An interesting slide that Nvidia slotted in its presentation of the Tegra 4i was the CPU architecture comparison between the Tegra 4 A15, the Tegra 4i and Qualcomm’s Krait 800, its top of the range system-on-chip which was launched earlier this year at CES in Las Vegas.
Unsurprisingly, Nvidia puts its own chips in a positive light. The 800, which is expected to appear in retail products in mid-2013, is expected to have a performance of 0.33 SpecInt per MHz and at 2.3GHz, hitting 762 SpecInt. When comparing the core size, that’s a perf per mm^2 of 293.
The 800 is comfortably overrun by the A15-based Tegra 4 which runs “only” at 1.9GHz and manages to hit 1168 SpecInt thanks to a superior IPC (almost twice that of the S800, according to Nvidia). Even the “mainstream” Tegra 4i has a higher absolute Max Perf compared to the S800 at 920 SpecInt.
The gulf between the respective chipsets is even greater once we look at performance per unit area with the Tegra 4i hitting 800 SpecInt per mm^2, ahead of the Tegra 4 and the S800 at 432 and 293 respectively. Still those are synthetic tests and have yet to depict anything in real life. All the theoretical benchmarks are per core and all three SoCs have four cores.Leave a comment on this article