Japan 5G network tests Arm chips, claims power draw down by 72%

NTT Docomo and NEC have conducted tests that demonstrate Arm-based Graviton2 processors consume 72 percent less power compared to X86 chips while running NEC's 5G core software as part of the NTT Docomo 5G network. The two companies said they have completed a trial to test the energy efficiency and performance of AWS Graviton2 …

[Anonymous Coward]

"compared to X86"

Why would you use x86 to boot? The Japan of the 80's is long gone.

[Kevin McMurtrie]

I don't understand this article. Is NTT and NEC trying to run 5G protocols on general purpose CPUs and motherboards rather than 5G networking SOCs? That does seem like it would run hot.

[Justthefacts]

Open RAN

Yes, it’s a key part of 5G. The reasoning is: being able to run at least the Core Network entities, and also the protocol and upper layers of what used to be known as the RAN, on general purpose hardware makes it much cheaper than buying special purpose hardware from what was basically a triopoly: Ericsson, Nokia, Huawei. Most of the network is now being virtualised. From the operator point-of-view, this network architecture change is going to be what drives cost-savings over the next decade, and is the important part of 5G, while the use of 60GHz radio is really a sideshow.

By the way, it’s not the really specialist radio processing algorithms being accelerated, those are still in specialist hardware on the NodeB.

[Anonymous Coward]

Re: Open RAN

"From the operator point-of-view, this network architecture change is going to be what drives cost-savings over the next decade, and is the important part of 5G"

I'm a big believer in open standards based architectures especially where multivendor interoperability can cost effectively deliver real benefits. But who benefits here and what are those alleged benefits?

Cost savings? Or is it more likely to be vendors looking for increased kit sales?

Will it lead to lowered end user costs? Doubt it. That was promised with 3G and 4G wireless/mobile and even with Fixed Wireless Access broadband, What benefits actually were there in the UK or elsewhere?

Will it lead to service improvements e.g. better and more reliable coverage? Doubt it (the smartmeter networking fiasco might have been an ideal opportunity for joined up thinking, but we don't do that kind of thing here).

Will they lead to disappointments in unrelated areas, e.g. continued decrease in quality of, and ultimately the end of, free to air broadcast TV across the UK? I expect so - the nice people at Ofcon have flogged off some of the current digital broadcast bandwidth (e.g. where COM7 lived?) to be used for 5G instead (whatever 5G might mean in the real world).

The monopolies of concern in this picture don't look like Ericsson etc, not from where most people are anyway.

[druck]

% less power

This is great news, but what is 72 percent less power? Is it using 72% of the power that the x86 uses, or actually 72% less which is 100-72 = 28% of the power of the x86? If its the latter, why not just say 28%, or is this yet another Americanism?