G3's - they just keep getting better
Now the PowerPC's were ready to evolve again -- this was the 3rd generation of PowerPC's. After the highly successful second generation of PowerPC's, IBM and Motorola decided to split out development and create more processors. This family of processors became known as the G3. There are both low-cost/low-power versions and higher performance versions. The G3 was going to have both 603 and 604 derrivatives.
The first was the 603 derrivative. These processors got some changes to the core (the way it excecutes instructions) to optimize the processor for the Macintosh Operating System (1). This of course resulted in a large performance boost, above and beyond the performance boosts offered by the new backside cache.
(1) This is why the G3 is faster on the Mac that the 604's (in tests like MacBench and other application tests), eventhough the 604's are faster on theoretical benchmarks like Spec. Which just goes to show that specialized benchmarks (like Spec) can be too abstracted from the real world to hold much value.
At first the 604e's die shrink and speed up (the Mach5) was going to be called a "G3 Processor", since it used the newer manufacturing processes, and it was a fairly major performance bump. But since there was no change to the "core" of the processor, and it didn't use a new backside cache, this idea was scrapped.
The G4 project was progressing well (with a totally new core), Sommerset (the design lab for Apple/IBM/Motorola, now owned by Motorola) was working on multiple core versions of the PowerPC, and AltiVec was getting completed. So all these projects looked like they would come to fruition about end of '98 or early '99. The 603 derived G3 performed very well with its backside cache, and was very cheap to make and quite scalable by just adding more L2 cache (or faster L2 cache). Apple killed clones and focused the product lines, which all reduced demands for as many different high-end desktop PPC's. The end results being that the 604 derived G3's (code named Habanero), and some of the other flavors (like ones with better MP support (2)) were scrapped in favor of focusing on the G4's. Which makes sense, considering these other processors wouldn't be coming out until basically the same time as the G4's anyway, and you shouldn't split into that many different development efforts (waste of money).
(2) It is not true that the G3's don't support MP. They just don't support it as easily as they could, and there are some implementation details that may effect performance.
The G3 came out in two flavors (both 603 derived), the 740 and the 750.
The 740 (code named Arthur) is fast and extremely small and efficient. This tiny bugger is currently out-performing Pentium II's while using less than 1/5th the amount of power and size (power, size and cost are all related, so this is important). The 740, and has a similar layout as the current generation of PowerPC's to allow for replacement of current processors with the 740.
This 750 processor is a variant of the 740 that has a fast method of access to the L2 cache (called a backside cache) that allows higher performance. This L2 cache runs much faster than most -- and at speeds up to the clock rate of the main processor (or many different fractions thereof). This cach system really speeds things up, but it also requires more electronics (and pins) than the 740, so while the chip cost isn't much more, the added cache can drive the cost of the system up (and increase the total power usage).
Both processors offer excelent performance, at very low power, and even lower costs. Especially compared to comprable PentiumsII's, and even more so when you look at portable performance. These RISC processors are just crushing anything in the x86 world in the Portable market -- while the same processor is performance competitive with (or superior to) the most expensive desktop processors that Intel makes. But these processors are just a small stepping stone to the G4's, and may have a very short lifespan indeed.