Intel’s Menlow: a primer on MID power

1 Comment

SilverthornebackHaven’t had enough Menlow information yet? You’re in luck because AnandTech has a high-level piece with a some detailed tidbits sprinkled in on the new mobile processor platform. While we’re all hoping for UMPC-nirvana, AnandTech appears to have lower expectations:

"The Menlow project began back in 2004, so what you’re hearing about today is much like what Centrino was back in 2003.  We’re on the verge of something very big …It was nice to see Menlow up and running, but our real excitement isn’t about these Menlow based MIDs but what is coming in the next 3 years."

If you’re looking for a good "Menlow 101" article, this is a good place to start. Although it’s far too early to tell, my concern for Menlow is a majority of battery life advantages coming from reduced performance similar to the A11x devices. Then again, more of the Menlow focus at CES was around MIDs running lighter environments as opposed to UMPCs running full desktop operating systems.

1 Comment

Chris K

I think that you’re probably jumping the gun by a fair bit, but then again, I’m consistently of the belief that IO matters more than CPU after a certain point. (That point is somewhere past the Geode LX chip, but I’m not even entirely convinced of that.)

Intel has a significant advantage at this point in the embedded space, since their lowest-power x86 chip is still based on a significantly superior floating-point unit. That has little to do with clock speed, and that’s where things get interesting:

For general purpose computing, you need a responsive ALU, but not necessarily a blazing fast one. You just need to keep it well-fed, and that takes RAM, and decent IO to storage. Basically, in the cycle of fetch, decode, and execute, fetching is a giant bottleneck for these cases. The neat part is that we fix a lot of this by shrinking the data, so a 5-minute video can be fetched quickly, but now decoding and execution take a while, since they use floating point calculations.

So, where does this place Intel? Simply put, they’re cramming the Core 2 Duo into a slower, smarter package. That doesn’t sacrifice the FPU performance on a clock-by-clock basis, and that’s going to be absolutely huge, since that’s where video operations go when the graphics chipset can’t handle them. Flash video will decode faster. Quicktime will decode faster. Movies will fit in your palm on a lower-clocked processor.

We can’t ignore that the average CPU speed has gone DOWN over the past few years in consumer systems, as Intel has recovered from the P4 debacle. Slower is actually better at this point, since we’re approaching a human bottleneck. As a user, I can only perceive a certain amount of responsiveness, and the rest is gravy. Windows need to swap and redraw quickly, yes. Movies need to render at 30FPS at 480p on a handheld computer, but that’s it. If that can be done with an 800 MHz part that’s brilliantly efficient, then we’re in good shape. The computer’s job is to process and present data quickly, and while the P4 was a Ferrari compared to the Core Duo’s 18-wheeler, I know which choice I’d pick when I need to move a couch at a minimum speed of 55mph.

Comments are closed.