The common trend here seems to be a concern over not cannibalizing existing markets. Two examples are the case of flash memory being delayed by a decade or more, and also Microsoft releasing a bloated Vista with the expectation that faster processors were on the horizon. Too bad there can't be some legislation to encourage a synergistic effect.

Hopefully, netbooks and FLOSS can free us from this x86-only world. I would love to have an 8-core ARM in my netbook.

I remember my IBM z50. It used some parts that were used in a Pentium notebook sold in some markets. Because of this part sharing with a Pentium notebook, the MIPS processor in it could run for 10+ hours on a single charge.

Why do you prefer ARM over x86?

No ball-and-chain of Windows compatibility holding you back.

No possibility of being lazy and using a traditional PC BIOS + ACPI etc. architecture. x86 doesn't even have mature alternatives: EFI doesn't really help anything, and OLPC's OpenFirmware approach was implemented poorly. With ARM there's already a bunch of well-implemented variant architectures from multiple vendors with full support in dozens of operating systems.

Yep, on ARM you don't have those problems; you have different ones: every system has unique firmware and requires a special bootloader and kernel, making it impossible to create one OS image that will boot on many ARM systems.

http://lwn.net/Articles/364654/

There is hope, but I doubt it will arrive in time for the first wave of ARM "smartbooks": http://lwn.net/Articles/367752/

Also, Ubuntu and Chrome OS seem to have divergent plans for firmware.

If OLPC pulls the XO-1.75 off, there will be a reference design for ARM-based notebook that is open-source from top to bottom.

Lower power consumption, I'd guess.

I also like the cleaner instruction set. It may make compiler writer lives easier, render better optimization strategies, among other things.

It's maybe a little nicer for assembly programmers, but x86 is not that bad any more, and the predication and auto-shifting features in ARM are not that easy to deal with for compilers.

The main thing that may be problematic for x86 is that the individual cores may get smaller and simpler, which means that the decoding hardware will take up a proportionally higher portion of each one.

The print version was covered here:

http://news.ycombinator.com/item?id=980927

With or without Larrabee, the number of cores and threads per core on x86 processors will not decrease.

FWIW, the January 6, Stanford EE380 is about Larabee. See http://www.stanford.edu/class/ee380/

Yes, it's open to the general public and will be webcast. (The latter may, or may not, work on "not Windows" machines.)

Thank you. Can it be made available for download after the event?

With few exceptions, all EE380 talks are available for download "indefinitely". Often the slides are available separately.

Thanks. It's great to know. I will schedule it so I can see as soon as possible.