Toshiba Satellite P755D: Nearing the End of the Road for Llano

Gaming Performance

While the CPU half of Llano has always been a substantial bottleneck, the GPU half is able to do wonders to make up for the CPU's shortcomings. Unfortunately something I've experienced anecdotally that doesn't manifest itself on our benchmark charts is just how great the difference is between the Radeon in an AMD APU and Intel's Sandy Bridge IGP. Intel's HD 3000 doesn't ever look completely awful on charts, but in practice I've found it to have a tremendous amount of peaks and valleys that make it less ideal for even casual gaming than Intel would have you believe. Worse, I've gotten the distinct impression that geometry continues to be a very serious weakness for Intel.

The net takeaway here is that, performance numbers notwithstanding, Llano's graphics hardware actually provides a significantly more fluid and responsive experience the Intel's HD 3000 overall.

Now this is interesting. What you're essentially seeing is the IGP version of the Radeon HD 6650M beat it in certain benchmarks. This massive difference can likely be attributed to two things: the older (and essentailly non-updateable) Radeon drivers on the Sony Vaio Z2, and the Z2's HD 6650M being bottlenecked by the Thunderbolt interface to PCIe x4 speeds. (That laptop has to transfer the frames back to the integrated display over Thunderbolt, which appears to hurt performance.) The ultimate result is that even though the Z2 should trounce the Satellite P755D's A8-3520M on paper, in practice things aren't so clear. The VAIO SE on the other hand puts in a better showing in pretty much every game with the HD 6630M—just as you'd expect.

For academic purposes (and to prep our charts for Trinity), I've also run our Mainstream class benchmark suite.

Unfortunately it doesn't really matter what you want to point the finger at, be it low memory bandwidth or just a weak GPU overall, the A8-3520M just isn't capable of pushing 900p gaming at playable framerates. Our best case scenario is Portal 2 on Valve's aging Source engine, and even that game is just on the cusp of playability. Llano remains fine for entry level gaming, but any more stress than that is ill advised.

You may or may not have noticed we don't have any testing results for Batman: Arkham City. The reason for that is a simple one: the legitimate license key that I've been using has basically been locked out. I ran out of activations, went through their arcane key revoking process, and the best part? When I try to revoke a key on a system I'm done testing, SecuROM actually tells me I've revoked the key too many times! So congratulations to Warner Brothers Games, you have successfully prevented a legitimate user from playing your game or letting it get any kind of positive exposure.

We're currently working with NVIDIA on getting keys blessed with an arbitrarily large number of activations so that we can continue to do our jobs, so if that comes to pass, results will be added for the P755D. It does bear mentioning that the driver developers on NVIDIA's mobile team are actually having the exact same problems with Arkham City's DRM that we are (thus the impetus for them to press WB for keys without such draconian limitations), so if the people who are supposed to be optimizing their hardware and software for your game can't even actually run it, clearly your DRM scheme is a smashing success.

And by "smashing success," I of course mean "complete and utter failure." Here's a novel idea: Steam already has DRM in place so that legitimate Steam users can only play a game on one PC at a time—log in elsewhere and you get logged out. And really, who cares how many times people install a game? If someone is looking to pirate your game, giving them an arguably better experience for doing so only adds to the desire to avoid legitimate channels. This stinks of the same profit taking that's bringing about such useless innovations as Blu-ray's Cinavia protection.