AMD's oncoming new graphic cards that aren't that, you know, oncoming.
http://en.wikipedia.org/wiki/Southern_Islands_(GPU_family
I have an aging HD4850 that I'd really like to replace with something newer as an interim step to moving to Ivy Bridge next year, so I've been waiting for these to appear. The word was september, but apparently the word was a bit hasty. Certainly moving to a new smaller process is likely to be one reason, but competition strategy might be in the running as well.
At this point no-one knows their real world behavior.
Now since I'll be moving to an Intel processor with an updated IGP on it I'm wondering if the entry level Southern Islands cards are worth their combined price, energy use, heat production and the place they'll take up. A smaller PC (currently rocking an ancient p180) would be something I might just sell the missus on.
This, of course, depends on Ivy Bridge which has also been delayed and also has little actual real-world behavior known.
So, should I get my hopes up for vaproware 1 or vaporware 2?
What about connections? I have a dvi flat panel- will that need replacing? Is it likely Ivy Bridge GPUs will have good multiple monitor support?
Posts
What are you looking to spend on a card?
I've never been happy with graphics card cooling so normally have to splurge on a third party cooler. I'm hoping the new 28nm process will get me a passive card in that price range.
Performance wise I'm not looking at any great improvements over the HD4850. So if the Ivy Bridge IGP can top the HD4850 that's plenty. At 150 euros more for a discrete GPU I'm looking for at least double the performance over the IGP in openGL.
http://www.anandtech.com/show/4585/intels-haswell-igp-to-feature-directx-111-increased-professional-application-support does seem to indicate that a GPU could beat the IGP at openGL by such a significant margin. But it's all a bit pie in the sky without tests.
To be honest I'm rather disappointed that these two architectures have been plowed into obscurity with such ease. The whole reason tech sites are there is to help me decide where to spend my money- I've never known such a lack of information on tech that is literally around the corner. If anything it leads me to suspect- along with 22nm ssd problems- that the semiconductor industry has hit a technical limit a few orders of magnitude larger than they want known. And the traditional business response has always (and recently again) been to ship anyway and let the early adopters pay the prize for fixing the process.
Also, sorry I thought I posted earlier.