YANL

I started a term when the 9800 GTX was first released known as YANL: Yet Another NVIDIA Launch.  Basically it's a derogatory term poking fun at a company that produces a LOT of different brands and products; some with very little separating them from a technical perspective. 

Yesterday news started leaking about a new GPU from NVIDIA called the GeForce 9800 GTX+ - essentially an overclocked G92 part built on a new 55nm process technology that will sell at a LOWER price than current 9800 GTX cards on the market.  Interested yet?

NVIDIA GeForce 9800 GTX+ 512MB Details

We have already discussed the 9800 GTX+ card and analyzed images of the die in some recent news posts, so I'll be reusing that data below. 

Get ready for the details:

• MSRP: $229
• Core clock: 738 MHz
• Shader clock: 1836 MHz
• 512MB GDDR3 Memory: 1100 MHz
• 2- and 3-Way SLI support
• kick in the ass

NVIDIA GeForce 9800 GTX+ card unmasked

NVIDIA GeForce 9800 GTX+ GPU @ 55nm

NVIDIA GeForce 9800 GTX GPU @ 65nm

NVIDIA GeForce 9800 GTX+ (top) versus NVIDIA GeForce 9800 GTX (bottom)

A little photoshop action putting them side by side; 55nm on the left, 65nm on the right

By my crude measurements, the 55nm die is 15.2mm high, or 231
mm^2 in comparison to the 65nm G92's die area of 324mm^2 - a savings of
approximately 40%. 

So that seems pretty complete for a pictorial guideline, right?  The real question you might be asking is why would NVIDIA NOT want to push the point home about the 9800 GTX+ being a 55nm part? 
Obviously there are some advantages to it: lower power, less heat,
faster clocks (obviously). 

My guess is that the marketing team at Big Green didn't want to
one-up the recently released GT200 parts - some people might assume
that singe it was built on the 65nm process while the new 9800 GTX+ is
on the 55nm, that the GT200 would be inferior in some way.  As we saw
in our performance review of the GeForce GTX 280/260 cards, that's obviously not the case but I can see how some readers might make that conclusion. 

From what I heard while at Computex about this part, NVIDIA wasn't
sure about releasing it even as late as a couple weeks ago.  NVIDIA was
probably using the 55nm shrink of the G92 core as a "practice run" of
sorts for 55nm production and design but then was impressed with the
results they got back from the fab.  The, with the help of some
pressure from AMD with the HD 4850 and HD 4870 launches next week,
NVIDIA decided to pull the trigger on production.

I will be curious to see how the 55nm transition of the GT200 part
goes, if NVIDIA will try to make significant modifications in the
design before production or simply attempt to ramp it up ASAP to help
with costs and yields.