Wednesday, July 2, 2014

Nvidia Replaces 64-bit Denver Chip With Server Development

Nvidia was previously planning on developing a 64-bit CPU processor but, according to recent reports from the company, has decided to abandon that project for servers. To make matters worse, this announcement comes just three years after the company promised that it would build such a chip. According to Vice President of Accelerated Computing at Nvidia Ian Buck, "That's not something we're doing today." Instead the company is focusing its latest 64-bit Tegra chips on mobile and embedded devices, according to Buck.

Back in 2011, Nvidia announced that it would be creating a 64-bit ARM-based chip named Project Denver. The chip was slated to go into mobile devices, PCs and servers. Nvidia stated that it was looking to pair Tegra chips with GPUs in servers, though those plans have yet to come to fruition. The company's latest Tegra K1 chip, which includes a 64-bit CPU and is slated to ship later this year, was created for smartphones, tablets, cars and other products.

Nvidia has previously stated that the mobile K1 chips could very well make their way into microservers, though the company will not be developing a specialized server CPU. Moreover, Nvidia's departure from this field only leaves four ARM server chip makers left: Advanced Micro Devices, AppliedMicro, Broadcom and Cavium. Nvidia has always been known for its graphics cards but has instead shifted focus to building Tesla high-performance GPUs for ARM servers. "We should continue to focus on building great GPUs for them," Buck added.

The interest in ARM servers has rapidly been growing for web hosting and cloud computing in addition to being a low-power alternative to the dominant x86 servers. However, there is also some doubt surrounding ARM servers, especially do to the fact that they are still being tested. In addition to that, the platform software has yet to fully mature and ARM server pioneer Calxeda shut down late last year due to a lack of funds. As a result, a lot of questions were raised about the viability of the products themselves.

AMD is betting its future in servers on ARM chips while AppliedMicro is taking a more cautious approach. According to the company, it's plan is to have its first few chips help the company gauge interest in ARM servers and then determine whether or not to continue based off of that data.

Only a small number of 32-bit ARM servers are currently available from vendors like Boston Ltd. and Mitac. The first 64-bit ARM servers from Cirrascale and E4 Engineering are slated to ship sometime this year with servers running AppliedMicro's 64-bit processors and Nvidia's Tesla graphics cards. At this time, none of the companies have offered pricing details.

Servers that are running Nvidia's Tegra chips have been developed in the past with supercomputers at the Barcelona Supercomputing Center in Spain being based off Tegra 2 and 3 to test the energy efficiency of ARM chips. However, that project was replaced by a supercomputer running ARM-based Exynos smartphone chips developed by smartphones.

It is still unclear as to how Nvidia dropping its Denver chip project will ultimately affect the business, especially considering the fact that Nvidia went back on a three-year promise by doing so. Server business is doing fairly well right now so Nvidia is probably banking on that to come through for them.

Need To Rent A Server? Try 800-736-8772

No comments:

Post a Comment