I
believe that Nvidia, ATI (err, AMD), and the power supply manu-
facturers are playing with our emotions. In the space of three
years, we’ve gone from needing two four-pin Molex connectors to
power high-end GPUs to needing a single six-pin, then two six-pins,
and finally four six-pins to run two GeForce 8800 GTX cards in SLI.
Don’t get me wrong; I’m not complaining about the 8800 GTX
cards—they bring a smile to my face—but I am confused by the
apparent lack of coordination among the various vendors.
Just when you thought it was safe to go back into the power-
supply waters, there’s a plan to change the connectors yet again.
Apparently, 75 watts from each six-pin plug and 75 watts from the
PCI-E slot isn’t enough.
Next-generation cards from AMD, I’ve been told, will need 300
watts of power, and Nvidia has been hunting for more power, too. In
fact, an early engineering-sample 8800 GTX card I saw had a six-pin
and an eight -pin connector.
Power supply vendors tell me the design was yanked at the
last minute out of fear that a tyro PC builder would jam the existing
eight-pin plug that should go into the motherboard into the GPU.
It’s tough to do, but with enough force it’s actually possible, andthe result would be disastrous. A former coworker of mine once
jammed a Molex plug into a solid-state hard drive upside-down
and whoosh!
So what’s in store? No one knows for sure yet. I’ve heard dif-
ferent stories from different power supply vendors, but a decision
may come soon. The PCI-SIG, the user group for PCI, is currently
working on a new spec that should standardize power connectors
as well as address thermal issues and the mechanics of heavy
graphics cards. The early draft I saw showed only a single eight-
pin connector delivering power for 225-watt and 300-watt cards,
but is it enough? One PSU vendor I spoke with said high-wattage
cards would likely need eight-pin and six-pin connectors. The spec
will be debated and finalized later this year, but let’s hope they
build in enough headroom to support at least another generation
of cards.Think your one-kilowatt power supply delivers enough
juice? A new spec may render it useless6 MAXIMUMPC april 2007
iH the F;< Real-WoRld testing: Results. analysis. Recommendations
An early engineering-sample GeForce 8800 GTX sports a six-
pin connector as well as a new eight-pin connector.GORDON MAH UNG
Wonders If Your
Power Supply Is
Already Obsolete
I^
went into the Lab thinking I wouldn’t hear a bit of difference
between Apple Lossless, open-source FLAC, and Microsoft’s
WMA Lossless. And I might not have had I used something less
than Ultrasone’s Proline 750i headphones and Creative’s X-Fi
Elite Pro soundcard (the Proline 750i’s are divine, and the Elite
Pro is outfitted with higher-quality DACs than other X-Fi’s).
Even then, I could detect sonic differences only with repeated
A-B-C comparisons of Ben Harper strumming his acoustic guitar
to produce a sustained chord. Without vocals or other instru-ments, the guitar just hangs there, naked. Once the rest of the
band kicked in, I couldn’t discern any of the compressed tracks
from the tracks on the CD. So why pick FLAC over the others?
Because in the back of my mind, I know I’m missing something,
even if I can’t detect it in most listening sessions.
Besides, FLAC is widely supported by audio-streamer manu-
facturers, including Slim Devices, Roku, Olive, and Sonos. Roku
and Slim Devices also support Apple Lossless, but support for
WMA Lossless is much harder to find.
Apple Lossless is the big winner on the portable-device
front, thanks to the iPod. The TrekStor Vibez and Cowon iAudio
X5 are among the few players that support FLAC, and Toshiba’s
Gigabeat S-series of players are the only devices we know of that
support WMA Lossless. Note to Microsoft: If you want the indus-
try to support your lossless codec, try supporting it in your own
hardware instead of consciously stripping it out of the Gigabeat-
based Zune.Michael Brown
Tests Lossless Codecs
FLAC is yet another open-source success story