Home » Opinions » 40 Years and Counting » 40 Years and Counting

40 Years and Counting


Moore's Law and Silicon Scaffolding

Though he admits that 10GHz CPUs may not be reached anytime soon, Pat still sees some healthy progress in the advancement of Moore's Law. You know, the one that says the number of transistors on an integrated circuit will double approximately every two years.

Pat explained that twenty years ago '1 micron was challenging and 100nm looked impossible'. However, 'now we casually talk about what it takes to get to 10nm'. He also goes onto use rather a neat analogy to explain how the 45nm chips we see in the shops now relate to what Intel is working on in its labs that we'll see in the future. '{it's} similar to driving down a road on foggy night where you are able to see 100 yards and as you cross the first 50 yards, the next 100 yards become clear'. Right now we're at 45nm and the 100 yard mark is 10nm. Neither Intel nor we can speculate on what will come next, merely that a few years down the line things will have progressed another 50 yards and the road ahead will become clear once more.

He also goes onto explain that increasing wafer size is of paramount importance to progressing economically and that we'll soon be seeing wafer sizes increase to 400mm (the current standard is 300mm) and even 450mm in the coming years. The consequence of foundry's increasing wafer size will also 'drive further industry structure changes' or in other words fewer and fewer, larger and larger foundry's will be needed to produce silicon at the right quantity and quality.

Tera-age of Computing

Pat admits that the market for semiconductors is in a strange period of transition at the moment. With so little software taking advantage of multi-core processors and single threaded speeds having hit a ceiling of around 4GHz, the benefits of upgrading your CPU are limited and there's little Intel can do about it at the moment. What is needed is for 'software programmers to take advantage of parallelism' ushering in a whole new era of software such as 'search{ing} across model based data, visual computing applications {and} physics, etc.'

For what it's worth, Intel is doing its darndest to push things forward with development of specialist multithreaded compilers and programming environments but the point at which the benefits of these developments filter down to the everyday user still seems to be some way off. Indeed the whole terascale concept is one that Intel has been banging on about for a while now but we've yet to see any results. We still hold out hope, though.

Power of Compatibility

Taking inspiration from lessons learned in the past Pat also sees the importance of maintaining compatibility into the future with the ultimate aim being 'to deliver a solution for everything that computes'. Essentially, Intel would like the world to use its chips for everything, 'literally from the very smallest platforms such as MIDs and phones to the very largest super computer' and ensuring compatibility between all these platforms is essential in encouraging uptake in all areas of the market.

IA Everywhere, 24 7, the Modality of Life

Moving on from wanting all computing devices to use Intel Architecture, Pat would also like to see 'internet connectivity for every modality of life'. By enabling more and more devices to communicate in sophisticated ways we may finally get to that utopian goal of having devices that talk to each other, telling themselves what they need rather than needing us lowly humans to get in the way. Lights will dim themselves, power grids will self manage and everybody will live in peace and harmony. At least that's the theory.

In fairness Pat concedes that 'we are only 4 or 5% complete with our task' and 'there is a lot of growth left to do' so he's not completely away with the fairies on this one. Certainly it would seem, though, that there's plenty of new and exciting developments in the pipeline that will keep this stalwart of the technology industry innovating for another 40 years.

comments powered by Disqus