Giving Credit Where It's Due: Intel 45nm

There's a lot we take for granted, especially when it comes to Intel. AMD has always enjoyed the benefit (of the only benefits) of being the underdog in that when it does something, we all cheer, whereas it takes a more revolutionary feat for us to really appreciate Intel.

Intel did pull some nice tricks out of its hat back in the Pentium 4 days, but the chip was just not very good and couldn't really compete with AMD. It wasn't until Core 2 that we, as a community, really started considering Intel once more. But since the Core 2 launch, we've treated most of Intel's advancements as meh-points.

I'm especially guilty of this; I remember finishing Penryn testing and thinking "that's it?". Here I am complaining about a 5 - 10% increase in performance that will most likely be given for free compared to today's prices, how spoiled I've become.

Then there's the bigger item of 45nm in general. I touched on this briefly in my Phenom Preview, but Intel has put itself in a pretty damn good position right now. The pressure is on AMD to continue this price war that no one can win (other than us), and Intel's cost structure simply gets better with the move to 45nm - which by the way is coming along quite nicely. For such a dramatic change to the transistor, Intel has executed very well on its 45nm roadmap; I've had Penryn for a while, as has Intel's partners, and honestly it's been performing just as good as final silicon this entire time.

I did a quick performance comparison between A0 and A1 Wolfdale (dual-core Penryn) and for the most part there was no performance difference, Oblivion and Half Life 2 showed a ~1 - 2% increase in performance but I don't consider that significant. The technology is mature, clock speeds are artificially limited, and as I mentioned in the Phenom preview: I believe we would've seen the entire Penryn lineup earlier if AMD launched Phenom this summer.

Intel is sitting pretty right now, honestly the only area where I don't see a clear strategy is in its foray into the CE market. We've heard the codenames (e.g. Silverthorne) but how Intel plans on getting these chips into DTVs, cell phones, etc... isn't as clear cut as AMD. AMD will simply offer two versions of its Xilleon processors that are already used in these devices, one with an integrated x86 core and one with a MIPS core - to the hardware designers it's a simple swap.

Intel doesn't have as big of a presence there as AMD simply because of the ATI acquisition, and it's not really a detriment today because the core that would go into those devices isn't ready yet. I'd say I'm more curious than anything because as I've mentioned before, I'm really interested in seeing higher performance, x86-cores, in these devices with hopes of the oft promised "digital home" finally coming to fruition.