X

HP ponders mysteries of dual-core

Stephen Shankland Former Principal Writer
Stephen Shankland worked at CNET from 1998 to 2024 and wrote about processors, digital photography, AI, quantum computing, computer science, materials science, supercomputers, drones, browsers, 3D printing, USB, and new computing technology in general. He has a soft spot in his heart for standards groups and I/O interfaces. His first big scoop was about radioactive cat poop.
Expertise Processors, semiconductors, web browsers, quantum computing, supercomputers, AI, 3D printing, drones, computer science, physics, programming, materials science, USB, UWB, Android, digital photography, science. Credentials
  • Shankland covered the tech industry for more than 25 years and was a science writer for five years before that. He has deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and more.
Stephen Shankland
2 min read

Dual-core processors are arriving in the mainstream, bringing new uncertainties about how to price software sold on the basis of how many processors it's running on. Now a Hewlett-Packard executive has chimed in with an opinion that agrees with that from Sun Microsystems, Microsoft, Intel and Advanced Micro Devices--a processor should be defined as that which fits into processor socket--but that suggests eventually sidestepping the issue altogether.

Oracle is on the other side of the dual-core debate, preferring instead to charge on the basis of processor cores. That's likely to make other company's software more competitive on chips such as Sun's upcoming Niagara chip, which has eight cores, or on forthcoming Intel chips with four and eventually more. IBM is conflicted about the idea, calling dual-core x86 chips such as AMD's newest Opteron a single processor but its own dual-core Power5 two separate processors.

It's not surprising there's confusion: Those who sell chips and servers like to tout the improved performance of next-generation processors, while software companies generally are inclined to charge as much money as possible.

Rich Marcello, head of HP's high-end server group, ponders the pricing conundrum on his blog and concludes per-socket pricing is the way to go for now for now.

"Having worked on microprocessor designs for the first ten years of my career, I tend to think of the number of cores as an implementation detail in order to stay on the current performance improvement curve," he said. In other words, adding cores is just the next step in making chips faster. And pricing per-socket sidesteps the tricky issue of another potential pricing complication called 'multithreading' that lets a single core look in some ways like more than one.

It's not an easy decision, though: "Even with these relatively strong arguments, I'm not sure I'm right about this. There are some good arguments on the other side as well," Marcello adds.

Ultimately, cores and threads probably are beside the point in the long run. "Instead of thinking about processors as physical entities, maybe the right model is to define processor units virtually and to charge customers according to a use-based algorithm," Marcello said.

Sun is a couple steps farther down that road. It wants to sell software based on how many employees a company has--no tedious usage metering, much less processor counting.