were truly just reacting to the market situation, we wouldn't have had these kinds of extensive engineering plans in place.''
Still, if Intel really had been prepared, it wouldn't have placed such emphasis on the 4-gigahertz chip, nor would it have dedicated a costly engineering team toward achieving that goal. At several events during the past year, Intel President Paul Otellini touted the fact that Intel would hit the 4-gigahertz milestone with its Pentium 4 chips.
In May, Intel made a similar gaffe and explained its strategy shift in a similar way. It canceled two chips in development and said it would accelerate the development of an alternate design that puts two processors on one chip. By contrast, AMD hasn't made the same kind of U-turns. And AMD's chips are so popular that AMD reported it gained market share in the third quarter in processor revenues.
Most industry observers agree that megahertz, or clock speed, is no longer a good way to measure a computer's performance. Making chips that sacrifice power efficiency and other features in the name of megahertz no longer makes sense, said Bernie Meyerson, chief technologist at IBM's Systems and Technology Group, in a speech last week at the Fall Processor Forum in San Jose.
Intel's Kirby said the new 3.8-gigahertz chip will likely be faster than the canceled 4-gigahertz chip would have been because of the extra memory cache. But the 3.8-gigahertz chip will come out relatively late, it may be more expensive to produce and it isn't clear if Intel will be able to charge equivalent prices for this kind of chip.
Jonathan Seckler, a product manager for Sunnyvale-based AMD, said Intel's move was a Band-Aid for a variety of problems. He noted that by adding more memory to its chips, Intel will make them bigger and more expensive to manufacture. A chip's size is directly proportional to its costs, since it takes more material to make a bigger chip. Seckler said chip size may be the real point of Intel's new design.
Intel invested nearly $20 billion in the past four years in 90-nanometer and 300-millimeter chip manufacturing technologies. That has given Intel a several-fold increase in capacity at the very time when sales of personal computers have slowed, said Michael Masdea, an analyst at Credit Suisse First Boston. On Tuesday, Intel said demand for the fourth quarter was relatively weak.
Making bigger chips
Intel has to do something with its extra capacity. Making bigger chips takes up more of a factory's output than making smaller chips. So Intel can make use of the excess capacity and churn out faster chips. The company said its decisions aren't driven by capacity alone.
The drastic changes that Intel is making suggest that, on one hand, it doesn't know how to plan its chip designs properly and it wasted engineering resources on the 4-gigahertz chip.
On the other hand, the company always leaves itself with contingency plans. If demand grows and the company's factories use more of their capacity, then Intel could hold off on the plan to include more memory in its chips. Since Intel can adapt its plans to changing market needs, it has more flexibility than AMD, said Intel spokesman Robert Manetta.
Changing course again, Intel on Thursday canceled its much-promoted 4-gigahertz Pentium 4 microprocessor and said it would switch to a new chip design.
The scuttling of the personal computer processor was another embarrassing move for the world's biggest chip maker and the latest in a series of missteps. The change in strategy is the result of big bets that Intel made over the past several years in both chip designs and expanding its manufacturing capabilities. The full significance of the shift remains to be seen.
To replace the discarded chip, Intel will create a 3.8-gigahertz chip with a double dose of memory in the first half of next year. This memory, known as cache, helps a processor run more efficiently by storing frequently used data close to the processor so that it doesn't have to retrieve data from the slower main memory chips.
If Intel is switching to a new design to make better use of its manufacturing plants, then the strategy seems more like a knee-jerk reaction to falling demand for its chips. But if the company has planned more deliberately, the new design could pay off in the long run.
``It's embarrassing to Intel because they talked about this 4-gigahertz goal publicly for so long,'' said Dean McCarron, an analyst at Mercury Research in Scottsdale, Ariz. ``On the other hand, it could be a non-event financially.''
Intel said it decided it could create chips with more built-in memory that didn't run at 4-gigahertz speeds but still delivered performance that would satisfy customers. The company also said it is designing chips that exploit manufacturing capacity that has become available more quickly than anticipated.
For Intel, the scuttling of the chip raises the risk that it might fall behind rival Advanced Micro Devices in the quest to provide the fastest processors. The move also means consumers won't get the fastest chips Intel had promised to make available for the holiday season.
Intel's detractors viewed the move as a stop-gap emergency measure to deal with the fact that its designs aren't as competitive as those of Advanced Micro Devices. AMD has begun to gain market share with chips that deliver more efficient performance than those made by Intel. Critics also contend Intel is switching gears to deal with a looming factory capacity problem.
Spin
``Everybody is putting their spin on this, like a presidential debate,'' said Kevin Krewell, editor in chief of the Microprocessor Report.
Intel said its moves were tactical, not desperate.
``We have planned for this,'' said Bill Kirby, Intel's director of platform marketing. ``If we were truly just reacting to the market situation, we wouldn't have had these kinds of extensive engineering plans in place.''
Still, if Intel really had been prepared, it wouldn't have placed such emphasis on the 4-gigahertz chip, nor would it have dedicated a costly engineering team toward achieving that goal. At several events during the past year, Intel President Paul Otellini touted the fact that Intel would hit the 4-gigahertz milestone with its Pentium 4 chips.
In May, Intel made a similar gaffe and explained its strategy shift in a similar way. It canceled two chips in development and said it would accelerate the development of an alternate design that puts two processors on one chip. By contrast, AMD hasn't made the same kind of U-turns. And AMD's chips are so popular that AMD reported it gained market share in the third quarter in processor revenues.
Most industry observers agree that megahertz, or clock speed, is no longer a good way to measure a computer's performance. Making chips that sacrifice power efficiency and other features in the name of megahertz no longer makes sense, said Bernie Meyerson, chief technologist at IBM's Systems and Technology Group, in a speech last week at the Fall Processor Forum in San Jose.
Intel's Kirby said the new 3.8-gigahertz chip will likely be faster than the canceled 4-gigahertz chip would have been because of the extra memory cache. But the 3.8-gigahertz chip will come out relatively late, it may be more expensive to produce and it isn't clear if Intel will be able to charge equivalent prices for this kind of chip.
Jonathan Seckler, a product manager for Sunnyvale-based AMD, said Intel's move was a Band-Aid for a variety of problems. He noted that by adding more memory to its chips, Intel will make them bigger and more expensive to manufacture. A chip's size is directly proportional to its costs, since it takes more material to make a bigger chip. Seckler said chip size may be the real point of Intel's new design.
Intel invested nearly $20 billion in the past four years in 90-nanometer and 300-millimeter chip manufacturing technologies. That has given Intel a several-fold increase in capacity at the very time when sales of personal computers have slowed, said Michael Masdea, an analyst at Credit Suisse First Boston. On Tuesday, Intel said demand for the fourth quarter was relatively weak.
Making bigger chips
Intel has to do something with its extra capacity. Making bigger chips takes up more of a factory's output than making smaller chips. So Intel can make use of the excess capacity and churn out faster chips. The company said its decisions aren't driven by capacity alone.
The drastic changes that Intel is making suggest that, on one hand, it doesn't know how to plan its chip designs properly and it wasted engineering resources on the 4-gigahertz chip.
On the other hand, the company always leaves itself with contingency plans. If demand grows and the company's factories use more of their capacity, then Intel could hold off on the plan to include more memory in its chips. Since Intel can adapt its plans to changing market needs, it has more flexibility than AMD, said Intel spokesman Robert Manetta.