There is a battle being fought at this very moment over the mighty megahertz measuring stick used by both Transmeta and Intel. Each company uses this one aspect of measuring a systems performance to help convey the performance of their products to the end consumers.

At face value this would seem to be an infallible method of comparison. Megahertz is a number (or more correctly a unit of measurement) so it either is, or isn't accurate. There is no need, nor is there room for interpretation - a processor either operates at a specific value of MHz or it does not. Right? Well not always...

For instance, right now I'm typing this article on a TM5600 Crusoe powered NEC notebook that I know for certain is operating at 600MHz (or more precisely 592.66MHz). Transmeta's power management utility, Long Run is on, but I've configured it so that the notebook operates at full tilt. The program itself gives me a visual clue as to the general operating characteristics as I type each key, or move the cursor around. But to really drive the point home that this computer is operating at 600MHz (or precisely 592.66MHz) I need only run a program called WCPUID/CLK - a real time clock checker.

I can run this test with the NEC notebook running on its internal batteries, or while connected to the wall via the AC adapter and the results will be the same. I can even run this program on my old 366MHz Celeron based Acer notebook to see that it is indeed running at 366MHz, whether plugged into the wall or while operating on its batteries.

Where recent conflicts have begun to arrive are with Speed Step enabled Intel mobile Pentium III processors.

As with most conflicts in the tech industry, this one is completely marketing driven and has to do with the way some notebooks are represented to the consumer, and specifically how they compare to Crusoe-based notebooks.

Intel offers a range of fixed speed, mobile PIII processors that operate at between 400MHz and 500MHz. All other current mobile PIII processors operate with a technology called Speed Step. In Intel's own words, what Speed Step enables the mobile PIII to do is "operate in two performance modes - maximum performance mode for near desktop performance, and battery optimized performance mode for lower power consumption and improved battery life."

Essentially, when the notebook is in "maximum performance mode" (such is the case when connected to AC power) the processor by default operates at the full clock speed. When the notebook is operating in mobile mode, or "battery optimized mode" (such is the case when relying only on the internal battery), it will by default, operate at a lower clock speed to conserve electricity - a la Speed Step. See the conflict?

When the notebook is operating in "battery optimized mode" (mobile mode) is when it is actually acting like a notebook, but at a substantially lower clock speed than it is generally referred to. Of course, a notebook with a mobile PIII can have Speed Step manually overridden so it will operate at the "maximum performance mode" while just on battery power, but this conflicts with Intel's notion of mobile mode, and comes with a sacrifice to battery life.

If you're confused don't worry, there is a lot of emphasis being put on word play here. Hopefully the following chart will clear things up, and underline the difference in default clock speeds for Speed Step enabled PIII processors operating under AC and battery alone.

Processor	AC power	Battery
Fixed-Speed PIII
PIII	500MHz	500MHz
	450MHz	450MHz
	400MHz	400MHz
Normal PIII Mobile
PIII Mobile	1.26GHz	800MHz
	1.0GHz	700MHz
	900MHz	700MHz
	850MHz	700MHz
	800MHz	650MHz
	750MHz	600MHz
	700MHz	550MHz
	650MHz	500MHz
	600MHz	400MHz
Low Voltage (LV) Mobile PIII
LV PIII	866MHz	533MHz
	750MHz	500MHz
	700MHz	500MHz
	600MHz	500MHz
Ultra Low Voltage (ULV) Mobile PIII
ULV PIII	700	300MHz
	600	300MHz
	500	300MHz

Of this list, the LV and ULV Mobile PIII processors are most directly comparable to Crusoe in terms of the applications and types of notebooks they will find themselves into. Generally speaking the ULV, and LV types are destined for sub 6lb notebooks and subnotebooks. The vast majority of the normal PIII mobile processors find their way into 6lb or heavier notebooks and are not directly competing with the Crusoe processor market. Regardless of type or die size (0.18 or 0.13 micron) each Intel processor equipped with Speed Step operates by default, at a lower "battery-optimized mode" clock speed.

Additionally, since Transmeta's Crusoe processors are equipped with an onboard north bridge chipset, directly comparing the power consumption of the Crusoe chips to that of the Mobile PIII is somewhat inaccurate, especially when notebook power consumption is concerned. Ideally, this type of notebook level comparison would see Crusoe pitted against the PIII processor and its' north bridge chipset. That chipset alone can account for anywhere between 1-3 extra Watts of power consumption in the Intel system.

As we said, the differences are primarily marketing driven, and the terms of conflict are almost to the level of legal specifics, but if you are seeing comparisons between these two processors it is important to keep these points in mind. After all, if notebook is meant to be used as a mobile platform then Speed Step seems to be more of a handicap than an advantage. Sure it scales the processor back into a "battery optimized mode," but this also scales back the performance of the machine as well. With that in mind, there are a couple of options for the consumer to consider; side with Crusoe-based notebooks that operate at their listed frequencies, or go for a 850MHz mobile PIII and save a few bucks over the 1.0GHz model - both scale back to 700MHz when in "battery optimized mode."

In this regard, Speed Step appears to function as both a technology for Intel-based notebooks and as a marketing tool. While it would be nice to see the notebooks classed by only their mobile speeds this seems unrealistic. Faced with no other alternative it is good to remember that in terms of power consumption and MHz, the differences are not always what they appear to be at face value.