IT's a great time to be in IT

IT’s good to be in IT.

If you have been in the IT game for a while you might remember the hey-days of the Original Athlon, or the introduction of the Opteron. These were times when it was seriously exciting to be in IT. It seemed the world was at your feet and there was nothing you could not do with a computer. In more recent times the launches of the first consumer 64-bit operating systems and Multi-Core CPUs and GPU rigs have also caused this type of euphoric feeling.

But now, today we stand on the brink of a real revolution in technology. I can feel it, the signs are there. The PC (and Mac) as we know it is going to really change into something different. 

On the materials side we are seeing real advancements in the use of Carbon Nano Tubes (CNT) these small extremely thin tubes can be used to build interconnects inside CPUs.  Meanwhile Graphine [a single layer carbon molecule] can be used to build the transistors. These two materials can be used to build CPUs smaller and more efficient than ever before.  CNTs and Graphine do not have the leakage problems like coper, HKMG or other metals, they have a form of conductivity that has been dubbed “ballistic” this means that no electrons are blocked or released they simply flow from point a to point b.

On the architecture side, CPUs are taking over more and more functions of the mainboard. As of the launch of Lynnfield the PCI-e controller has found its way inside the CPU. In future generations we will see the actual GPU bolted next to the CPU core. This is only a predecessor to fully integrating the CPU into the core [as AMD has been talking about for years]. The reasons for this are varied but most boil down to efficiency. By removing the need to send graphical signals down the traces of a mainboard [and through other electronic signals] you lower the time it takes for them to be processed. 

Again GPUs on the CPU is just the first step, the next logical set is to being the rest of the PCI pipelines into the CPU. From there you can either pull almost all peripheral controllers into the CPU or create interconnects that allow you to shift them with each board you use. So the only thing you will need is for the board to identify the available peripherals to the CPU and you are off and running.

This type of architecture is already here although not on any scale like a standard desktop or server. These are smaller devices like nVidia’s Tegra. Tegra is a complete System on Chip design.  You have all of your core components shoe-horned directly into the same piece of silicon.  The idea is great and as we have seen from recent design wins and actual products released [Like the Zune HD] it really can handle the work in an efficient manner. 

Now Tegra is great and offers us a glimpse into the design of the future but as previously mentioned the manufacturing process is not there yet to do anything like this on a larger and more powerful scale. Even running at a future process of 28nm there is simply not enough room to do this economically. Now my talk of Graphine and CNTs makes more sense and fits in with what I have seen in the future of computing [the next 6-10 years]. Added into the equation is Kingston and RAMBUS working on ways to make memory as multi-thread efficient as possible. This new technique reduces latency and improves performance and power usage all in one swoop.

Still hardware is not everything in computing, we can have the best hardware out but if software is bulky and inefficient then the gains are lost. Enter the concept of “Singularity”. Here we have the missing piece of the puzzle. Singularity is a concept of a complete yet lightweight operating system kernel. This kernel is the foundation that allows other operating system layers to ride on top of it in much the same way that virtual machines sit in a layer above a hypervisor. 

Singularity is not a hypervisor though; it is something new and different. It is, as I mentioned, a complete kernel all on its own. It can exist without any layers above it but is more functional with then installed. You see in the end Singularity will allow for “virtual” operating systems to ride over the kernel and have direct connections to hardware through it. This gives the possibility of having multiple live operating systems running at the same time without the normal performance hit you would see with traditional virtual machines.  It breezes by the complications of application compatibility by allowing them to run in their own space as needed.  You could potentially have these running under a unified desktop so that all are visible as their own instance on a single GUI yet each is running in their own virtual OS underneath. This will be a massive leap forward for computing once this is brought to the consumer.

So today we can see the glimmerings of the future in products like Tegra, VMware’s Unity, Intel’s Lynnfield and Clarkdale CPUs, AMD’s Fusion, Graphine and CNTs and Microsoft’s Singularity project.
If you combine these trends and look at where they can take us, you will understand why the next 6-10 years will be some of the most exciting in the history of computing.