Mobile Sempron vs. Celeron M: better chip for battery life efficiency?

[Ionizer86]

When I read Anand's $600 laptop roundup, I was suprised that the Celeron M based Compaq V2000 was seriously outlasting the V2000Z based on the mobile Sempron. I thought that Speedstep and PowerNow makes a huge difference in battery efficiency, and so the Celeron M would do much worse than the Sempron due to lack of Speedstep.

Let's couple this with a Toms article showing comparable runtime between a mobile Sempron (a 25W chip, much like the Turion MT), a Turion MT, and even a Pentium M / 533. This article shows that all three chips, which have Speedstep or PowerNow, post similar battery efficiency numbers.

Finally, to complete the loop, I saw an article on Laptoplogic comparing the Turion ML to the Pentium M / 533. Here, these two chips post similar battery life times.

To put details together: we know the Pentium M outperforms the Celeron M by a lot in battery life.
We also see from the Toms article and the Laptoplogic article that the Turion ML compares well with the Pentium M / 533.

Now, we also know that the Turion MT is more battery efficient than the Turion ML. And from Toms Turion article, both the 25W Turion MT and 25W mobile Sempron compare similarly with the Pentium M / 533.

How does the mobile Sempron do so poorly in Anand's roundup? It seems like a break in logic:

Celeron M > mobile Sempron
+
Turion ~= obile Sempron ~= Pentium M > Celeron M
=
mobile Sempron > mobile Sempron?

 

[Furen]

The CPU is but one part of the power consumption game. Some chipsets are remarkably crappy with regards to power consumption, a driver/bios problem could make the Sempron system just not downclock properly, there could be a hardware problem somewhere... heck, the PCB design can also mess this up. Also, the Sempron could be the 62W (maybe 130nm) part, which is $12 cheaper than the 25W one.