Why Intel's new chip names are so bad

Why Intel's new chip names are so bad

Summary: commentary Speed isn't a measure of speed. That's Intel's message with its new naming scheme for its mobile CPUs.

SHARE:
Rafe Needleman commentary Speed isn't a measure of speed. That's Intel's message with its new naming scheme for its mobile CPUs.

While it's true that clock speed tells you less than it used to about how well a particular chip will perform, this new scheme will likely confuse buyers more than it helps. Here's Intel's justification for it: According to Intel VP Anand Chandrasekher, speaking at the launch of the Dothan (a.k.a. 700 series) CPU, the new numbers are not about performance so much as "goodness."

When I buy a car, I don't want to know how good the manufacturer thinks the engine is. I want to know how powerful it is--how much horsepower it has. Even BMW's cryptic naming scheme makes more sense than Intel's new plan, because the last two digits in most BMW model numbers indicate engine displacement in (a BMW 325 has a 2.5-litre engine), which is a pretty good indication of how much oomph the car has. By comparison, an Intel-series 745 chip runs at 1.8GHz. Do you see the correlation? Me neither. Because there isn't one.

Of course, with car engines, we care about linear things -- horsepower and torque. With CPUs, we care about performance when playing games and performance when processing videos, and top-of-the-heap performance in one doesn't necessarily mean best-of-class in the other. But even so, fast is fast. What Intel is doing with its new scheme is taking into account other factors that have nothing to do with speed. For example, Chandrasekher says the new scheme takes into account power draw (and thus battery life).

This is a good thing to know, but it's like BMW naming its cars based on fuel economy.

Intel's new scheme also considers things such as cache size and bus speed, which do affect performance. And Intel has to do this. The problem, if it can be called that, is that the new Pentium M chips run faster than older Pentium Ms with the same, or even faster, clock speeds. So Intel needs to attach bigger numbers to the new parts to telegraph to consumers that they are faster.

This scheme also appears to be a shot at AMD, Intel's chief competitor. AMD's chip-naming scheme is grounded in its competition with Intel. For example, an AMD Athlon 64 3200 is sold as a direct competitor to an Intel Pentium 4 running at 3.2GHz -- even though the AMD's internal clock speed is 2.2GHz. With Intel's new series-based naming scheme, AMD will have to figure out some new way to compete.

Ultimately, just as engine displacement really doesn't equal horsepower (some engines wring more power out of less, and the number of cylinders and turbochargers and things like that throw off any displacement rating), because of the large impact of things such as the size of a processor's cache, CPU processor speed is meaning less and less in terms of performance. And Intel is working on dual-core processors, which will essentially (in theory) double performance at a given clock speed.

So what we need is an agreed-upon measure of processor performance. It doesn't have to be exact. Car engines are rated in horsepower, a performance measure the public understands, even though a car's overall performance will vary depending on a car's total weight and many other factors. Chips can be benchmarked on simple or complex tasks--from a raw number such as MIPS (millions of instructions per second) to something more real world, such as the BAPco SysMark tests we run on desktops and laptops.

I'd like to see the chip companies adopt a performance number from such a standardised test. It won't tell you precisely how well a chip will perform, but it will put you in the ballpark, just like a horsepower rating does. And it'd be a lot easier to decipher than an arbitrary number that seems destined to confuse buyers more than help them understand what it represents.

For more information on how these chips really perform, see our Tech Guide on Intel's new Pentium M processor.

What do you think? Is Intel's new naming scheme helpful to you? Or is it confusing? TalkBack to me below!

Topics: Processors, Mobility, Networking

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

4 comments
Log in or register to join the discussion
  • People shouldnt care about engine displacement. The figures they are actually interested in are primarily Power to Weight, Fuel Consumption, and some sort of acceleration measurement 0-60 in ...

    Saying a car has a 3.8L engine does not mean it out-performs a car with a 2.5L engine, however all things equal, there is a correlation between displacement and performance.

    Likewise, there may be little performance difference between two CPUs with a difference clock speed. In fact, I have seen a 2.8GHz outperform (by a significant margin) a 3.2 GHz CPU in a hyperthreading demonstration.

    A lower clocked P4 will outperform a Celeron on certain tasks because the P4 will have more L2 cache, but on other tasks, the Celeron will win out.

    I agree a better index is needed then GHz, but what? Benchmarking is fine but the losing chip will always claim foul. How can a person wanting to run a web browser and word possibly use the same performance index as a person wanting to do video encoding? What relevance does a chips floating point division speed have for a user only interested in word? What relevance is a benchmark without such measurements to someone encoding videos?
    anonymous
  • What's in a name? A rose by any other name would smell as sweet! Well, that's ok, but only so long as the name doesn't relate to a new Intel processor, or so it would appear.

    Here's a question for Intel's CEO. Perhaps you'd like to pass it on. "How is one supposed to guage a processor's performance from an apparently, completely unrelated code number?"

    Gee! Hope they don't lose sales over this blunder....
    deando_au
  • The example of the BMW naming system, while appropriate, is not entirely accurate. Mercedes uses a similar system, but like BMW, the numbers don't necessarily equate. The BMW 320i has a 2.2l six, the 545i has a 4.4l V8, Merc S350 has a 3.7 V6, the SL65 AMG (performance branch of Merc) has a 6.0l twin-turbocharged V12. However, this is just the engine: a bigger, more powerful engine does not mean a better car at all. It can, but generally doesn't.

    My computer has an AMD Athlon XP 2800+, and works very well under my general usage. A friend of mine has a virtually identical spec PC but with a 2.8GHz HT Pentium 4. Under a benchmark test I found, my system turned out to be 8% faster. And the 2800+ has a clock speed of 2.083 GHz. Which is the superior processor? I think we can tell. Oh, and it's significantly cheaper, too.
    anonymous
  • I look at what I'm going to use the pc for. I don't play games. Just suft the net and do word processing etc.

    So to have a whiz bang processor and heaps of ram are of little importance to me. I prefer to know that the processor, ram are going to work with sata and other features. Having a larger cache is just as important. Why have a fast processor if it has 128kb of cache. Would it not be better to have a 512 or 1mb cache and a bit slower clock speed?
    Then what about hyper threading and dual processors again you have to factor this in.
    No I think raw spped is not an indicator.

    Cars can go to 220 km/h, but you can't use it on the surburban road.
    anonymous