Xbitlabs: AMD Expects Next-Gen Chip to Cut Power Consumption by Half

[Saico]

You mean AMD cut the chips performance to reach 5w. This is an accomplishment . Not ! Its the same as intel does with atom . No microarch change just. respin of same old chip. intel not to its credit did little on the 32nm Atom to improve it. It will be seen as a massive error in the near future on intels part.

Wrong.

The 5W version of the C-50 does not suffer in the performance department, still being a dual-core clocked at 1GHz, no. The power consumption has been cut via cutting back on features, such as having a simpler memory controller (less support for lager memory, clock speeds and memory timing), as well as I/O support cut (no multiple ports of the same type – ie. one USD, one HDMI etc.)

 

[LOL_Wut_Axel]

If MBA was a windows based laptop I would have already bought it. It's just the kind of laptop I'm looking for. However I'm going to make a prediction now:

All windows based ultrabooks will use a 1366x768 screen (instead of 16:10 1440:900 in the MBA) that is the cheapest available = bad TN Panel ultra glossy. Which will make them pretty much useless and similar to 95% of the other unusable crap. Also I predict they won't be much cheaper than the MBA.

Just FYI, the 11" MBA uses an 11.6" 1366x768.

Intel is making $1000 as the min recommended, but most manufacturers seem to be having problems getting them down there without going to a slower CPU like a Core i3. I think the MBA will prove to be the most popular due to having a good price in comparison to their ultra-book competition. That's not to say that I don't find the ultra-book concept over-priced (they're only somewhat faster and somewhat thinner while losing I/O and battery life), but within that small niche the MBA seems like a decent buy.

 

[LOL_Wut_Axel]

Wrong.

That's true. It should be clarified, though: the 5.9W TDP part being talked about is the Fusion Z-01, which is based on the Desna platform and not Ontario.

 

[Nemesis 1]

Wrong.

Link and Benchmarks please. How did AMD reduce the power consumption ? Cutting memory shouldn't hurt performance . I want to see benchmarks . What I really want to see is a review site test these cpus the way AMD did befor release . Benched with other apps running along side the GPU while gaming . I can go get the utube video . Not 1 review duplicated AMDs test in reviews. My neighbor has one and these things are junke if you run 1 compute programm along with a game . Nothing like what AMD said . But I will take your word for it even tho I see no benchmarks . Just text in a forum

 

[podspi]

Link and Benchmarks please. How did AMD reduce the power consumption ? Cutting memory shouldn't hurt performance . I want to see benchmarks . What I really want to see is a review site test these cpus the way AMD did befor release . Benched with other apps running along side the GPU while gaming . I can go get the utube video . Not 1 review duplicated AMDs test in reviews. My neighbor has one and these things are junke if you run 1 compute programm along with a game . Nothing like what AMD said . But I will take your word for it even tho I see no benchmarks . Just text in a forum

What model does your neighbor have? I imagine the performance delta between an A4 and an A8 is pretty large in multitasking scenarios. Also, I can say from experience that the C-class does not perform like that. Honestly it performs like my C2D in battery mode (clocked @ 1ghz).

Which isn't that bad, honestly. But nothing like AMD's youtube video of the A8 vs SB (forget the model).

 

[gdansk]

My neighbor has one and these things are junke if you run 1 compute programm along with a game

And what are you comparing it to? At the AMD C-series power envelope and die size it should be compared to a single core Atom processor. I happen to have a few Bobcat (E-350 and C-50) devices and a Atom netbook (N550). The superiority of Bobcat is pretty apparent, especially when trying to watch HD video or play (older) games. Also, thanks for the link earlier. I had heard nothing of Intel's plan to move to a new notebook target.

As for the original claim that a version of Trinity will use 17.5W... Trinity is Bulldozer-based which means even if it as fast as the K10.5 in Llano it should have better resource sharing. By simply sacrificing a bit of clock speed on the CPU, GPU and the memory, it should be possible. The question is now, how much? I really hope they bring in some process improvements to make the impact even less, but... AMD should just really be focused on getting its products to the markets at this point.