Ya probably on 28nm in sept. It should slap fermi around pretty good . Since I pretty much shut down my computer from spring to fall ,I won't even look at fermi. I would rather see AMD do it first on 32 with HK/Metal gates . It will likely be 28 nm Bulk. With a 50% increase in what makes great Video cards great. Good back to school refresh
Ok, but you're assuming nVidia won't have updates to Fermi ready by the end of the year. Unlike you shutting your computer down in the spring to fall months, nVidia doesn't close down their operations for half of the year.
Keep in mind the issues with the Radeon X1800 didn't affect the X1900 by a huge amount. The X1900 was released right about when it should while the X1800 was massively late. In other words, don't count on the successor/refresh to Fermi to be very late. I'm not saying it'll be here early, just that it might not be held up as long as AMD (or the fanboys) hope.
The 5x00 series is great price/performance part and currently the best high end cards available. But I have to ask: how many times can an architecture be updated/refreshed before it starts showing it's age, i.e. bottlenecks and design limitations. They've been on this current architecture for quite awhile now and even though Fermi's not out yet, Nvidia will likely bring Fermi's refresh to market sooner into Fermi's life cycle than in G80's and GT200's cycle.
If AMD can keep refreshing/updating their cards as opposed to bringing a completely new architecure to market and stay competitive then that's really good for them and us. But I just have a feeling that, by this fall, upper and high end must-have cards will be squarely back in Nvidia's corner. I guess if they are coming out with a completely new architecture in about a year, then it'll probably work out great for them. We'll see though.
The problem for AMD is that short term, they have nothing to worry about but it is also my feeling that AMD really needs to refresh their architecture. The new info on Fermi that was recently released pins some of the blame of why it was late on the massive changes between the GT200 cores and Fermi. That doesn't help us in the short term if we want something from nVidia but long term, it might provide huge performance benefits for nVidia GPU's. Similar to what we saw with the Geforce 8 series. And the size hit of Fermi will become less important as TSMC moves At this point, we simply can't say.
As much as I've said that currently, GPU computing is not that important yet, and as much as I've said that physics acceleration isn't important yet, it will become important after 2010. More and more programs will start to optimize for GPGPU and games will start seriously considering physics acceleration. While not having a cohesive physics acceleration package/strategy hasn't hurt AMD, it will if they still don't have their act together by the end of the year. And I'm talking not just announcements but actual development kits and support. Same with GPU computing as some companies like Adobe are already using CUDA in CS5 or other products.
Just beware of the red PCB
Also note: ATI Catalyst drivers are buggy and slow.
After my x800 experience and catalyst Im proud to say I will never ever go ATIhell again in my life. nVidia 4 life fellaz....
Yes. Let's perpetuate falsehoods like ATI has crap drivers when, for the most part, their drivers are equal to nVidia. Just because you had a bad experience with them five years ago must mean they have kept on sucking since then right? Forget the many issues that nVidia have had with drivers since the release of ATI's X800 GPU. Not that ATI hasn't had their issues as well but dammit, only ATI has crap drivers!
🙄