Intel is conceding that a number attached to the word
Gigahertz
The Santa Clara, Calif.-based chip making giant revised its CPU numbering system for its various Pentium and Celeron brands last week. Intel said it will continue to use the Pentium M, Pentium 4 and Celeron brands, but starting with its upcoming
Dothan mobile chip, each desktop and notebook CPU will now have an additional three-digit suffix. The first digit indicates the CPU series – 3 for Celeron, 5 for Pentium 4, 7 for Pentium M on notebooks and Pentium 4 Extreme Edition on the desktop. The company is adding on second and third digits to indicate the features, including clock speed, bus speed, cache size and micro architecture.
Now any novice could tell you that a chip running at 4.0 GHz is faster than one running at 1.0 GHz, but Intel said higher-numbered processors in a lower series will sometimes outperform lower-numbered processors in the series above, while identical numbers for desktop and notebook processors may represent very different feature sets.
If it sounds confusing, don’t worry just yet. Analysts at Gartner Dataquest say while Intel’s new numbering system aims to differentiate value-added features, the changes are meaningless to most PC buyers and users. They will continue to focus on clock speed as the main CPU functional parameter.
“Intel’s aim is reasonable, but its execution is flawed,” Gartner said in a report issued to subscribers. “Higher numbers indicate more features, not more performance, and users will inevitably be confused. The numbers don’t explicitly reflect clock speed, and users will have to use look-up tables to interpret them. We believe some feature sets would be better represented by an alphabetical suffix, which could align with associated chip sets, and would also avoid numbers that suggest relative performance. We believe users will continue to focus on the clock speed, which will still be stated for CPUs, rather than the new, three-digit number. We predict that Intel will be forced to change the numbering to represent relative performance.”
In-Stat/MDR Microprocessor Report principal analyst Max Baron says Intel’s decision is also an acknowledgement that the high performance semiconductor industry has entered the system-on-chip (SoC) era and that “a product’s performance will not be measured in RPM of the engine but in how well the car performs in its intended market.”
“I believe that new criteria will replace or be added to raw frequency, which is comprised of style — of which Apple is quite ahead in this respect — size, noise, the number of processors in a box, peripherals, and frequency.”
Baron says the perceived performance of the internal core becomes “translated” to the outside world in performance terms that are “not a direct function of the core’s operating frequency.”
“The core’s internal microarchitecture, its supporting on-chip memory hierarchy, contention with other on-chip cores, external I/O, and the speed of external memory come into play,” he said. “System and application software will perform differently on different implementations that run at the same frequency. Power dissipation and temperature are becoming major factors influencing frequency of operation.”
Intel’s choice to drop the Gigahertz is very reminiscent of last year when AMD shifted its model numbering system to coincide with the release of its 64-bit Opteron processor. Under AMD’s model numbering system the first digit in the model number represents the maximum number of processors that can be supported in a system. For example, the company’s AMD Opteron processor 100 Series (i.e. 1XX) is designed for a 1-way server, the 200 Series is designed for a 2-way server and its 800
Series (i.e. 8XX) supports up to an 8-way server.
AMD also uses second and third digits communicate relative performance within each product line. AMD started numbering the last two digits at 40. In this case, the company says an Opteron model 244 will offer higher performance than an Opteron chip model 242. However, the company says the model numbers are not directly related to frequency speeds.
“This gives AMD flexibility to describe AMD’s server processor performance without potentially confusing end users by starting at 10, 20, or 30, because users might mistake ‘Model 224’ with a 2.4 GHz processor,” the company said in a statement last year.
So with less conventional names, does this mean that AMD, Intel, and IBM
could actually benchmark their similar processors in a more
open manner? Baron says no.
“I believe the customer will be better informed but, all other things being equal, will still make decisions based on brand names. And, published road tests of complete systems — not just the processors — will become important,” he said.
The name game has also been convoluted with the expansion into 64-bit processors for the desktop. AMD was first to jump on this by marketing AMD64 as its version of x86 architecture running on 64-bits (generically known as x86-64) for its Opteron and Athlon processors. Intel has taken a different path picking Intel Extended Memory 64 Technology — or Intel EM64T — as its moniker for its 64-bit extension technology (previously code-named Clackamas Technology or “CT”).
Intel said its EM64T will be included in its upcoming single-processor and dual-processor IA32 server and workstation products (code named Prescott and Nocona). The company said it will bundle several features into these chips including its Hyper-Threading technology, PCI Express, DDR2 support, enhanced power management, SSE3 instructions, high-definition audio, faster bus speeds, faster frequencies, enhanced security (LaGrande) and virtualization (Vanderpool) technologies.
To reduce confusion in the enterprise, Gartner said vendors should use the processor number to more clearly specify CPUs in purchase orders.
“Clock speed alone isn’t a sufficient specification, but should also be clearly stated,” Gartner said in its report.