Update: Be sure to read our full review of AMD's E-350 here.

Last week I mentioned that I had recently spent some time with AMD down in Austin, TX, benchmarking its upcoming Brazos platform. The Brazos platform is composed of an AMD Zacate or Ontario APU and the Fusion Controller Hub (a South Bridge based on the SB800 series). Brazos systems will run the gamut of mainstream notebook, netbook and nettop segments ranging from $299 to around $500. While AMD let us reveal the fact that we tested Brazos, we weren't allowed to publish numbers last week. Today, we can.

I didn’t have much time with Brazos. The AMD briefing started at 9AM, but AMD wanted to go through some marketing slides and answer questions before letting us at Brazos. Going into this whole thing I was worried that I wouldn’t have enough time to run everything I wanted to run. You see, the system I had access to wasn’t pre-configured. It had Windows 7 x64 loaded on it, drivers installed and PCMark Vantage - but everything else was up to me. Despite having a 128GB Crucial RealSSD C300, installing a dozen applications and games still took hours on the system. I asked AMD if I could at least begin copying/installing some applications before we started the briefing, they gladly entertained my request.

I brought an SSD full of applications, games and benchmarks that I wanted to run on the Brazos platform. I purposefully avoided any large test suites (PCMark Vantage, SYSMark) because they would eat up a lot of time and I had no idea how long the rest of the benchmarking would take.


The Brazos test platform

I also didn’t run any of our media streaming suite. The Zacate/Ontario APUs feature AMD’s UVD3 engine and should, in theory, have similar media playback features to the Radeon HD 6000 series. Of course once we have final systems it’ll be easier to put this to the test. I was mainly interested in characterizing the CPU and GPU performance of Brazos, the two major unknowns.

I didn’t get into the full swing of testing until just before 11AM, and we had a hard stop at 5PM. That didn’t leave a ton of time, but I believe it left enough to get a good idea for what Brazos will perform like in the real world.

As I mentioned in Part 1 of our coverage, the system felt snappy. I had the 11-inch MacBook Air on hand (it served as my Excel-runner while I benchmarked) and interacting with the OS felt no different between the Brazos system and the 1.6GHz MBA. That being said, the MBA is technically much quicker (and more expensive).

AMD Brazos Lineup
APU Model Number of Bobcat Cores CPU Clock Speed GPU Number of GPU Cores GPU Clock Speed TDP
AMD E-350 2 1.6GHz Radeon HD 6310 80 500MHz 18W
AMD E-240 1 1.5GHz Radeon HD 6310 80 500MHz 18W
AMD C-50 2 1.0GHz Radeon HD 6250 80 280MHz 9W
AMD C-30 1 1.2GHz Radeon HD 6250 80 280MHz 9W

The system I tested had AMD’s E-350 processor, the highest end APU you’ll find on a Brazos. This is the chip you’ll find in $400 nettops and notebooks in the $400 - $500 range. This puts its direct competition as really expensive Atom based netbooks, Pentium dual-core notebooks and low end Core i3 notebooks. While the latter two should easily outperform the E-350 in CPU intensive tasks, the GPU comparison is another story entirely. It’s also worth noting that the E-350 carries an 18W TDP (including graphics). During my testing I measured a maximum total system power consumption of around 30W (including the 1366 x 768 LCD panel) while playing games and around 25W while encoding H.264 on the two Bobcat cores. The system idled around 15W however AMD cautioned me that this number was unnaturally high. Final Brazos systems will be far more power optimized and AMD expects numbers to drop down to as low as 5.6W.

AMD is confident we will see Brazos based systems deliver well beyond 6 hours of battery life. AMD's goal is to deliver Atom like battery life and form factors, with a real GPU and hopefully better than Atom performance. We spent our time in Austin trying to find out if its goals were realistic.

Setting Performance Expectations
Comments Locked

207 Comments

View All Comments

  • IMPL0DE - Tuesday, November 16, 2010 - link

    ...underwhelming. I was hopeful to see more performance out of this one. Let this not be the sign of the things to come. I still have faith in Bulldozer.
  • Dark_Archonis - Tuesday, November 16, 2010 - link

    If Bulldozer underwhelms or disappoints, AMD might be finished. I don't see how AMD could continue to be in business if two of their major new products are both underwhelming.

    For AMD's sake, Bulldozer needs to extremely competitive, at the very minimum.
  • wiak - Tuesday, November 16, 2010 - link

    AMD Radeon is one line you should not underestimate ;)
  • GeorgeH - Tuesday, November 16, 2010 - link

    Bulldozer is only a piece of the puzzle. AMD's killer chip is the one that seamlessly combines what will (hopefully) be a Bulldozer core's awesome integer performance with the floating point performance of a GPU, all with competitive mobile power consumption and a process node that will allow them to make a decent profit while competing with Intel on price.
  • Kevin G - Tuesday, November 16, 2010 - link

    That will come in 2012 with the successor to Llano.

    The thing that AMD needs to be worried about is getting Llano out in a time frame to be competitive. It has seemingly been pushed into the middle of 2011 instead of the early 2011.
  • Dark_Archonis - Tuesday, November 16, 2010 - link

    Great, we've been hearing that for years. First we heard that AMD's "killer product" was going to be a combination of a CPU and GPU, except they got beat by Intel on that front.

    Now, AMD's "killer" product apparently will be a Bulldozer CPU combined with a good GPU.

    It seems that AMD's "killer" product is a continually changing target, and is something that may never arrive.

    What if the competition is much stronger by the time AMD releases a Bulldozer Fusion product? What will the "killer" product for AMD be then I wonder?
  • Oxford Guy - Tuesday, November 16, 2010 - link

    Well, years back, AMD was generally crushing Intel with its crappy NetBurst line. That may never happen again, but it was fun while it lasted. The problem for AMD is that Intel has to really screw up for it to have a chance, given Intel's leverage. An absurd design like Prescott is unlikely to come again.
  • Dark_Archonis - Tuesday, November 16, 2010 - link

    That's really the point. Even if Bulldozer is very impressive, for AMD to make financial or market share gains, Intel has to stumble. By all accounts, Sandy Bridge will be a very strong product.

    As I say, Intel stumbling on the scale of the P4 Netburst fiasco is unlikely to happen again. Intel is very well known in the industry for learning from their mistakes.
  • GeorgeH - Tuesday, November 16, 2010 - link

    AMD hasn't been beaten to anything yet. CPU/GPU hybrids aren't about slapping a classical IGP together with a CPU. CPU/GPU hybrids are about combining the compute performance of a CPU with that of a GPU into a single flexible unit.

    Intel is moving towards that target with good CPUs and substandard GPUs. AMD is moving towards that target with substandard CPUs and good GPUs. If/when/where they meet is the point where AMD has the potential to deliver an awesome product.

    Sandy Bridge, Llano, and the like are only the first few baby steps towards the goal, and it's silly to claim a "moving target" just because those steps aren't as large or as relatively impressive as we might like.
  • Dark_Archonis - Tuesday, November 16, 2010 - link

    AMD should have never had so much hype behind Fusion in the first place then, as the product that you describe won't be on the market for several years.

Log in

Don't have an account? Sign up now