All News | Boards | Chips | Devices | Software | Archive | About | Contact | Subscribe
Follow LinuxGizmos:
Twitter Google+ Facebook RSS feed
(get email updates)

Nvidia unveils Tegra K1 with 192 GPUs, preps 64-bits

Jan 6, 2014 — by Eric Brown — 1,747 views
Tweet about this on TwitterGoogle+Share on FacebookShare on LinkedInShare on Reddit

[Updated Mar. 17, 2015] — Nvidia unveiled its Tegra K1 SoC, which adds a 192-core Kepler GPU to a Tegra 4-like quad-core Cortex-A15 design, and it’s also working on a 64-bit “Denver” version.

At the Consumer Electronics Show in Las Vegas, Nvidia showed off a prototype of the Tegra K1 system-on-chip, its fifth-generation Tegra design. To emphasize the Tegra K1’s revolutionary advancement, Nvidia broke with its typical naming scheme, and promoted the product by creating a crop circle in a field in Salinas, Calif., with clues to the design written in braille. The 32-bit version of the Tegra K1 will be very similar to the Tegra 4, but with one huge exception: It swaps out the 72-core GeForce GPU for a Mobile Kepler GPU that integrates 192 CUDA graphics cores, making it the most advanced mobile GPU around.

Mobile Kepler was announced by Nvidia last July, back when the Tegra K1 SoC that most expected to be called the Tegra 5 was referred to as Project Logan. As a result, the real news here is that the Tegra K1 will also be released in a 64-bit “Project Denver” version.

Nvidia did not mention ship dates, but according to an Engadget report, the Tegra K1 will ship in phones be sold by AT&T and Vodafone by the end of the second quarter, together with a separate LTE chip. It was unclear whether that timetable applied to the pin-compatible 64-bit Denver version as well as the 32-bit design. One of the Tegra slides suggests Denver won’t hit until 2015, and will feature a next-generation “Maxwell” GPU, as well as a Tri-Gate FinFet CPU design.

On the Tegra K1 product page, Nvidia lists specs only for the 32-bit version, which aside from the new GPU, appears to use the same 4-Plus-1 design as the Tegra 4. This combines four 28nm-fabricated Cortex-A15 cores clocked at 2.3GHz, up from 1.9GHz on the Tegra 4, along with a fifth power management core. As before, it supports HDMI resolutions up to 4K UltraHD, and it doubles DDR3 RAM support to 8GB.

The Denver design will use a dual 64-bit cores clocked to 2.5GHz, and will offer 7-way superscalar technology instead of the 32-bit version’s 3-way superscalar processing.

Whereas the 32-bit version offers two 32KB L1 caches, the 64-bit design offers one 128KB cache and one 64KB cache, according to Nvidia slides reproduced by ZDNet. Presumably, the 64-bit cores are variations on ARM’s 64-bit Cortex-A53 design, although no more details were forthcoming.

Tegra K1 simplified (left) and detailed function diagrams

The 32-bit Tegra K1’s Mobile Kepler GPU is a variation on Nvidia’s desktop-level Kepler engine. The 192-core GPU offers better graphics performance than the Nvidia GeForce-driven PlayStation 3, and is even more powerful than Nvidia’s desktop-oriented GeForce 8800 GTX graphics card, according to the company. Mobile Kepler also adds a new low-power inter-unit interconnect and mobile optimizations that greatly reduce power consumption, says Nvidia. Last summer, Nvidia claimed the 192-core GPU used a third of the power of the Imagination Technologies PowerVR SGX 554MP4 GPU in the iPad 4.

Tegra K1 block diagram showing interface connections

Mobile Kepler and the Tegra K1 also provide unprecedented API support compared to earlier Tegras, including support for Khronos’s new OpenGL 4.4 graphics specification, as well as OpenGL ES 3.0 and DirectX 11. Various eye-witness reports from the CES presentation said there were stark differences in Nvidia’s side-by-side graphics comparison with the Tegra 4.

Nvidia Shield v1

As had been previously rumored, the Tegra K1 will be used in the second-generation Nvidia Shield game console. Other news tipped at the Nvidia event is that Epic Games will port the Unreal Engine 4 to Tegra K1 for game development.

Tegra K1’s role in automotive

According to CNET, Nvidia CEO Jen-Hsun Huang used the Tegra K1 event to discuss an upcoming G-Sync display technology that will help reduce lagtime in gameplay. He also promoted the Tegra K1 as an automotive processor, and mentioned an upcoming Tegra K1 VCM (vehicle computing module) reference design.

The VCM technology is said to support advanced driver assistance with pedestrian detection and blind spot monitoring, as well as other semi-autonomous driving features. He also mentioned a “Project Mercury” technology that will help industrial designers build digital dashboards for cars.

These designs will likely comply with Google’s newly announced Open Automotive Alliance, which ultimately aims to standardize Android automotive computers. Since Nvidia is the only other member along with a quartet of automakers, the initial spec will presumably be based on a Tegra K1.

Further information

More information on the Tegra K1 may be found at Nvidia’s Tegra K1 product page, as well as this more marketing oriented Tegra K1 page.

(advertise here)


Please comment here...