Does it matter? Nvidia having burned the OEMs so many times in the past (according to Charlie) with false performance claims isn't getting any traction. They have one design win.
No, not really and these are Nvidia's numbers;) However, this is a generic A15 quad cluster so if these numbers are realistic then it gives it sense of what clover trail (and the 22nm Atoms) will be up against in the next year or so.
It would matter if true and it is an interesting first comparison.
There are some tough problems in fairly comparing systems.
All of the OS have power management systems that are configured by the user and can be modified to change the benchmark scores. There are configuration files that can set up to check for benchmark programs and override user defined power settings.
The operating conditions can be different and affect the results. The device can be in airplane mode or using resources to connect to a network. Cycles spent on the net are not available for the benchmark.
I expect the K900 (smartphone) has power management settings for smart phone. I would expect the Tegra 4 tablet reference platform to have settings for a tablet (I think that implies a larger battery).
The HTC Honor 2 was able to bump their AnTuTu scores from 12,757 to 14,948 by adjusting their power setting dynamically. They watched for AnTuTu running and then overrode the power settings to run in benchmark mode.
More data points will be needed for comparison but that will not insure more wins.