Nvidia released Tegra, lied about power usage, performance and availability. Due to these lies it looked great on paper and loads of customers started making designs.... it was late, over power and not as fast as they said so loads of customers got screwed BUT everyone makes a mistake. So Tegra 2 came along. Nvidia did the same thing, same result, same customers getting screwed. Tegra 3 came along, the PR train of BS was still in full swing, some customers had wised up, a few more got screwed. Tegra 4 came along, more lies, more lateness..... few to no customers got screwed, why, because Nvidia burned all their bridges, no one believed Nvidia, no one made products based on Nvidia's lies, Tegra 4 crashed and burned and cost Nvidia massively as no one was buying it... hence Nvidia making loads of their own Tegra 4 products.
Tegra K1.... who knows, they've burned all their bridges, the PR is still obviously BS, no announcements from customers on design wins despite supposedly being available in the first half of the year. Previous years such chips available in h1 would be being talked up in dozens of products at CES, k1/Nvidia products, nadda.
Even if Nvidia finally made a good product when you lie and burn all your customers for 4 years straight, no one actually wants to work with you even if you have the best product, more importantly, no one at all believes your numbers.
Anandtech did a Tegra 3 or 4(forget which) preview which showed competition beating numbers but with no power.... months later it turned out Nvidia burned Anandtech, used their relationship to have Anandtech push out a BS article comparing sub 2W apple/other chips vs a 10W version of Nvidia's chip... yeah, no **** it was faster.
Anandtech REFUSED to do a k1 preview because Nvidia was unwilling to tell them how much power their test system used... fool me once, shame on you, fool me twice, shame on me. Yet another person they work with lied to who now won't be their shill any more.
Aside from the fact Nvidia made the claim about console level gaming for both Tegra 2.(in passing comments rather than numbers, really, and had Carmack suggest as much in a video... which is probably why Carmack is calling shens), and actually for Tegra 3 they used numbers to show it was console competitive. We then have the k1 actual numbers. Different slides in which in some it says "this chip is X powerful" and in other slides it says "this chip uses 2W"..... but with no indication these are both the same thing. This is exactly what they did to Anandtech, they had out numbers this chip can achieve(at 10W + ) and tell everyone how much power it will use when running in a phone (2W) and never actually physically say, these two things are impossible to achieve together.
Take any Apple/Qualcomm chip, stick a heatsink that can dissipate 20W on it crank the clocks up 30% cpu 100% gpu, and have zero throttling and you'd get 3-4 times the performance AT LEAST. Nvidia is comparing this unregulated, fully clocked, insane power version to the competitors throttled down clocked phone versions.
For instance the gpu Apple has in their current best chip runs at essentially half the full clock the chip was designed for, and half the gflops, and will certainly throttle inside a phone when under TDP limits and almost no cooling. But that gpu by design can go easily twice as fast and unthrottled, no problem being several times faster.
