MaximumPC 2007 10

(Dariusz) #1

watch dodogg MAXIMUM PC TAKES A BITE OUT OF BAD GEAR dog


And the company had the gall to tell me to go back and
buy another product that I basically already own? For
now I’m going back to XP, which is not an easy task.
Please bark up the ATI tree (I already did) to get
it to change its website notice, make WDM drivers
to support its current products for Vista, and maybe
offer rebates or discounts because of the lack of
product support.
— Scott Schwarz

The Dog had a lengthy discussion with AMD/
ATI officials about the situation, as it seems
perfectly reasonable to support a year-and-a-
half-old product with Windows Vista drivers. But
the answer is a bit more complicated than you
would expect. AMD officials told the Dog that
the problem is not a simple driver rewrite—it’s a
problem with how the All-in-Wonder is designed.
The capture and tuner drivers operate as child
drivers of the main display driver. Under the
driver rules in Windows XP that’s fine. However,
in Windows Vista, when the child driver tries
to allocate a 3D space, it is unable to because
Windows Vista doesn’t give child drivers that
capability. “We have looked for a workaround
but have not been able to find one,” the spokes-
man said. Not all is lost though. The spokesman
said that AMD recently certified SnapStream’s
Beyond TV (www.snapstream.com) as a solu-
tion for All-in-Wonder cards under Windows
Vista. AMD was also in the process of certifying
Arcsoft’s TotalMedia 3 (www.arcsoft.com) as
well. The reason those two applications do func-
tion with the All-in-Wonder while Media Center
does not is because the pair of applications do
not run as a service, officials said. The com-
pany said support for the tuner was intention-
ally discontinued when it was unable to find a
workaround for the problem with Windows Vista.
Officials also wanted to point out that the 2D
and 3D functionality of the card does work and
is supported with drivers. What about rebates
or some way for AIW owners to get a break on
Beyond TV or TotalMedia 3? You have a bet-
ter chance of getting money out of a pyramid
scheme. Woof.

ERRATUM TO THE CORE
Dog, I tried contacting Intel about an erratum in
its Core 2 Duo and Core 2 Extreme CPUs but could
not get a response. The Erratum is called AK
and says: “AK39. Cache Data Access Request from
One Core Hitting a Modified Line in the L1 Data
Cache of the Other Core May Cause Unpredictable
System Behavior. Problem: When request for
data from Core 1 results in a L1 cache miss, the
request is sent to the L2 cache. If this request hits
a modified line in the L1 data cache of Core 2, cer-
tain internal conditions may cause incorrect data
to be returned to the Core 1.”

Intel says the fix is a BIOS workaround. But I
want to know if the BIOS workaround affects per-
formance. I don’t see how it couldn’t. I also want to
know when a fixed stepping is expected?
— David Galbi

The Dog queried his contact at Intel, who
tersely said that the workaround did not cause
a performance hit. The spokesperson also said
that the company does not comment on step-
ping releases, so the Dog can’t provide the
scoop on that topic. Is Intel telling the truth
about the BIOS fix? Probably. Intel and AMD
(as well as Nvidia and ATI) are very closely
watched companies. If a BIOS workaround had
any noticeable performance impact on a CPU,
people would bark. Heck, people howled to the
moon about the minor L2 cache changes in the
65nm Athlon 64 that slowed things down by a
tiny margin in only a few benchmarks, so the
Dog believes that testing would have revealed
a problem by now. The Dog should also point
out that if Intel were trying to hide such a flaw,
it probably would not publish it as a known
erratum on a document that it distributes to the
public and its hardware partners.

Tuner functionality in the All-in-Wonder
series was broken by Microsoft Vista,
not ATI.

Intel says an
erratum on its quad-core
CPUs will not impact performance.
Free download pdf