Rambus (RMBS)

[elixe]

Cor, Regarding Performance graphics

From what I’ve seen, performance graphics memory represents about 4 - 5% of W/W DRAM revenues.
(About $1 billion this year).

GDDR3 is expensive, but so are performance graphics cards. Predictions are that approximately 15 million High-end Graphics cards will ship this year. That represents about 6% of forecast graphics shipments of ~260 million products, including embedded graphics. (Intel has ~43% of the graphics market).

ATI and nVidea will ship a combined total of about 110 million graphics cards this year. So it’s obvious that Intel, ATI and nVidea pretty much own the market.

A Performance graphics card with 256-bit GDDR3 will typically have a bandwidth of 35 GBps.
(For example the ATI X850).

If ATI or nVidea were to produce a 128-bit solution with XDR-3.2 it would have a bandwidth of 51.2 GBps.

Both solutions would have eight DRAMs, and I don’t think COG is a major issue.

The question for me is how would the XDR solution benchmark vs GDDR3 given it’s higher granularity?

Considering that GDDR3 Graphics solutions are in volume production, what does it take for XDR to displace them?

BTW, can I assume you’re still gathering inputs for your Rambus report?

All JMOs, of course.