Welcome to the S|A news roundup! We hope to bring you some of the biggest stories of the week as a supplement to our own. There are some good stories out there that we just don’t have the staff to cover.
First we turn to Microsoft’s upcoming Windows 8 product. Bloomberg released a story this week detailing Microsoft’s plans to have products based on the rather divisive OS by October of this year. The initial batch of systems will unsurprisingly be based on mostly classical x86 designs. A small number of ARM based devices will also make it into the market place. Microsoft has apparently set very strict standards for ARM based Windows 8 tablets, which is the same approach it took with Windows Phone 7 based devices. Launching Windows 8 in October is a smart decision for the Redmond, Washington based Microsoft which followed a strikingly similar schedule with its Windows 7 launch in 2009. It remains to be seen whether or not Microsoft’s tight hold on its ARM based OS is the right long term decision, especially when many attribute the wide spread use of Windows to its loosely controlled hardware ecosystem.
The news item that I’m sure you’ve all read quite a bit about by now is Nvidia’s recently launched GTX 680 GPU. Essentially what Nvidia’s done is taken its Fermi architecture, removed most of the compute oriented features, and slimmed down its individual CUDA cores in terms of both power use and die space consumption. This is a massive turn around in terms of design goals compared to Nvidia’s previous offerings, and it makes the GTX 680 well suited to the high-end PC gaming market. The most prominent example of an improvement made to Kepler over Fermi is the new memory controller which can run GDDR5 chips at 6Ghz, rather than topping out at about 4Ghz like Fermi’s memory controller. The Anandtech and [H]ard|OCP reviews pretty confirm what we’ve all been expecting since late January, Kepler wins.
More rumors are flowing from the Far East about AMD’s yet to be seen HD 7990. Inpai’s Zhang Fan had a short article that basically confirmed what we all expected from AMD top end part. Core clocks are down slightly from the HD 7970’s 925Mhz to 850Mhz. The HD 7990’s GDDR5 will be clocked at a flat 5Ghz and the article mentions an April release, which is in line with previous rumors and mutterings. The biggest question mark left on the HD 7990 spec sheet is TDP. Will AMD take a HD 5970 like approach and cap the card at a mere 300 Watts; or will they once again set the stock TDP to 375 Watts and include a BIOS switch with a 450 Watt profile? In all likelihood it will be the latter of the two alternatives, but at this rate we won’t have to wait much longer to find out.
Claims of AMD’s upcoming Trinity APU bringing a 29 percent improvement in CPU performance and a 56 percent increase in GPU performance over its predecessor, Llano, were widely discussed this week. Tom’s Hardware had the original article, in which they paid a visit to AMD’s booth at CeBIT. They were lucky enough to be walked through AMD’s Trinity demo, and pick up a pair of slides while they were at it. In addition to that article, PCEVA.com a Chinese site, released what appear to be benchmarks of the top end Trinity SKU, the A10-5800K. This chip is said to be clocked at 3.8GHz stock, with TurboBoost taking it up to 4.2Ghz under certain conditions. This chip is about 10 percent faster than Llano in SuperPi’s 1M benchmark. Make what you will of that, but at the very least it’s encouraging to see some leaks.
After a brief announcement last November at SC11, AMD finally launched its Opteron 3200 series of AM3+ compatible server chips this week. These chips all fit in to 45 Watt and 65 Watt thermal envelopes, for more on their specifications and target market Charlie has you covered. eWeek’s Jeffrey Burt also got some quotes from AMD’s John Fruehe on the subject. Really the most interesting characteristic of these new Opterons is their ability to use a desktop CPU socket. It’s hard to say exactly where this will lead, but a lot of desktop users were very happily using Opteron parts back in the Socket 939 era. Perhaps we’ll see the same sort of ecosystem start to develop around these Opteron three thousand series parts.
This concludes the news roundup for the week of March 19th.S|A
Latest posts by Thomas Ryan (see all)
- Intel’s Turbo Boost Max 3.0 and the i7-6950X - Feb 24, 2017
- Performance Per Watt with the i7-6950X - Feb 20, 2017
- Sony Stacks Memory on its Sensors to Fight Rolling Shutter - Feb 9, 2017
- Intel’s Latest Flagship Server Chip is the Xeon E7-8894 v4 - Feb 9, 2017
- AMD Offers its ReLive Tool to the Pros - Jan 30, 2017