A monolithic semiconductor substrate

Chip Design Journal

Subscribe to Chip Design Journal: eMailAlertsEmail Alerts newslettersWeekly Newsletters
Get Chip Design Journal: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn

Chip Design Authors: Jnan Dash, Jason Bloomberg, Trevor Bradley, David Strom

Related Topics: Chip Design, AMD Virtualization Journal

Chip Design: Article

Only Linux Can Get Us From Here to There

Only Linux Can Get Us From Here to There

With increasing frequency, we are now seeing stories that have some mention of how a particular problem was solved only through Linux and its openness.

The Hollywood animation studio ILM admits they couldn't have done some of the really tough load balancing code used in their render farms without being able to modify the task scheduler code in Linux; and the Internet caching folks at Akami note how they were forced to choose Linux because nothing else could be made to do what they need – the irony of which is now many Microsoft web pages are actually served from Linux.

We're going to see more and more of these stories and today's announcement about a new chip that U. Texas and IBM are working on may be yet another of these stories eventually. The chip uses a new concept called “block-oriented execution." The technology could lead to a 500Mhz chip with 4 cores that each can perform 16 operations per cycle. The chip therefore can do 64 operations per cycle and, at 500Mhz would be able to perform 32 billion operations per second, in theory.

We don't currently know a lot about this chip in two important ways – 1) the facilities it provides for communication between the cores and 2) the methods it uses to execute 16 operations per core. For example, the Athlon executes 9 instructions per cycle and the P4 does 6 so bumping up to 16 instructions is quite a change and there's no saying that these instructions are apples to apples comparable. It's far too early to tell how optimal this chip design will be in the real world.

What we do know is that the software needed to take advantage of this environment is likely to be Linux or some other open OS. Linux already works on everything from wrist watches to super computers and has been moving with increasing speed into new markets including data centers, embedded devices, and cell phones. With Linux ported to so many different platforms, it's reasonable to expect that when an exotic new architecture arrives, Linux will be the first to support it - and perhaps the best as well.

The open source developer community demonstrated ability to rapidly port Linux to new platforms with the 64-bit Opteron from AMD. AMD announced specifications for the Opteron to the Linux community at the LinuxWorld Expo in August of 2000. Using that information, Linux developers set to work coding a kernel modifications and related utilities that allowed SUSE Linux to ship a 64-bit Opteron-compatible version of Linux in the same month as the Opteron launch in April of 2003. The comparable Windows solution remains months away.


What does this all mean?

In the future, when all the high-volume markets are dominated by open operating systems, chip architects should have increased flexibility to implement the designs that make sense. Designers will be able to assume that operating systems will keep pace. A processor like the one announced by UT/IBM today may still require new application software to take full advantage of inherent parallelism in the CPU but, in the general case, we can expect more flexibility in hardware approaches than we have seen in the past.

Linux won't tie us to the x86 platform like we have been and while this may shuffle the deck a little for Intel and AMD, it should eventually add fuel to a whole new round of processor innovation and advancement. And that could lead to at least one more round of revolutionary (as opposed to evolutionary) hardware price/performance improvement as fundamentally new and different technology floods the market.

It's an odd turn but open software not only looks like a better model for software development, it also opens new doors for better, cheaper hardware.

More Stories By Paul Nowak

Paul Nowak first used Linux in 1995 while migrating from Sun to Linux at the University of Michigan. He used Linux in subsequent IT projects including web, telecom, telemetry and embedded projects and is currently CIO of a small professional association based in Washington D.C.

Comments (5)

Share your thoughts on this story.

Add your comment
You must be signed in to add a comment. Sign-in | Register

In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.