bit-tech.net

Intel has answers for AMD’s Fusion

Intel has answers for AMD’s Fusion

Otellini said that IGP process technology will catch up with CPU process technology by 2009-2010.

Paul Otellini said this morning that Intel plans to increase integrated graphics performance by a factor of ten by 2010, while also changing the rate of adoption of new process technologies in its integrated graphics chipsets – right in time for when AMD plans to introduce ‘Fusion’.

In the past, we’ve seen Intel’s integrated graphics chipsets falling one or two generations behind the company’s bleeding edge process technology – the current IGP chipsets are built on a 90nm process for example. However, recognising the need for improved graphics capabilities, he said that we’ll see 65 nanometre integrated graphics products early next year – this transition will deliver roughly twice the graphics performance we’ve got today in integrated graphics.

Frankly, this is nothing to get excited about because it’ll be twice as much as almost nothing, so although it sounds like a big jump, it’s not going to cure cancer overnight.

The big performance gain comes in the first half of 2009 when Intel introduces integrated graphics chipsets manufactured on a 45nm process – this will be the first time that the IGP will be manufactured on the same process as the CPU. The reason for this is simple: the IGP will be integrated into the processor – you’ll remember that Intel told bit-tech that Nehalem will feature an integrated graphics processor back in March this year.

This move will increase performance by a factor of over six compared to Intel’s 2006 IGP products and then a 10 fold performance increase will be reached when Intel moves both its CPUs and IGPs to 32nm in 2010. At this point, both CPU and IGP are designed from scratch to hit the same early launch point on a new technology. He acknowledged that this wasn’t good enough though, and that’s where Larrabee comes into play.

Discuss in the forums

3 Comments

Discuss in the forums Reply
Bladestorm 19th September 2007, 02:35 Quote
Hmm, even 10x almost nothing .. in 3 years ? they'll probably still be at the point that major game engine developers say they wish they'd cut it out and stop making them ;p

Larrabee sounds vaguely interesting, though even if it is relatively easy to develop for, presumably a cpu-based implementation will still be vastly different from what major graphic engine developers have been designing and working with and the hardware industry has been developing graphics cards around ? surely ?

Intel might have the corporate power to launch physics processors into the mainstream though.

Edit add : morning tim :)
Bluephoenix 19th September 2007, 16:37 Quote
the IGP on the CPU is interesting because if done right it will remove any process delay caused by the FSB or similar.

I think discrete graphics will still have a place for those who need more powerful solutions (however this will certainly be a kick in the pants for notebooks)
iwod 23rd September 2007, 14:00 Quote
I argee.. 10x increase in 3 years is almost nothing. 20x to 30x would be nice to get gaming on IGP going.
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums