With the announcement of the Tegra K1 at last year's CES, Nvidia seemed to have pulled of the impossible. The Tegra K1 was the first of its kind in mobile chips to bring desktop grade graphics performance to the mobile platform, courtesy of the 192 Kepler GPU cores present on board. Supplemented by the ARM 64-bit Denver based CPU, the Tegra K1 was a chip that seemed to have it all on board. The first main stream market device to feature the Tegra K1 was Google's Nexus 9, launched last year and designed by HTC. With the launch of the Nexus 9, it was clear that mobile graphics and computing would never be the same.
Things were taken to the next level however, with Apple's launch of the iPad Air 2 in September. With the iPad Air 2, came Cupertino's very own A8X SoC. The A8X at the time of the iPad launch was claimed by Apple to offer 40% increased CPU and 2.5% increased GPU performance over its predecessor, the A7. With both the devices in the open, pretty soon the benchmark war started and we began to receive results showing which processor had the clearest lead when it came to performance and graphics. You can take a look at our benchmark comparisons for the A8X and the Tegra K1 here, here and here.
But that's all about the Tegra K1 vs the A8X. Nvidia has announced the Tegra X1 mobile SoC today at the CES in Las Vegas, and with it, the company seems well poised to take the mobile computing platform to another level completely. The Tegra X1 comes with 256 Maxwell cores on board which not only increases the amount of cores found on the Tegra K1 but also ends up delivering twice the performance efficiency and 40% improved overall performance.
Not to mention the Tegra X1 is manufactured on the 20nm process, which will improve the power efficiency for the chip, which is something that had been a thorn in the Tegra K1's side. With all these specifciations, the question remains, how do these numbers and facts translate into performance? Well, hardwarezone has managed to capture Nvidia's official benchmarks for the Tegra X1 and we're here to show you just how vast the performance difference really is on Nvidia's latest.
Starting from Manhattan and 1080p Manhattan Offscreen, the Tegra X1 manages to get overall scores of 3,391 and 4,077 respectively. FPS scores for the SoC come out to be 65.8 for 1080p offscreen and 54.7 for Manhattan. When we compare these to earlier Apple iPad Air 2 and A8X benchmarks the difference between the two is astounding.
The Tegra X1 literally doubles the A8X's scores when the two are put together. The scores for another benchmark testing for the A8X found here, also end up solidifying this fact. Nvidia's Maxwell architecture on a mobile GPU simply does not have a parallel in the mobile SoC market right now. Moving on from Manhattan Off and Onscreen to T-Rex, the results end up solidifying the Tegra X1's lead once again, though not in such an impressive manner as above. Take a look at the A8X's T-Rex scores below.
The Tegra X1's T-Rex Onscreen and Offscreen scores stand out to be at 59.8 and 124.2 fps respectively. Compare these to the A8X's 53 and 72 fps above and you'll see the difference for yourselves once again. Looks like the Maxwell's Streaming Multiprocessors and the 256 CUDA cores on the Tegra X1 end up making all the difference in the world when it comes to mobile graphics processing. The Tegra X1's Render Output Unit and Texture Mapping Units also receive an increase to 16/16 from the K1's 4/8 and the difference made by this increased pixel mapping is evident. Another area where the Tegra X1 doubles the A8X's score is 3D Mark's IceStorm Unlimited, where the A8X and the iPad Air 2 scores 21655.
With the 20nm manufacturing process, not only does the Tegra X1 bring increased performance, but the SoC can also be expected to perform much more efficiently than before. We've managed to get our hands (courtesy of hardwarezone once again) on a graph comparing the power efficiency of the GPUs on the Tegra X1 with the A8X and you can see the difference for yourselves below.
GPU Power consumption is another area where the Tegra X1 almost doubles its lead over the A8X. With these increased GPU consumption readings, and Nvidia's ability to finally bring Maxwell to mobile computing, things have moved very much ahead in the mobile computing industry then previously thought of. So let us know what you think in the comments section about the facts above.
The real question now remains is that how many devices will end up adopting the Tegra X1 for their CPU, which is an area in which its predecessor didn't do that well. Nevertheless, the chip packs a heavily solid puch when it comes to graphics and computing, and with the average power consumption of 1.4 W as shown above, looks like Nvidia has make its mobile SoCs power efficient as well. All eyes lay on Apple and the rest now. Cheers.