Conclusion: Llano Brings A Lot Of Potential To The Table
Indisputably, the GPU is the most crucial component of AMD’s Fusion initiative, and Llano will stand or fall according to its graphics ability. Fortunately, the A-series APU really does bring discrete-class graphics to the table, and the economy of combining these components can’t be denied.
For as much as we would have preferred a desktop part, it’s no surprise that AMD gave us a notebook to test first. The associated power savings of a single-chip solution are very attractive in this application. We’ve already seen the advantage of adding graphics to AMD’s low-power C- and E-series APUs in the netbook space. Llano will leverage the same benefit in the more performance-driven laptop market. Frankly, there’s not much out there in the $500-700 range with discrete-class graphics. And Intel’s Sandy Bridge architecture actually does a fair job with its HD Graphics 3000 engine. But Llano blows that away in 3D workloads, using less power in the process. If the suggested prices are correct (and that’s a colossal if; laptop vendors will determine pricing, not AMD), then an A-series laptop for as little as $500 could churn through a 3D game with decent performance. That’s an exciting prospect that could have a tremendous impact on the landscape. Of course, our recommendation has to be to wait until notebooks are actually available. What manufacturers do with Llano is almost as important as the potential AMD hands to them.
When it comes to the desktop space, Llano’s prospects are decidedly less impressive in light of the competition. These APUs make for an ideal solution to replace entry-level PCs with crappy integrated graphics. And, they certainly could introduce a lot of graphics muscle to a segment historically light in that regard. If Llano catches a foothold there, the APU could impact peoples’ expectation of what a PC can do. Developers might start targeting a higher lowest common denominator in their games, and that’d of course be great news for PC gaming.
But once you reach outside of the budget basement and consider folks willing to use discrete graphics, the A-series’ utility is hamstrung. It’s easy to put an $80 Radeon HD 6670 in a cheap OEM box and walk away with something that easily trumps AMD’s product in both processing and graphics benchmarks.
If Llano has an Achilles’ heel, it is the relatively old Stars CPU architecture. Obviously, this is where a comparison to Intel results in humiliation by Sandy Bridge. Our A8-3500M APU is the only 35 watt quad-core model, and we hope that this restricted power ceiling is responsible for the disappointing results in processor-oriented benchmarks. We never saw this CPU come close to approaching the performance of a 2.4 GHz Phenom II X4, which is what we would have expected, at least in single-threaded tests like iTunes and LAME, where Turbo Core should have kicked in. Perhaps AMD should consider removing Turbo Core as a marketing bullet point on SKUs destined to never enjoy its benefit.
AMD showed us a number of interesting applications in development expected to take advantage of its APP initiative. But it’s going to take a lot more than potential to make a general-purpose approach to programming for graphics resources successful. Just ask Nvidia. CUDA has been around for longer, Nvidia seems to throw a lot more resources at software developers, and arguably the most exciting win we’ve seen on the desktop is CUDA support in Adobe’s CS5 suite. I’m not sure what it will take (Ed.: I am: money), but a broad install base of APUs certainly won’t hurt.
And then there’s the future to consider. Yes, AMD’s Stars design is being retired in lieu of Bulldozer, and Llano’s successor, code-named Trinity, will cough up the old CPU design and take the newer architecture instead. But Bulldozer remains a question mark; we have no idea how it will perform yet. Intel isn’t sitting on its hands either, and the 22 nm Ivy Bridge die shrink should be in volume production by the end of this year. Llano is AMD’s first 32 nm part, and representatives from Globalfoundries indicate that AMD won’t be on a 20 nm node for two more years. Fourteen nanometer manufacturing won’t be online until two years after that. And this is if AMD’s predictions are accurate. The company has a habit of pushing back manufacturing advancements beyond initial expectations.
On the other hand, graphics is a perpetual thorn in Intel’s side. We’ve seen the company try, try, and try again, always deemphasizing the importance of graphics in the shadow of its CPUs. This is what AMD appears to be exploiting today. Yes, Intel’s processor is superior. But the real question is: would you be more inclined to notice AMD’s slower CPU or Intel’s slower graphics engine? Would it be AMD’s longer battery life during graphics-intensive tasks that caught your attention, or Intel’s stronger application performance?
As is so often the case, there is no right answer to those questions. It’s a matter of individual usage patterns. To be frank, most folks aren’t going to be able to distinguish between a Sandy Bridge-based Core processor or a Llano-based A-series chip when it comes to Web browsing or composing a document in Word. Then again, if you fire up a game, the A-series APU surges past Intel’s HD Graphics implementation. Play that same title on a mobile system using battery power and you’re treated to another surprise: improved performance is complemented by longer battery life.
Intel addresses processing-intensive workloads more adeptly. Quick Sync makes transcoding much more viable on a mobile platform, too. There’s no getting around the superiority of the Nehalem architecture over Stars. But, at least in the notebook space, graphics ability and battery life are going to be priorities for most folks. The fact that 3ds Max doesn’t run as well on Llano is of less consequence.
So, what’s the final verdict? Llano has the potential to introduce impressive graphics-oriented value to the low/mid-range market; this APU easily outclasses Intel HD Graphics 3000. Its power-saving advantage is definitely appreciable in the notebook space, and we’re hoping that more generous desktop thermals give the processor side of these chips a little more room to be competitive. At the same time, Llano will forever be squeezed into the lower echelons of mainstream on the desktop by a superior Intel CPU matched up to faster discrete graphics cards. This doesn’t matter much to the folks chasing low prices, but we’ll have to wait for Trinity to see if AMD can come up with an APU that can simultaneously challenge Intel’s processing capabilities as it wipes the floor with its on-die graphics. A no-compromise part would almost certainly guarantee more success than the give-and-take that is Llano.
....big win there...
Ditto on the "Good Job AMD" definitely on the right track.
The NDA is up on the 30th.
1) What happened to the Game Charts results for the Radeon HD 5570, when the games were benchmarked? I thought you made a point to say you were going to compare the APU's 6620G with a discrete card (that has the same number of SPs and same clock). So much for that, unless you thought only comparing the two with a synthetic test was enough. Oh well. Tom's can be such a tease!
2) I'm just a little disappointed that the APU's graphics power was not able to double Intel's.... Under the best of circumstances, AMD's latest integrated graphics came close to being twice as fast, but i guess that is ok since we are not playing horseshoes. I just thought it would be nice if it had made a nice even doubling, or more. Now, i'm worried IVY BRIDGE will beat it....
we ARE playing horseshoes...
and i have to give credit where credit is due: props to AMD for almost doubling Intel's HD Graphics in the integrated space....