I love how the crossfire config is used as a "case in point", when crossfire/sli have always had frame time issues, and are heavily
dependent on application profiling. Perfect bait and switch here, not to mention the card it's being compared against is the $1,000 Titan. Of course it's going to turn out this way, but are you going to spend $1,000 to get that experience? I'm sure as hell not.
Now when we read the aforementioned article in full context we see:
When I asked Nvidia about this game, they said they believe there may be some problems with the application itself.
This pretty much validates my point about frame latency issues being primarily software/driver based. Question: Is the 7950's hardware still being considered deficient in Borderlands 2? That's what you're insinuating anyway, when it's quite obvious that scenario was driver dependent. Do you know what REALLY is hardware deficient? AA, compute, resolution, and maybe even tessellation in the 660 Ti. There is a real
deficiency there with a crippled memory bus. Framerate benchmarks show that, latency ones don't. Frame rates aren't just important, they are vastly
more important than frame times, as there is less chance for bias and config rigging. I'm not saying frame times aren't useful, but they need to be taken in proper perspective, as there are way more outside variables that effect frame time than frame rate. I agree that there is some hardware connection to frame times, like Fermi memory limitations, or Amd's VLIW cards tessellation performance, but for the most part frame times are software / driver issues, and TR is trumping them up WAY more than is necessary outside of brand bias. AMD isn't going to win a competition based on drivers, so we might as well call it right now for Nvidia. Nvidia wins. Forever.
I also like how the Fermi issues in BF3 were brought up, but nowhere to be seen in the 7950 / 660 Ti report. They didn't exist. The Fermi issues might have been from the card running out of memory, but since none of the larger 2-3gb memory cards were tested, we'll never know. Lack of perspective is a serious problem, which is a good reason to totally disregard most testing based on frame times. Frame times complement frame rates, not the other way around.
The product that works best now must be assumed to be superior until proven otherwise.
*COUGH* Tomb Raider *COUGH*
Frame Times are 100% driver dependent with new releases.