With no immediate pressure from the competition and a healthy mainstream product lineup, there is no reason for NVIDIA to rush new cards out to the market
This article was first published in German on K-Hardware.de.
In talks with industry insiders at
CeBIT I was able to confirm that NVIDIA's upcoming mainstream DX10
G84 and G86 cards would have been ready in time for a CeBIT launch.
NVIDIA, however, decided to postpone the launch to April 17th for
various reasons.
I wouldn't put much doubt in the
quality of the chips of the cards at this point, since reputated
video card manufacturers confirmed to me that these are up and
running. MSI was already able to show custom designs of G84 and G86
based retail cards, not reference cards. On top of that, various companies
exhibited notebooks equipped with the mobile variants dubbed G84M and
G86M, running all day long under Windows Vista without a crash.
After all, the PCB for GeForce 8600 on down is based on the stable GeForce 7600-series line, which was already a derivative of the GeForce 6600-series layout. The only component that would need any fine tuning at this point is the GPU -- and that's been stable for months.
One possible reason is the recent delay
of AMDs R600 generation to the second quarter of the year. There is no pressure from the competition to rush out the
cards to the market. Instead NVIDIA can lean back and happily sell
current GeForce 7 mainstream cards. Meanwhile their GeForce 8
mainstream cards based on G84 and G86 are in production. Basically
NVIDIA is preparing for another hard launch.
According to industry sources I've talked to, NVIDIA will be able to deliver enourmous amounts of those
cards when they launch - they will basically flood the whole market.
If you talk to the AMD crowd, however,
you get other reasons, which are also somewhat realistic. Apart from
various problems with the hardware, NVIDIA would delay the launch as
long as possible to have a working driver ready. The
current state of G80 drivers is not at the level of quality
one would expect.
Though AMD may be suffering from rather low yields on TSMCs 65nm process, it decided to employ the new process on the upcoming R600 chip family, even on the largest high-end chips. This approach may put ATI ahead technically, but it's risky with regard to the fact that NVIDIA can put massive quantities of older components at lower prices on the market very quickly.
"I modded down, down, down, and the flames went higher." -- Sven Olsen
|
|