| « Front Page Extensions for Linux | Dedicated unmetered 100Mbps » |
January 15th, 2007 [by Doug Alder]
There is a paradigm shift taking place in the way computer and server CPUs are built and in the way you, the consumer need to think about them. In the past the focus has been on ever higher speeds achieved through faster GHz processors, the higher the number the faster the processor. However Moore’s law and basic physics are finally catching up to that way of doing things. The faster you make the core the more heat it generates and the harder it becomes to disperse that heat. To make the core faster you need to pack more transistors in a smaller space which means thinner layers in the die which in turn leads to problems with power leakage, data corruption, data synchronization, and heat dissipation. The same holds true for scaling down transistors to the nanoscale level (and eventually to the atomic level where you transistors made up of just a couple of atoms (or maybe a single atom of an exotic compound)
One of the major challenges in integrated circuits that use nanoscale transistors is increase in parameter variation and leakage currents. As a result of variation and leakage, the design margins available to do predictive design is becoming harder and additionally such systems dissipate considerable power even when not switching. Adaptive and statistical design along with leakage power reduction is critical to sustain scaling of CMOS. A good treatment of these topics is covered in Leakage in Nanometer CMOS Technologies. Other scaling challenges include:
- The ability to control parasitic resistance and capacitance in transistors,
- The ability to reduce resistance and capacitance in electrical interconnects,
- The ability to maintain proper transistor electrostatics that allow the gate terminal to control the ON/OFF behavior,
- Increasing effect of line edge roughness,
- Dopant fluctuations,
- System level power delivery,
- Thermal design to effectively handle the dissipation of delivered power, and
- Solve all these challenges with ever-reducing cost of manufacturing of the overall system. (more…)
The added complexities of trying to continuously go for faster single cores must drive the cost of making those cores up exponentially.To counteract this direction, both AMD and Intel have switched their main focus to multi-core CPUs. Each processor includes two or more complete cores per physical processor, enabling server platforms to execute more tasks, software threads and/or applications simultaneously. RackForce has opted to go with Intel’s implementation of this technology as you can see from our extremely popular 5100 Series of IBM Dual Core servers.The pace of progress in multi-core chips is nothing short of phenomenal. Less than a year ago the dual cores came out, now we have quad cores about to make the scene and RackForce will be one of the very first data centers anywhere to have Intel’s new quad core CPUs and the first to have IBM’s implementation of that technology in our new 5300 series of servers coming online this week. Already Intel is working on an 80 core chip due out in 4 years.More cores means more work done in less time. A quad core processor running at 2GHz (yes that right just 2GHz) is up to 4 times faster (depending on the process) than a single core 2GHz CPU. That’s the equivalence of an 8GHz single core CPU, something that doesn’t as yet exist.That’s what RackForce is all about. Bringing you the best so you can be your best. RackForce brings you advanced technology so you can get your work done more efficiently, whether that work be website hosting, streaming media, database crunching or any other application for a server you can imagine, we are here to help make it possible for you.
