Tag: SoC

Intel’s Xeon SoC to ARM wrestle

Posted on March 10, 2015 by - News, Products

arm-wrestlingIntel has lifted the veil on a new Xeon D family of processors which are the company’s first ever Xeon-based System-on-Chip (SoC).

The news is bad for ARM because it is wanted to dominate the microserver market and this package is exactly what it does not want out there.

The Xeon D line is built on Intel’s 14nm process technology and combines the performance of Xeon chips with the size and power savings of a SoC.

Intel says the Xeon D delivers up to 3.4x faster performance node and up to 1.7x better performance per watt compared to the company’s Atom C2750, which is part of Intel’s second-generation 64-bit SoC family.

Xeon D is the third generation and it’s actually based on Intel’s 14nm Broadwell architecture.

This puts Intel in the running for those customers who want low-power, high density infrastructure products. In fact Intel says that it can deliver  server class reliability, availability, and serviceability (RAS) features in an ultra-dense, low-power device.

Cisco, HP, NEC, Quanta Cloud Technology, Sugon, and Supermicro have sworn their loyalty to the chip, before all their dark gods, and are committed to building microserves based on Intel’s new Xeon D options.

This means ARM has not got much time before actual products are out there.

Diane Bryant, senior vice president and general manager of the Data Center Group at Intel said that the growth of connected devices and demand for more digital services has created new opportunities for information and communication technology,” said.

“By bringing Intel Xeon processor performance to a low-power SoC, we’re delivering the best of both worlds and enabling our customers to deliver exciting new services.”

Intel’s kicking things off with two Xeon D processors, the D-1540 with 8 cores, 16 threads, 2GHz, 45W TDP and D-1520 with its 4 cores, 8 threads, 2.2GHz, 45W TDP. These have memory controllers capable of up to 128GB of addressable memory.

They also feature an integrated platform controller hub (PCH), integrated I/Os, and two integrated 10 Gigabit Ethernet ports.

All of this is based on Intel’s Broadwell so should give a reasonable performance per watt.

Intel shows off 14nm processor

Posted on January 6, 2015 by - News

Intel-Core-MIntel has taken the wraps off its first 14nm processor, the Core M at the CES show in Las Vegas.

The CPU is designed for the ultramobile, low-power market and is an  SoC die shrink of the existing architecture.

Where it does something interesting is on the GPU side with shedloads of improvements and advances,

The video decoder block has been beefed up with dual bit stream decoders, we are expecting to see other improvements but Intel has been quiet on exactly what it has tweaked and changed to date.

Intel does appear to have clear ground ahead of it. TSMC’s 20nm is only a slight improvement on die size and power consumption, Intel’s new Broadwell cores are 37 percent smaller.

What is odd, is that although more than half the chip seems dedicated to its GPU, Intel has made no announcement about supporting Direct X12. Nvidia and AMD has announced full DX12 support for their various current hardware,

Intel claims its new chip can boost battery life by 1.5 hours, speed video conversions, and offer a whopping 22 percent improvement to 3D performance. It has been a while that we have seen a significant productivity jump and we were not expecting to see much in the way of improved CPU performance from this chip anyway.

We are expecting Intel to announce the shipment of its Cherry Trail platform today too.  This is the  14nm Bay Trail die shrink that’s been on the roadmap for a little while and is not really news.  What is strange is how Intel has been quiet about the specifics of Cherry Trail so far. We have no benchmarks, statistics, or anything.

This is odd given Intel’s commitment to tablets and that pushing the Atom down to 14nm is a significant achievement. It should mean some performance gains and it is not clear why Chipzilla is not trumpeting it more.

AMD, Intel in tablet spat

Posted on December 11, 2014 by - News

tweedledummBitter semiconductor rivals Intel and AMD are set to up the stakes in 2015 with a fresh assault on the tablet market.

Both companies are often seen as the Tweedledum and Tweedledee of the chip market, continually foraying into battles where no one actually gets hurt.

But Digitimes, which is privy to both companies’ future roadmaps, said that they’ll foray out once more in an attempt to capture some slice of the tablet pie – estimated to represent 200 million units in 2015.

Intel has been forced to provide subsidies to companies in an attempt to bolster its rather feeble market share in the tablet and mobile markets.

It will kick off the show by releasing a system-on-a-chip (SoC) device codenamed Cherry Trail, which will be built using a 14 nanometre process and may be with the world as early as March next year.  The chip will have lower power consumption and support Windows and Android operating systems, said the wire.

But AMD is not going to stand still after receiving that SoC on the jaw.  It will introduce an X86 device dubbed Nolan, and an ARM device called Amur in the second half of next year.

Future HP servers will learn from smartphones

Posted on June 24, 2014 by - News

HPHere in sunny Barcelona @ the ETSS conference, Mark Potter – HP’s CTO for its Enterprise Group was expounding his company’s vision for how the future will look for file servers.

Basically, the IT industry appears to have woken up to a technology which smartphones have been using for ages – the SoC (System on a Chip). Not revolutionary concept for the fact that Potter suggested that HP would use ARM cores in addition to Intel based processors. Given that Intel and AMD were major sponsors of this show, we’re sure he couldn’t elaborate more. Anyway, Potter was keen to pump up what HP is calling The Machine as a new era in server technology.

HP seems a bit ambivalent in its approach to The Machine because Dr Tom Bradicich, vp for Moonshot Engineering with HP servers suggested that Moonshot was the first step towards The Machine model. Except it isn’t The Machine.

Curiously, HP appears to have announced the Moonshot last month at an event hosted by our old friends – Citrix.

Well, the Moonshot is now a reality and shipping but for some reason Bradicich revealed that initially this product would only be made available through his company’s “US partners”.

Which is a bit strange because most resellers that ChannelEye was speaking to here claimed to have a US presence. So they should all be able to source the product.

Besides SoCs, the other technology which Potter was bigging up was photonics.

We’ve all heard of it before but HP seems to be suggesting that you’ll see products actually utilising photonicsavailable from HP before Q1 2015.

Which will be great because photonics are a key part to HP’s dream of The Machine.

Thanks to photonics, a server won’t have to write information to anywhere – It will simply reside in shared memory.

Sounds impressive but the major advantage to photonics is the energy saving, Potter quoted somebody else as saying that “copper is an energy sucking devil”.

But he’s got a point. Potter revealed that if cloud computing was a country it would be the fifth largest electricity consumer above the UK and just below Japan.

If The Machine really can return such massive energy savings through the use of photonics which HP is suggesting, then the ROI for customers of its channel partners will be significant.

As Potter hinted, a great way of getting enteprises – which have recently been low spenders, to update their data centres.

Intel confirms $200 Android notebooks

Posted on April 26, 2013 by - News, Products

Intel-logoIntel is working on $200 notebook designs, powered by cheap chips and Google’s free Android operating system.

Dadi Perlmutter, Intel VP and chief product officer told CNET that $200 notebooks will predominantly be Android products based on Atom chips.

Now the ball is in Microsoft’s court, if Redmond wants to gain a toehold in the ultra cheap notebook market it will have to make its operating systems a bit cheaper.

Perlmutter said the price of Windows 8 gear depends on how Microsoft prices the OS, and it could end up slightly pricier. In addition, Perlmutter pointed out prices of notebooks based on Core processors should go down to $399 to $499.

As netbooks are about to die a quiet death, $200 droidbooks might be the best way to replace them. Upcoming Atom chips are roughly on a par with ARM SoCs in terms of power consumption and performance per watt, so from a technical standpoint Intel should have everything covered.

The big question is software. Android is a good operating system for touch enabled devices, but it has yet to prove itself on more traditional form factors, that is,  notebooks.

AMD to cut reliance on PC chips

Posted on April 24, 2013 by - News

AMD, SunnyvaleAMD might be on the verge of its biggest strategic shift in ages, as it starts to embrace ARM processors and more frugal chips. The company hopes to make as much as half of its money from console chips and ultra-low power processors by 2016. That is on top of ARM-based server chips which are already in the works.

At the moment, AMD generates the vast majority of its revenue from x86 processors and GPUs, but this year it is expected to ship millions of custom APUs for gaming consoles. On Tuesday the company launched a range of embedded x86 chips based on the new Jaguar core, but it also hinted at upcoming embedded chips based on ARM designs.

AMD already made it clear that it intends to use upcoming 64-bit ARM cores in its server parts, but the decision to design embedded APUs with ARM cores could have far reaching implications. In technical terms, these chips will have a lot more in common with consumer application processors than server chips. They are also expected to feature the latest generation Radeon graphics. In other words, AMD will develop ARM based SoCs, but it is still unclear whether it will target the consumer market.

Several years ago AMD sold its Imageon mobile graphics division to Qualcomm, and Qualcomm put it to good use, churning out millions of mobile SoCs with Adreno graphics, courtesy of AMD IP. However, AMD insists that it could catch up with relative ease.

Sasa Marinkovic, AMD’s Technology Marketing Lead, told Forbes that chip architecture at the time of the Imageon sale was full of bottlenecks and it has moved along since then.

“We sold some graphics IP, but we didn’t forget how to build it,” he said.

AMD already has some x86 designs capable of hitting sub-5W thermal envelopes required by tablet manufacturers, such as the 4.5W Temash SoC. However, ARM based chips could offer even better power efficiency and end up with sub-3W TDPs.

On the console front things are looking even better. AMD expects sales of custom APUs for the Playstation 4 and next-gen Xbox to account for 20 percent of its revenue by the end of the year. Similar chips based on the Jaguar core are coming to the consumer market as well.

Nvidia reportedly working on phone and tablet reference designs

Posted on January 28, 2013 by - Gossip, News

nvidia-gangnam-style-330pxNvidia has been frantically trying to carve out a foothold in the mobile SoC market over the past couple of years and with every new Tegra generation it saw its number of design wins double.

However, that doesn’t seem to be enough for CEO Jen-Hsun Huang. According to Mobile Review, the chipmaker is planning to kick things up a notch, by directly entering the smartphone and tablet business with a bit of help from its partners.

Nvidia is said to be working on reference smartphone and tablet designs. The designs will be built by contract manufacturers, under Nvidia’s direct control. The partners are apparently small regional players, basically retail chains who already offered cheap Chinese tablets under their brands. It is unclear whether traditional Nvidia AIBs will also join the effort, but it seems more than likely, as some of them have already started making inroads in the tablet space, without Nvidia’s support.

The idea seems to hold water. Nvidia could control the feature sets and performance of reference designs, while regional players could take care of distribution and retail. Nvidia partners could end up with cheap, yet competitive devices capable of taking on first-tier devices with fancier brands. The drive is said to be scheduled for May and June, roughly Computex time, and Mobile Review’s Eldar Murtazin believes tens of 7- and 10-inch tablets based on Nvidia’s reference design could show up.

Nvidia already has extensive reference design experience and plenty of partners in the graphics market, so such an approach would be nothing new for the company. However, last year Nvidia also offered Kai, a reference tablet design based on the Tegra 3 SoC. It did not find many takers. Back at CES, Nvidia unveiled Project Shield, a Tegra 4 gaming console built under its own brand, but we haven’t seen any Tegra 4 design wins yet. The chip is apparently a couple of months behind schedule, due to some technical issues which necessitated a re-spin.

We should see more Tegra 4 gear at the Mobile World Congress and Nvidia is also expected to launch a cheaper, A9-based quad-core SoC in the latter half of the year. The reference design approach could help Nvidia gain quite a bit more design wins this time around, provided it doesn’t flop like Kai.