It's fairly simple. Poor scaling in COD Ghosts , Batman: AO + AC IV and no crossfire profile for Guild Wars 2. Stating that those games are "weird", well I dunno. I don't like COD ghosts all that much but GW2 was a massively popular game. It doesn't seem too weird to me. I mean other websites still test hitman absolution which is a fairly incredibly old game. Or Dirt 3, Dirt showdown, I could name more that are old games but for whatever reason are still tested by some sites. Then you have Hexus testing Just Cause 2. I'd say those are weird and old games. Do you think those are "weird" games? But the testing suite is what it is, and AMD saw fit to send a review card to techreport. That means AMD more or less knew what was going to be tested, so there's nothing weird about it. If AMD knows, then AMD can't say it's "weird". They have perfect working knowledge of the test suite to be used prior to sending their review cards out, i'd assume. They have to.
I mean, do you think they should do a test suite of nothing but AMD GE games and call it a day? Please. To say it's "weird" strikes me in a strange way.
Honestly, not having crossfire available for Guild Wars 2 is somewhat surprising. It was a major MMO game which sold a ton of copies. I'm sure it would do fine given a crossfire profile, but alas, it is what it is. Whether that's excusable for AMD or not excusable I have no opinion, but there's nothing weird about that. It was a good MMO which sold very well, although i'm sure to be bombarded with messages about how horrible GW2 was.
Basically, I don't see a reason to resort to excuses. Clearly the 295X2 performs very well where crossfire is working and scaling well, and cools better than I thought it would. Kudos to AMD on that. It seems to be a great card. But this scaling stuff. It is what it is and doesn't warrant excuses, really. For this, we need to look at what i've mentioned numerous times in the past. AMD's software team. They just need to step it up. Now they're definitely better than they were a year or two ago, that's for sure. But not where they need to be. If they were where they "needed" to be, you wouldn't have these major games which don't scale properly in crossfire. Or completely lacking a profile as in GW2's case. These are popular games, so you can't say that these games are unknown indie titles. Again. Place blame where blame lies = AMD's software team. Don't blame the review website. Period. If AMD wants to blame the review website, then we have to ask why AMD sent the card to them in the first place when they had PRIOR KNOWLEDGE of what would be tested. Now that's a question worth asking. If the review website is to blame, and AMD sends them a card anyway while knowing what is to be tested. What is up with that.
Of course, this is one of the things that happens with mGPU in general. Some games don't scale to the same extent and some don't scale. That's why you see some on this forum with a clear preference for single GPU. Despite the issues mentioned above with those games, AMD has improved in software over the past year for SURE (not enough though). That's the main thing to look towards with these types of dual GPU cards. Hardware seems to be sound. But I still think the AMD software team needs to step it up way more.