Image source: Intel.
Earlier this year, chip giant Intel (INTC 0.07%) disclosed to investors that it had axed its "tick-tock" product development methodology in favor of a new one known as "process-architecture-optimization".
Until Intel formally unveiled its seventh-generation Core processor family -- the first "optimization" that Intel is bringing to market -- it wasn't clear what this development methodology really entailed. However, the details behind this strategy are now much more apparent.
Although it's a reasonable change, I believe that Intel would be best-served with additional investments to make it even better. Allow me to explain.
Understanding the current strategy
Under Intel's process-architecture-optimization development methodology, here's what the company is expected to deliver with each step:
- Process: This is a direct replacement for the "tick" under Intel's previous methodology. What Intel delivers here is a slightly enhanced processor core, as well as a new graphics processor manufactured on a significantly denser manufacturing technology than the product that came before it.
- Architecture: And this is a replacement for the "tock". In this step, Intel launches a "new" processor core with substantial performance and feature enhancements over the prior-generation. Intel is also expected to deliver a new graphics processor architecture in this step, along with other significant platform feature enhancements. It is likely that, going forward, Intel will build this chip on a performance-enhanced version of the manufacturing technology that made its debut in the "process" step.
- Optimization: Based on what Intel delivered with its seventh-generation Core processor family, it would appear that here, Intel takes essentially the same architecture that came in the previous step and implements it on a performance-enhanced version of the manufacturing technology that was used in the "architecture" stage.
With this methodology, Intel can deliver new and better products each year, so in that respect, it is a clear win. However, there is room for Intel to enjoy further improvements -- if it's willing to make the required investments.
Optimization could enjoy architectural benefits
During the optimization stage, Intel's products see performance benefits associated with the new manufacturing technologies. In effect, the company can run these chips at higher speeds while keeping power consumption the same. But it might make sense for Intel to try to be more aggressive on the architectural side of things here, too.
Remember that under "tick-tock", Intel's architecture teams planned for all-new CPU architectures every two years and new graphics architectures every three years.
Indeed, it's not as though the "tick-tock" methodology was derailed by execution problems on the part of the architecture teams. Indeed, the problems are clearly related to the inability of Intel's manufacturing group to make generational manufacturing technology leaps at the same speed they used to be able to.
So, going forward, Intel would probably be better served by adopting the following methodology:
- Step 1: Move to an all-new manufacturing technology, tweak the CPU cores a bit, and introduce a new graphics processor. This would be analogous to a traditional "tick" under the prior methodology (in other words, think Intel's third generation Core processors and its fifth generation Core processors).
- Step 2: Move to a performance-enhanced version of the manufacturing technology in step 1 but also deliver significant generational leaps in both CPU, graphics, and platform architecture more than is typically expected. This would basically be the same as a "tock" under Intel's previous methodology (think fourth generation Core and sixth generation Core processors) but with the added boost from a performance-enhanced version of the manufacturing technology.
- Step 3: Move to an even higher-performance variant of the manufacturing technologies in step 1 and step 2, tweak the processor cores a bit, and introduce a new graphics architecture. This would also be like a "tick" under Intel's previous development methodology, but the main difference would be that the manufacturing technology delivers better performance (which is all the end customer really cares about) but doesn't deliver the cost reduction that comes with a smaller manufacturing technology.
Under such a methodology, Intel's core intellectual property teams don't have to change the pace at which they introduce new technologies, but the performance enhancements that the chipmaker will be able to deliver with each generation, and cumulatively over a three-year/generation period, would be much better.
This would ultimately mean better products and stronger competitive positioning for the chip giant in the years ahead.
