Centrino

Whats all the fuss about? When I was reading reviews of the new Pentium M chip from Intel, it seemed like a great chip. The more I thought about it, the less I thought of Centrino. Now I may not know as much about laptops as I should. At [H]ardocp today there was news of a 1.6 Ghz Centrino laptop keeping within 90% of a P4-M 2.26 laptop in 3D Mark 2001. It seems like the centrino is getting most of its praise from its high IPC because a more efficient chip will give more battery life in a notebook. After reading the post at [H] I went searching for a review of a 1.6 Ghz Athlon XP (2000+). At Anandtech I found the perfect review: a 2000+ versus a 2.26 ghz P4; just like the comparison from [H]. Here is a quote from Anand, "Both the Athlon XP 2000+ and the Pentium 4 2.2GHz processors are very close performers in most respects." After reading this it seems as though the Pentium M has very similar IPC to the Athlon XP. The athlon xp-m's have similar power saving techniques as the Pentium M's speed step. My question is, other than the fact that a Centrino Laptop comes with an intel northbridge and wlan integrated, why is the Centrino so much better than the athlon xp-m. They have similar IPC, but AMD has raised clock speeds much higher (2400+).

because it has intels name. and it was built from the ground up to be a mobile chip, not just a lower voltage desktop chip with some new instructions. blah blah blah. mostly its just getting praise because intel advertises like no other

I agree that it is a total waste. I would stay away from it.

Doesn't the Centrino also have 1MB of cache?

Thats another good point. If you took an Athlon XP at 1.6 Ghz and gave it 1 MB of cache it would probably be a 2800+ or something. So once again AMD's processor is a much more efficient cpu in terms of IPC.

Considering how little Barton's 512KB of L2 cache helped the architecture, what makes you think doubling it again would help more? Let alone less.

Either way, the Pentium-M design is not exactly an engineering marvel. It's just a chip designed with a different purpose in mind. And to that end, it accomplishes its goals, but it's the natural conclusion of the design goals they had. It provides good performance at lower power levels, and for those of us carrying around a bulky Dell that can't get 2 hours of battery life, it'll be a welcome addition to the market.

Actually, I saw a review comparing Intel's cache to AMD's cache. They apply the way the processor uses it differently. So they compared a Williamette (256K) P4, a Northwood (512K) P4, a (Palomino or T-bred, I forget), and a Barton.

The Intel's saw a significant gain from their increase in cache, because of the way they use it, the AMD's didn't benefit as much from the increase.

I've put a link to Anand's review of Barton along with comparisons of cache benefits:
Review
The reason spans far beyond the inclusive/exclusive use of cache, the two major reasons that the P4 design benefits from a larger cache better than the K7 is 1. it has a 128-byte cacheline vs 64-byte on the K7 and 2. it has a relatively higher clockspeed.

Of course, the Pentium-M is not a P4 design so we don't know exactly how much the 1 MB of L2 cache helps.