Thoughts
Performance wise, there's one thing that stands out from the analysis and observation of P-M's scaling with nothing more than its base frequency: it's crying out for core logic to let it breathe and perform to its fullest; there's definitely plenty of extra performance to be had by pairing P-M with Alviso core logic. There's also extra performance to be had just by upping the CPU's bus clock, so the upcoming 200MHz Pentium-Ms, using Alviso, should give very fine performance, comparatively.P-M's shorter main pipeline, compared to the processors I put it against, mainly allow it the higher performance per clock you can see, moreso than L2 cache performance. From there, it's pretty simple to see the processor's appeal.
The heat output and its frugal design in terms of power, especially when combined with a platform that can enable power and heat saving measures like EIST are the icing on top of an already impressive cake.
Pentium-M is on the way to a 200MHz base front side bus in coming months to partner speed hikes, followed by the dual-core, shared cache, Yonah Pentium-M on 65nm in 2006. The future's bright for the CPU created by Intel's Israeli design team, and it's all down whether Intel see fit to bring it to the desktop space. Intel's investment in Netburst and the Pentium-4 that implements it are the biggest barriers to that.
The most glaring omission in Yonah's feature set is x86-64 compatibility, as it implements SSE3, increases the performance of the SSE unit in general, and improves the performance of the FPU.
Whatever happens, Pentium-M is a valid choice for those in the know, looking to build high performance PCs that aren't noisy, hot or chew through your electricity bill. They're not cheap at the time of writing, but they're available and the boards to put them in are getting more numerous by the day.
We'll endeavour to cover new Pentium-Ms as they're released, on its new core logic platform, along with any new mainboards that make it happen.
