MaximumPC 2006 12

10MP Camera Phone Arrives

Samsung’s B600 breaks the 10-megapixel barrier

Google + YouTube =?

Google to buy out YouTube as sixth sign of the

coming apocalypse

If you can’t beat ‘em, buy ‘em. At least that

seems to be the thinking over at Google these

days as the company enters into a $1.6 bil-

lion deal to acquire its scrappier and more

popular online-video competitor YouTube. The

move has obvious benefi ts for Google, whose

own video-streaming service—launched last

January—couldn’t touch the traffi c heading to

YouTube. YouTube streams a whopping 100 mil-

lion videos daily. Perhaps even more important

are the partnerships YouTube forged with most

major motion picture studios, music labels, and

television networks—as well as the potential for

lucrative advertising contracts.

What’s less clear is how this deal will ben-

efi t consumers. If Google absorbs YouTube into

the Google Video brand, it will leave a massive

chasm where YouTube used to be. But it’s only

a matter of time before some newer, scrappier

service fi lls the gap.

Retailers Cry

Foul over

Online Movies

In the wake of movie down-

load launches from Apple

iTunes and Amazon Unbox in

September, retail giant Target

is pitching a fi t. In a letter

to movie studio representa-

tives, Target president Gregg

Steinhafel threatened to reduce

the amount of shelf space it

devotes to movies unless stu-

dios could ensure equity pric-

ing among the different movie-

distribution methods.

Frankly, we’re a little

mystifi ed by the concern. At

press time, the price difference

between online and retail ver-

sions of major new releases

was negligible. For instance,

X-Men 2: The Last Stand sold on

Unbox for $14.95, while Target

offered it at $14.98, and the

same trend held true for just

about every new release we

looked up. If anything, we’re

annoyed that the DRM-encum-

bered online versions aren’t a

whole lot cheaper than the

easily ripped DVDs.

DECEMBER 2006 MAXIMUMPC

With a massive 3x zoom lens embedded in its backside, the Samsung
B600 is the world’s most powerful camera phone. For now.

B

y shipping its first quad-core processors in

November, Intel is beating AMD by at least six

months. AMD doesn’t expect to deliver its first quad

(Barcelona) until mid-2007. But Intel’s initial quads

are actually pairs of dual-core dies slapped together

in the same chip package, whereas Barcelona will

integrate four cores on a single die. Some critics

say Intel is cheating.

I don’t call it cheating, but I do use another

term Intel doesn’t like: “stopgap measure.” I think

a multichip module (MCM) is a compromise on the

way toward a tightly integrated quad-core die. Intel

argues that stuffing two dual-cores in one package

is an equally valid method of creating a quad-core

processor and even has some advantages.

If this debate sounds familiar, it’s because we’re

in the second round. Last year, Intel introduced its

first dual-core processors. The Pentium Extreme

Edition 840 (Smithfield) united two Pentium D cores

on a single die, but with minimal integration. The

Pentium Extreme Edition 955 (Presler) crammed

two Pentium 4 cores in an MCM. Six months later,

Intel rolled out its first “monolithic” dual-core chips,

which integrate two of the latest Core processors

on a single die.

Then, as now, Intel argued that pairing separate

dies in one package can save money by boosting

production yields. If one core is defective, Intel

discards only the die having that core, whereas if

one core on a monolithic die is defective, the whole

die is trashed, even if the rest of the chip is OK. In

addition, Intel can bin-sort the production output

to find the best-matched pairs of dies. Intel tests

and sorts the chips according to their clock speeds,

voltages, and other characteristics.

Intel’s points are sound. But it’s also true that

MCMs are more difficult and costly to manufacture,

which partly negates the advantages of bin-sorting.

Also, Intel’s MCMs are poorly integrated. The new

quads, like their early dual-core predecessors, must

route all core-to-core communications through the

front-side bus instead of through tightly integrated

interconnects. That detour saps performance.

Intel doesn’t plan to introduce monolithic

quad-core processors until at least the second

half of 2007. When those chips appear, don’t be

surprised if Intel touts the advantages of mono-

lithic integration.

Tom Halfhill was formerly a senior editor for Byte magazine

and is now an analyst for Microprocessor Report.

Quad-Core

Trade-Offs

FAST FORWARD

TOM

HALFHILL

FRONT BACK

09