Wednesday, 16 January 2013 14:21

Intel Vs. ARM; RISC against CISC all over again…

Written by

Reading time is around minutes.
Code

The rise of the ARM architecture has been a meteoric one and came as something of a shock to many of the mainstream players in the mobile industry. The concept for the ARM based SoC (System on Chip) is pretty simple: build a small, energy efficient device that is able to use memory and processor cycles as efficiently as possible.  This is what x86 CPU manufacturers are trying to do as well.  So why are there more manufacturers of ARM based SoCs than there are x86? Well, the problem here is one of licensing. You see, it is much easier and less expensive to license the ARM architecture than it is x86. Right now there are very few players that even have an x86 license. This limits the competition down quite a bit and also (unfortunately) slows progress. However, the fight between Intel and ARM is just heating up. And despite the small number of players in the x86 market, things are far from certain in this area as we enter into a new round of combat between RISC and CISC (Reduced Instruction Set Computing and Complete Instruction Set Computing).

The concept of the RISC CPU is an old one, and it is also a very efficient one. In RISC CPUs you have smaller workloads being run by each instruction, these instructions are also very often split into I/O and Memory to further eliminate overhead. This makes for very efficient use of CPU and memory time, but can also require some bulky code on the software side to make everything work. When RISC first was developed, it was the way to go simply because memory and HDD access was slow. The bulky and complex instructions found in an x86 CPU were cumbersome on older CPUs and could not keep up with RISC systems. This is one of the reasons we saw computers like the MAC and Compaq systems based on the DEC Alpha outperforming the Intel and AMD x86 systems on the market. It was not until Intel dived into the Pentium III and AMD pushed out the Athlon that we saw x86 truly outpace RISC in the workstation and server market. Although there are many reasons for this, one of the biggest was improved caching and memory efficiency on x86 CPUs. Once these items were “fixed”, the rest of the CPU could be tuned to compete with RISC and in short time the RISC CPU disappeared from mainstream computing.

Now in mobile devices (truly mobile devices like phones and tablets) you almost have to have a small efficient processor that can utilize memory, CPU cycles and power properly. The RISC architecture is perfect for this, as the size of the processor can be small. You do not require a ton of memory (RISC workloads are small and short), and you can optimize your OS and software to make the most of these two to conserve battery life. The use of RISC can also typically allow you to keep your transistor count down and reduce the size of your processor die, making these types of CPUs less expensive to manufacture. ARM and their partners have done a very good job of putting this philosophy into play when it comes to mobile devices; and when combined with their affordable licensing and widespread adoption in the market puts them in a very strong position indeed. In fact, their position is strong enough that they are working on invading the desktop and server space.
In retrospect, the assault on the server and workstation market was ill-conceived. ARM based systems simply do not have the power to compete head to head with Intel or AMD in this market, where you can virtualize operating systems and lower TOC exponentially by doing so. However, a few ARM manufacturers did attempt to make the argument that they were better and put up some numbers that were quickly debunked (simply because x86 servers are not used in single task mode anymore). This put ARM at a disadvantage and also allowed Intel (and to a lesser extend AMD) to jump into the mobile market with both feet.

Prior to ARM getting into the server market AMD and Intel were moving toward mobile, but at a much slower pace than they are now. Now, things have changed and Intel knows what to do to win this race. As they learned from their fight with PowerPC and DEC Alpha; Intel must commit to improving efficiency in terms of memory, cache, and power. Fortunately (for Intel), they have been working on doing just that ever since the embarrassing years of the Pentium 4 and RAMBUS. This commitment has yielded better caching, memory performance, and even the return of hyper threading as a viable feature. For the smaller form factors, Intel’s work on Atom has given them an insight into how to combine these higher-end features into small and energy efficient CPUs: this work has been integrated into Haswell, and as we saw at CES 2013, it will be a CPU to look out for in the future.

The ARM (RISC) Vs Intel (x86/CISC) war is going to be an interesting one moving into 2013. Both sides have learned much since their last encounter and ARM is the heavily entrenched incumbent this time. But make no mistake: Intel is hungry to grab a piece of the mobile market. They have the engineers, fabs, and other resources to make this happen. It might take a couple of years to truly get a good foot hold in this last bastion of RISC computing, but they will do it and it is only  a matter of time before they find the right mix of performance and power to attract OEMs and ODMs back to x86 for compatibility and performance. We have to wonder what the landscape will look like for RISC and ARM in 5 years and if they will still have a place in the market. Who knows, in 5 years ARM and other RISC systems might be looking to power a whole new family of products that x86 simply is not efficient enough to power and the cycle will start all over again.

What do you think about the Intel Vs. ARM war? Tell us in our Forum

Read 17166 times Last modified on Thursday, 17 January 2013 11:50

Leave a comment

Make sure you enter all the required information, indicated by an asterisk (*). HTML code is not allowed.