Vik

Solving the Automotive Bandwidth Problem: Aquantia Partners with NVIDIA for 10GbE
  • Comments are Closed

Solving the Automotive Bandwidth Problem: Aquantia Partners with NVIDIA for 10GbE

One of the lesser known topics around fully autonomous vehicles is one of transporting data around. There are usually two options: transport raw image and sensor data with super low latency but with high bandwidth requirements, or use encoding tools and DSPs to send fewer bits but at a higher latency. As we move into development of the first Level 4 (near autonomous) and Level 5 (fully autonomous) vehicle systems, for safety and response time reasons, low latency has won. This means shifting data around, and a lot of it. 


Bandwidth required, in Gbps for raw video at a given resolution and frame rate, also at a specific color depth. E.g. 720p30 at 24-bit RGB (8-bit per color) is 0.66 Gbps

Raw camera data is big: a 1080p60 video, with 8-bits of color per channel, requires a bandwidth of 0.373 GB/s. That is gigabytes per second, or the equivalent of 2.99 gigabits per second, per camera. Now strap anywhere from 4 to 8 of these sensors on board, the switches needed to manage them, the redundancy required for autonomy to still work if one element gets taken offline, and we hit a bandwidth problem. Gigabit simply isn’t enough. 

The announcement today is two fold: NVIDIA and Aquantia are announcing a partnership that means Aquantia based network controllers and PHYs will be used inside NVIDIA’s DrivePX Xavier platform, and subsequently the Pegasus platform as well. The second announcement is the new automotive product stack from Aquantia, AQcelerate, consisting of three chips depending on the automotive networking requirement. 

Aquantia AQcelerate for Automotive
  Type Input Output Use Case Package Size
(FCBGA)
AQC100 PHY 2500Base-X
USXGMII
XFI
KR		 10GbE
5GbE
2.5GbE		 ADAS
Cameras
Parking Assist
Sensors
Telematics
Audio/Video
Infotainment		 –
AQVC100 MAC XFI PCIe 2/3 x2/x4 7×11 mm
AQVC107 Both PCIe 2/3
x1/x2/x4		 10GbE
5GbE
2.5GbE		 12×14 mm

For the three new chips, one is a PHY, one is a PCIe network controller, and a third combines the two. The PHY can take a standard camera inputs (2500BASE-X, USXGMII, and XFI) and send the data through multi-gigabit Ethernet as required. The controller can take standard XFI 10 Gb SerDes data and output direct to PCIe, while the combination chip is as a regular MACPHY combo, converting Ethernet data to PCIe. All three chips are built on a 28nm process (Aquantia works with both TSMC and GloFo, but stated that for these products the fab is not being announced), and qualified for the AEC-Q100 industry standard. 

 
Click to enlarge block diagrams

The benefits of using multi-gigabit, as explained to us by Aquantia, is that it allows for a 2.5G connection using only a standard twisted pair cable, or 5G for dual pair, up to 10G for quad pair. Current automotive networking systems are based on single pair 100/1000Mbit technology, which is insufficient for the high bandwidth, low latency requirements that companies like NVIDIA put into their Level 4/5 systems. 

These chips were designed on Aquantia’s roadmap before its collaboration with NVIDIA, however NVIDIA approached Aquantia looking for something to work, given Aquantia’s current march on multi-gigabit Ethernet ahead of its rivals. We are told that the silicon doesn’t do anything special and specific with NVIDIA, allowing other companies keen on automotive technology to use Aquantia as well. With Aquantia’s lead in the multi-gigabit Ethernet space, over say Intel, Qualcomm, and Realtek, it seems that the only option at this point for wired connectivity, if you need to send raw data, is something like this. However, the lead time for collaboration seems to be substantial: Aquantia stated that NVIDIA’s Gary Shapiro recorded promotional material for them in the middle of last year, however Xavier was announced in 2016, so it is likely that Aquantia and NVIDIA were looking at integration before then. 

A quick side discussion on managing all this data. If there is 16 GB/s from all the sensors flying around, the internal switches and SoCs has to be able to handle it. At CES, NVIDIA provided a base block diagram of an Xavier SoC, including some details about its custom ARM cores, its GPU, the DSPs, and some about the networking. 


Image via CNX-Software

The slide shows that the silicon has gigabit and 10 gigabit embedded in (so it just needs a PHY to work), as well as 109 Gbps total networking support. On the Video Processor, it supports 1.8 gigapixel/s decode, which if we plug in some numbers (1080p60 = 124MPixel/s) allows for about a dozen or so cameras at 8bit color, or a combination of 4K cameras and other sensors. The images of the Xavier also show the ISP, capable of 1.5 gigapixel/s.

An mockup example from Aquantia showed a potential Level 4/5 autonomous arrangement, with 10 RADAR/LIDAR/SONAR sensors, 8 cameras, and a total of 18 PHYs, two controllers, and three switches. Bearing in mind that there is a level of redundancy for these systems (cameras and sensors should  connect two at least two switches, if one CPU fails than another can take over, etc), then this is a lot of networking silicon to go into a single car, and a large potential for anyone who can get the multi-gigabit data transfer done right. The question then comes down to power, which is something Aquantia is not revealing at this time, instead preferring to allow NVIDIA to quote a system wide level power.

The image at the top is the setup shown to us by Aquantia at CES, demonstrating a switch using AQcelerate silicon capable of supporting various cables, including the vital 2.5 Gbps over a single pair.

Related Reading

  • Comments are Closed

The EVGA X299 FTW K Motherboard Review: Dual U.2 Ports

The EVGA X299 FTW K is our first X299 motherboard from EVGA. The FTW K aims to bring users a solid power delivery, three-way multi-GPU capabilities, two M.2 slots, and a unique feature so far on our X299 coverage: two U.2 ports. The X29…

Intel Announces Q4 2017 and FY 2017 Results
  • Comments are Closed

Intel Announces Q4 2017 and FY 2017 Results

Today Intel announced their quarterly earnings, and 2017 was another record year for the company. Q4 revenue was a record $17.1 billion, and the full year revenue was a record $62.8 billion. For this quarter, Intel’s GAAP earnings took a hit due to a $5.4 billion tax expense thanks to the new tax reforms that were enacted in December, but as a one-time hit, it shouldn’t be a concern going forward. In fact, Intel is forecasting only a 14% tax rate for FY 2018.

In GAAP terms, Intel’s gross margin was 63.1%, which is up 1.4% from a year ago, and their operating income was up 19% to $5.4 billion. However, thanks to a 111.4% effective tax rate due to the one time $5.4 billion tax fee, Intel is actually reporting a GAAP loss for the quarter of $0.7 billion. Earnings per share were therefore down 120% to a loss per share of $0.15.

Due to the tax impact, it’s probably a good thing to look at Non-GAAP earnings as well which will exclude that one-time charge. In terms of Non-GAAP, Intel’s gross margin was 64.8%, which was up 1.7%. Operating income was up 21% to $5.9 billion. Intel’s Non-GAAP tax rate was 21.2%, which is actually higher than the 19.8% they paid last year, and net income was $5.2 billion, up 34% from a year ago. This led to earnings per share being up 37% to $1.08.

For the full year, Intel had $62.8 billion in revenue, with a gross margin of 62.3%. Operating income was up 39% year-over-year to $17.9 billion, although thanks to the tax hit their net income was down 7% to $9.6 billion. In Non-GAAP terms, net income was up 27% to $16.8 billion.

Intel Q3 2017 Financial Results (GAAP)
  Q4’2017 Q3’2017 Q4’2016
Revenue $17.1B $16.1B $16.4B
Operating Income $5.4B $5.1B $4.5B
Net Income -$0.7B $4.5B $3.6B
Gross Margin 63.1% 62.3% 61.7%
Client Computing Group Revenue $9.0B +1.6% -2.0%
Data Center Group Revenue $5.6B +14.7% +20.0%
Internet of Things Revenue $879M +3.5% +21.0%
Non-Volatile Memory Solutions Group $889M -0.2% +9.0%
Programmable Solutions Group $568M +21.1% +35.0%

Intel is still pivoting business away from the declining PC market, and they have done well to diversify, but still, their PC business is still the biggest piece of the pie. For the quarter, the Client Computing Group had revenues of $9.0 billion, which is down 2% year-over-year, but for the full year of 2017, their CCG was up 3% to $34 billion. It’s not quite dead yet. In an effort to show their diversification, Intel is now quoting their data in “PC-centric” and “Data-centric” and conveniently, Data-centric is all of their business outside of the CCG, but even so, the CCG still accounted for 53% of Intel’s revenue for the quarter.

But, the Data-centric group is certainly growing much quicker than the PC market which is still in a decline. The Data Center Group is by far the largest portion of Intel’s Data-centric efforts, and the DCG had revenues of $5.6 billion for the quarter, which are up 20% year-over-year. With the expansion of the cloud, this likely still has some ways to go before it hits a peak, so it’s likely just a matter of time before they finally surpass their Client Computing Group.

The rest of the Data-centric business was also up for the quarter, with IoT up 20% year-over-year, to $879 million. For the full year the IoT was up 20% and had revenues of $3.2 billion. Non-volatile memory was up 9% for the quarter to $889 million, and for the year it was up 37% to $3.5 billion. Programmable Solutions was up 35% for the quarter to $568 million, and for the full year, was up 14% to $1.9 billion.

Intel still hasn’t shipped anything on 10nm to the point where someone could go buy a chip, although we’re finally getting close, with Intel shipping 10nm to some of their partners. 14nm was delayed, but the delay in getting to 10nm has been longer than likely anyone expected. Intel has still made some bold claims about density, so it should be a good node for them if and when it arrives.

Intel’s forecast for Q1 2018 is for $15.0 billion in revenue, plus or minus $500 million, and despite the rough start to 2018, they are still forecasting another record year, with $65 billion in revenue, plus or minus $1.0 billion.

Source: Intel Investor Relations