With an odd sense of nervousness tied back to its mid-1990's Pentium debacle, microprocessor giant Intel on Monday delivered its next-generation Core i7 chips to customers. The i7 represents Intel's biggest architectural change since its mobile-oriented Core chips debuted almost three years ago. This time around, the base chips feature four processor cores, each of which is capable of executing two threads simultaneously, effectively providing the performance of 8 processors in compatible applications.
The Core i7 chips--codenamed Nehalem during development--have debuted to overwhelmingly positive reviews. But so did the original Pentium, which was quickly found to have a small but crucial floating point bug that necessitated a $400 million recall in 1994. Intel has completely overhauled the way it develops microprocessors since then, of course, but each new major architectural reworking triggers fears of another problem once the chips are out in the real world.
Intel's new chips are much more complex than the Core 2 Duo processors they'll eventually replace in the market, but they also provide dramatic performance benefits. There are several reasons for this improvement, including Intel's decisions to move the memory controller directly into the CPU die, bring back its Hyper-Threading technology from chips past, and speed the communication of data between each of the processor cores. The i7 can also selectively turn off up to three of its processor cores and throttle the remaining cores up to turbo speeds, increasing performance yet again.
What the current i7 really delivers is a baseline for future PC performance, those the initial chips, as always, are aimed at the workstation and server markets. The current first-generation chips easily break the 3 GHz barrier without any loss of power management efficiency compared to Core 2 Duo chips. But they can also easily be overclocked to above 4 GHz, suggesting even more impressive performance in the near future.