Pentium III launched Feb 28, 1999

26 years ago this week the Pentium III launched. It was noteworthy for being the CPU that broke the gigahertz barrier, but also for being a better chip than its successor. The Pentium 4 clocked higher, but a Pentium III at 1.13 GHz outperformed a Pentium 4 at 1.5 GHz. It wasn’t really until the Pentium 4 doubled the speed of the Pentium III that the P4 became a good CPU.

Competing with the Pentium III

Not only did Intel have trouble keeping up with the Pentium III, its competitors did too. AMD competed successfully with its original Athlon. Cyrix and Transmeta, on the other hand, both struggled.

Early Pentium III CPUs were very similar to the earlier Pentium II-branded processors. There were two notable differences. The first was the addition of the Streaming SIMD Extensions (SSE) instruction set to accelerate floating point and parallel calculations, taking MMX to a new level. The second, less desirable difference was the introduction of a serial number embedded in the chip during manufacturing. Some privacy-minded individuals swore off Intel at this generation and switched to AMD.

Even after the release of the Pentium 4 in late 2000, the Pentium III stayed in production with new models up until early 2003. Intel discontinued desktop P3 CPUs in April 2004 and mobile P3 CPUs in May 2007.

The P3 had four generations of desktop CPU: Katmai, Coppermine, Coppermine T, and Tualatin. The mobile version had two generations: Coppermine and Tualatin.

Katmai

Katmai had 2 million more transitors than a Pentium II and initially ran at 450 and 500 MHz running on a 100 MHz bus, matching the Pentium II’s top speed and exceeding it by one speed grade, respectively. Later revisions reached a maximum speed of 600 MHz running on a 133 MHz bus. They used the same Slot 1 connector as the Pentium II.

Coppermine

Coppermine ran at speeds ranging from 500 MHz to 1 GHz, on a 100 or 133 MHz bus depending on the model. It included an on-chip L2 cache, giving it a futher performance edge over Katmai. Initially Coppermine outperformed an AMD Athlon, but die shrink and moving L2 cache on-die gave the Athlon the edge. The Athlon also was able to reach a speed of 1.2 GHz, which the Pentium III never reached.

Intel produced Coppermine in both Slot 1 and Socket 370 packages, although the fastest models were only available in Socket 370.

Intel did release a Coppermine running at 1.13 GHz in mid-2000, but it didn’t run reliably. Tom’s Hardware and HardOCP collaborated on an expose around the problems with the 1.13 GHz Coppermine and its stability, or lack of it. Intel withdrew the 1.1 and 1.13 GHz Coppermines and released revised versions six months later.

The Coppermine T was a revised Coppermine that didn’t include any new speed grades.

Tualatin

Tualatin, the final revision of the Pentium III, ran at speeds ranging from 1 GHz to 1.4 GHz. It wasn’t fully compatible with earlier Pentium III motherboards. It was a process shrink from Coppermine, gaining speed from the process shrink but the only new feature was a tweak to the L2 cache to potentially improve performance slightly.

The privacy controversy

The Pentium III was the first x86 CPU to include a unique, retrievable identification number, which Intel called a Processor Serial Number (PSN). Software can read a Pentium III’s PSN through the CPUID instruction unless the end user or system manufacturer disable the feature through the BIOS.

This measure was controversial, in 1999, the European Parliament considered legal measures that would prevent these chips from being installed in the computers of European citizens.

Intel removed the PSN feature from Tualatin-based Pentium IIIs and omitted it from the later Pentium 4 and Pentium M.

Intel quietly added back a largely equivalent feature, the Protected Processor Identification Number (PPIN) to later x86 CPUs with little public notice, starting with Intel’s Ivy Bridge architecture 3rd-Generation Core i3, i5, and i7 processors. AMD’s Zen 2 CPUs have equivalent functionality.

Dave Farquhar

David Farquhar is a computer security professional, entrepreneur, and author. He has written professionally about computers since 1991, so he was writing about retro computers when they were still new. He has been working in IT professionally since 1994 and has specialized in vulnerability management since 2013. He holds Security+ and CISSP certifications. Today he blogs five times a week, mostly about retro computers and retro gaming covering the time period from 1975 to 2000.