As always, SiSoft Sandra’s processor benchmarks are a logical way to dive in to performance. While not a dead-accurate reflection of a processor’s real-world capabilities, SiSoft provides excellent raw data that can be used to judge how a particular part compares to its peers when asked to complete a demanding, heavily multi-threaded task.
The K10 architecture, combined with a base clock of just 1.5 GHz, results in processor performance that is well behind Intel’s Nehalem architecture, never mind Sandy Bridge. There’s really no way to twist these results – K10 was showing its age two years ago, and it is absolutely no match for the latest and greatest mobile processors in Intel’s stable.
Still, these benchmarks are just one part of a larger picture. The graphics performance of the A-Series is obviously an important part of the equation, and AMD has been careful to underline that point time and time again. Given the lack of raw processor performance, it’s no surprise that AMD would rather focus on graphics, but this talking point can be convenient for AMD’s marketing and still hold truth. Let’s move on to the general application benchmarks to see if the story differs.
The results here don’t do much to inspire confidence, as they’re generally quite far behind the competition from Intel. The Peacekeeper benchmark, for example, shows a massive gap between the A8-3500M and the Core i5 dual-core found in the ASUS K53E. That gap becomes a chasm when compared to Intel’s Core i7 processors in 7-zip, where the Intel quad-core in the Maingear eX-L 15 nearly triples the performance of the A8-3500M.
Yet PCMark 7 offers a glimmer of hope. We’ve used this new benchmark only with the MSI GT680R and Maingear eX-L 15, two gigantic 15.6” gaming laptops with high-end Nvidia graphics. Yet the simple 14” Compal testbed was able to make a decent showing. This would indicate that the dual graphics are doing what AMD hoped – making up for lackluster processor performance with the GPU.
It’s now time to move on to the gaming benchmarks, which will highlight the success or failure of Fusion’s mobile graphics. Since this system comes equipped with dual graphics, we have the opportunity to not only the combination of the APU and the discrete GPU, but also the APU alone (by turning off CrossFire). The results are split into those that apply to the Radeon HD 6620G (the APU alone) and those that apply to the Radeon HD 6690G (the APU boosted by the discrete GPU).
The gap between Intel’s HD 3000 and the Radeon HD 6620G is significant. The Radeon HD 6620G alone managed to beat the Intel HD 3000 graphics provided by the ASUS K53E across the board, and by no small margin. The 6620G scored over 1000 additional 3DMarks in 3DMark 06 and over 13 additional frames per second in the Far Cry 2 loop.
The greatest difference, however, was Just Cause 2. Although it’s been out for over a year now, Just Cause 2 remains relatively advanced in terms of the features used by its graphics engine, and it’s never played nice with Intel IGPs – in fact, the ASUS K53E was the first Intel system with an Intel IGP that could even launch the game. This gives some indication that Intel, despite massive improvements in driver quality and feature set, still has some catching up to do in those areas.
When combined with the Radeon HD 6630M in CrossFire to create the HD 6690G, the testbed system stands on equal footing with the Acer Aspire 5551G system we tested late last year. That means most games will be quite playable at the native resolution of 1366×768, and some will be playable at high detail settings. As you can see, both Just Cause 2 and Far Cry 2 were very playable with the HD 6690G fully enabled, and the 3DMark 06 and 3DMark 11 scores both enjoyed a significant boost.