Loved the article, I've been waiting on this for about a month now. I'm VERT impressed by the compute potential of this card. It matches the GTX770(a 330$ card) in compute tasks. And it's amazing how a card with only 640 cores is matching a card with 768 cores with 50% faster memory. Makes me wonder if I can hold off on upgrading my 780 when the 880 comes out. Hopefully it's not THAT much faster. But I have a feeling it will be.But I think Tom's messed up ONE thing pretty badly. When your testing a card, your suppose to use graphics settings optimized for THAT card, so you can get realistic results of what a user might achieve with it. Not for the best card in the whole round up. This card only has a 128bit memory bus, yet Tom's used AA in almost every benchmark accept for Metro LL. And I can't even use SSAA in Metro LL with my 780, otherwise Tom's would have used it. Some of the benchmarks ran with 4xAA! Who in their right mind is going to be using 4xAA on a graphics card that doesn't even need a 6 pin plug?The memory subsystem on this card is it's main bottleneck. Why Tom's chose settings that would bottleneck that card I have no idea.It matches the 650Ti Boost right now, just think if it had a 192bit memory bus like the Boost does. It would be faster than the GTX660. Now that's impressive.Take this same card and benchmark it with no AA, or at the VERY most 2xAA if you REALLY need it, and watch this card move up those charts considerably. It would be ahead of the 650Ti Boost for sure, but still behind the 265x, but not by nearly as much.Using 4xAA on a 150$ card lol, that's not right. I use Ultra settings and 2xMSAA when I play Battlefield 4 on my GTX780@1920x1080. Yet he's using 2xMSAA and Ultra on this little spec of a graphics card. I just don't get it. At least use Ultra with no AA, or maybe High with 2xAA. The settings he did lower don't make a significant difference in performance anyways so IDK why he lowered those and left the rest high.