Why aren't processors getting faster?

[Tiby]

The MHZ race ended long time ago (when dual core CPUs first entered the market). Then it was a race against multiple cores, each running at decent speed to provide enhanced performance.

Unfortunately software did not follow this evolution path. Few are the applications that indeed make use of multiple cores (besides professional ones such as Adobe suite, Autodesk ..).

To make it worse, mobile development started to gain ground with the announcement and release of the iPhone and android. Since then fewer and fewer apps were released for windows desktop platform, instead nearly all of them made a switch to either cloud computing or mobile.

Desktop apps weren't making developers money unless they were targeted to a specific business model, thus this switch was actually a very good one from their perspective. Having a monthly fee subscription for hosting your files or in game currency exchange for mobile games, was more profitable than anyone can imagine.

For this reason Microsoft silently tried to kill the desktop, giving existing developers to build mobile apps with the tools that they were already using. Not a bad ideea, but they were too late to the party.

The mobile development is so "in hype" nowadays that CPU builders that were long forgotten started releasing new products which focused on low power efficiency (anyone remember Via Chrome Chips)?

 

What could Intel do in this situation? Enhance the cores and overall core performance of CPUs or concentrate most of their efforts on lower power CPUs that could very well run on a tablet with decent battery life? 

They did both mind you  , however the first option will come at a premium price.

AMD followed the same route, more or less. APU's, low powered chips on AM1 platform that can be used on a mini ITX form factor, custom SOC's that were already integrated in consoles and the announcement of tablet pc's that will use their graphic capabilities are a confirmation of their direction. They knew this day will come since they aquired ATI.

[harrynowl]

The GHz war pretty much ended with Netburst and Cedar Mill. AMD had K8 which had far superior IPC and far superior power consumption characteristics. AMD then moved to make dual cores (Athlon 64X2) which meant Intel kind of strapped a 2nd netburst core to the die and it was called the Pentium D.

 

Intel also had Pentium-M on the mobile side, but they noticed that at much lower clock speeds it put out the same performance numbers as netburst. 

 

So basically since then Intel realized that AMDs strategy of improving the core rather than chasing clock 11GHz speeds was the better one and started working on the original Core which was based on Pentium-M.

 

Improvements from Sandybridge to Haswell have become marginal because as you increase complexity, it gets ever more difficult to gain performance. Not to mention most of Intels R&D is bent on getting into the mobile market which was pretty much taken by ARM.

[GeathX]

To be honest I don't know what they need to do to get back in front,  I can't help but wonder sometimes if the divestiture of global Foundries was their best move.  I know they were up against a wall however one of intels strong points is it owns all it's own manufacturing. 

 

Some advertising wouldn't go astray. I don't think I have ever seen an AMD advertisement other than a faint memory of a billboard in GTA3.

 

 

Improvements from Sandybridge to Haswell have become marginal because as you increase complexity, it gets ever more difficult to gain performance. Not to mention most of Intels R&D is bent on getting into the mobile market which was pretty much taken by ARM.

 

I don't completely buy that as the only reason of slowed progress. If AMD was actually trading blows with the i7 ( rather than very few specific situations where an 8350 can almost keep up ) Intel would be much more motivated to bring substantial upgrades to the CPU market. This hasn't really happened since the phenom II era. where AMD were rather close to catching up, but have stumbled a bit since. Hopefully having the console contracts will provide a injection of dependable funds to push CPU design. 

[Faa]

I tend to feel that Intel might have actually hit a wall.  It is getting increasingly harder for them with the limitations of manufacturing lithography.   Either way though, it is sad AMD haven't been able to keep up.  Competition always helps.

If they hitted the wall, plenty of time for AMD to get back on track, no?

Anyways if CPU's are going to evolve in terms of performance like this and gpu's getting massive performance improvements per gen, I wouldn't surprised over 10 years orsomething that any cpu out there would be a bottleneck for any gpu. Doubt Intel would do anything about it, they'd be just waiting for AMD and being criticized for making shit cpu's as well. The wan show with that engineer from Intel was complaining about games being crappy threaded or running awful, stupid comment seriously your own company is the only one to blame for having shit IPC. CPU monopoly being all about IPC is just plain silly.

[harrynowl]

If they hitted the wall, plenty of time for AMD to get back on track, no?

AMD hit a different kind of wall, the one that involves losing all your R&D money

[Humbug]

Physics; I think Intel and AMD are battling against physics and the nature of silicon when trying to make CPUs faster and faster. For the last few decades thay have been getting faster and faster, now we are hitting a wall finally.

 

The difference with GPUs / video cards is that they are all about parallelism. So every generation as we get to smaller process nodes AMD and Nvidia have been able to put more and more resources into these monster chips with massive parallel performance. We haven't hit the wall yet when it comes to graphics.

[Timopomer]

Intel and nvidea has to be fought off at once,its not so easy sometimes