Nvidia provides more details about its 64-bit “Denver” cores powered Tegra K1 chip
Nvidia’s latest chipset — the Tegra K1 — has been winning the company accolades from all reviewers because of its unparalleled GPU performance. However, instead of basking on its laurels, the company is already working hard on its next chip, a 64-bit version of Tegra K1 with two-custom “Denver” cores.
At the HOT CHIPS event, the company unveiled more details about its upcoming chip, which is slated to be among the first 64-bit chips to be available for Android devices.
Nvidia states that its custom-made Denver CPU in the 64-bit version of Tegra K1 will come with a 7-way Superscalar microarchitecture, 128KB of L1 instruction cache and a 2MB 16-way L2 cache for both cores. Other highlights of the architecture include Dynamic Code Optimization, low latency power-state transitions, extensive power-gating and voltage control and more.
Most importantly, the 64-bit Tegra K1 chip is pin compatible with its 32-variant, which makes it easier for OEMs to integrate it into their products.
Nvidia also states that the Denver cores can be clocked as high as up to 2.5GHz, and can rival desktop class CPUs in certain workloads “at significantly reduced power consumption.” On the GPU side, Nvidia is still using its 192-core Kepler GPU, which is already found inside the 32-bit Tegra K1 and is being hailed as a revolution in mobile GPU by many.
Nvidia states that the first batch of mobile devices based on its 64-bit Tegra K1 chip should release later this year. It also specifically states that it is already developing the next version of Android — L — on its latest chip.