Viewing entries in
Blog

Comment

NGCodec Joins the Alliance for Open Media

SUNNYVALE, Calif., – July 12, 2017 – NGCodec® Inc., a startup company focused on FPGA-based video codec development, today announced it has joined the Alliance for Open Media (AOMedia), a non-profit effort to develop a new open and royalty-free video codec. As a member of the Alliance, NGCodec will collaborate with industry leaders in pursuit of an open and royalty-free AOMedia video codec.

“Every day, YouTube viewers alone watch over a billion hours of online video,” said NGCodec CEO and Co-founder Oliver Gunasekara. “As an AOMedia member, NGCodec joins in the Alliance’s pursuit of an open and royalty-free next generation video codec. Combined with NGCodec’s 10x lower cost cloud encoding, AOMedia’s efforts will enable more customers to access high-quality, low-cost video that is optimized for Cloud delivery.”

The availability of AOMedia Video as an open source project is an important element in fulfilling the organization’s promise to deliver an open and royalty-free next generation video codec that is scalable to any modern device at any bandwidth.

"We’re pleased to welcome NGCodec to AOMedia, reflecting the growing importance of Cloud-based video encoding to achieve broad industry adoption," said AOMedia’s Executive Director Gabe Frost. "We look forward to NGCodec's contributions to this significant initiative, which will open the door to a new era of openness and interoperability for online video."

AOMedia’s Video 1 codec (AV1) is designed with a low computational footprint, optimized for hardware, and is capable of consistent, highest quality, real-time video delivery, making it flexible for both commercial and non-commercial content, including user-generated content.

 

About the Alliance for Open Media

The Alliance for Open Media is a project of the Joint Development Foundation, an independent non-profit organization that provides the corporate and legal infrastructure to enable groups to establish and operate standards and source code development collaborations. Launched in 2015, the Alliance for Open Media was formed to define and develop media technologies to address marketplace demand for an open standard for video compression and delivery over the web. For more information, visit http://aomedia.org/.

 

About NGCodec Inc.

NGCodec® has been in passionate pursuit of next generation video compression since 2012. With the support of investors including Xilinx, NGCodec’s agile startup team has created Reality Codec™, a compressor-decompressor technology optimized for ultra-low latency, high-quality applications. Headquartered in Sunnyvale, California, NGCodec leverages FPGA acceleration in the Cloud to lower encoding costs by 10x over traditional CPU encoders. Learn more at https://ngcodec.com or @NGCodec.

Comment

Comment

Very excited that NGCodec is one of the Red Herring Top 100 North America 2017 Finalists

100_N_America_Finalist_outline

Red Herring has announced the finalists for its Top 100 North America 2017 event, to be held later this month in California. The list, which has been whittled down from thousands of entrants, comprises many of the continent’s brightest and most innovative tech companies.

The finalists will find out if they have been crowned a Top 100 winner at the Red Herring Top 100 North America 2017 conference, held at the Marina Del Rey Marriott Hotel, June 12-14.

There, finalists will present their company’s business models to a host of industry judges. The event also comprises keynote speeches and roundtable discussions involving some of the local ecosystem’s most revered individuals.

The conference will conclude June 14 with a special dinner and awards ceremony led by Red Herring chairman Alex Vieux, who has singled out this year’s crop of entrepreneurial talent as one of the best he has seen in over two decades with the company.

“I’m absolutely thrilled to see all these fantastic companies and entrepreneurs under one roof,” he said. “California is tech’s powerhouse. And these guys are showing that there is plenty more to come from all over North America.”

Top 100 winners are judged on a wide variety of criteria, and go through a rigorous evaluation process. To see a full list of this year’s finalists, click here.

Comment

Comment

Live Video Encoding Using New AWS F1 Acceleration

In today’s mobile world, where live video is rapidly gaining ubiquity in everyday life, NGCodec is leading the charge to overcome the difficulties and sacrifices in quality associated with video encoding using traditional software methods. This white paper discusses the benefits of hardware encoding using Xilinx® FPGA in the new Amazon Web Services (AWS) F1 instances. We open with a background on video encoding and an overview of the encoding process. This is further contextualized with a discussion of the applications of cloud video transcoding and an exploration of the differences between file-based and live video encoding. Following on from this, we explore the limitations of traditional software encoding methods for live video encoding. Having established the drawbacks of relying on CPUs and GPUs, we discuss the superior results that can be obtained through hardware encoding with AWS FPGA F1 instances. Our paper goes on to delve into the methodology behind NGCodec’s FPGA F1 design using the Xilinx Vivado® HLS tool suite and to summarize how we ported our RealityCodecTM H.265/HEVC video encoder to AWS Elastic Compute Cloud (EC2) F1 instances in only three weeks. Finally, the paper outlines our roadmap for a new, twofold business model to make hardware encoding with FPGA F1 instances available to customers of all sizes and closes with an opportunity for readers to try out NGCodec’s video encoding capabilities for themselves.

Comment

NGCodec has joined as a founder of the VR Industry Forum

Comment

NGCodec has joined as a founder of the VR Industry Forum

The Virtual Reality Industry Forum is composed of a broad range of participants from sectors including, but not limited to, the movie, television, broadcast, mobile, and interactive gaming ecosystems, comprising content creators, content distributors, consumer electronics manufacturers, professional equipment manufacturers and technology companies. Membership in the VR Industry Forum is open to all parties that support the purposes of the VR Industry Forum. The VR Industry Forum is not a standards development organization, but will rely on, and liaise with, standards development organizations for the development of standards in support of VR services and devices. Adoption of any of the work products of the VR Industry Forum is voluntary; none of the work products of the VR Industry Forum shall be binding on Members or third parties.

The principal purpose of the VR Industry Forum shall be:

To further the widespread availability of high quality audiovisual VR experiences, for the benefit of consumers.

The VR Industry Forum focuses on content that is transmitted as audio and video, and it will monitor complementary technologies for inclusion in its scope, including those that enable augmented reality and mixed reality. The VR Industry Forum targets immersive experiences that typically require head-mounted devices, understanding that immersive VR content may also be consumed on “2D flat screens” (like tablets, mobile phones, PC screens, TVs) with navigation capabilities.

Comment

NGCodec Wireless VR demo at CES

Comment

NGCodec Wireless VR demo at CES

Celeno Communications hosted us in their demo suite in the 35th floor of the Venetian tower. We announced a partnership with them last month to demonstrate sub frame Wireless VR over 802.11ac 5Ghz WiFi. 

We scheduled around 40 meetings with most of the major VR/AR players and had an additional 10 meetings of Celeno referrals. It was a crazy busy 3 days. In total about 100 people tried our demo.

The NGCodec / Celeno Wireless VR demo at CES

The NGCodec / Celeno Wireless VR demo at CES

Diagram of the CES demo. Click to download the PDF CES flyer

Diagram of the CES demo. Click to download the PDF CES flyer

Our demo consists of an 'A' 'B' test. In one configuration the PC was directly connected to the HMD via HDMI. In a different configuration the PC was connected via our RealityCodec™ in a FPGA and WiFi show above. The Output of the PC was compressed to 15Mbps using H.265 and sent over Celeno 802.11ac WiFi to a video decoder and then sent to the HMD.

The goal was for people to see if they could detect which of the settings 'A' or 'B' uses our RealityCodec and Celeno WiFi. In setting 'A' the HDMI was using 3GBps and in settings 'B' it was using 15Mbps over WiFi. This is a 250:1 reductions. 

My Laptop (2016 MacBook Pro) Speedtest in Las Vegas Hotel.

My Laptop (2016 MacBook Pro) Speedtest in Las Vegas Hotel.

We asked people to look for compression artifacts, increased latency and wireless packet drops. 80% of people could not tell any difference or selected the wrong choice as Wifi. 10% detected packet loss. We are working on a solution to improve this with Celeno. 10% could detect a small increase in latency. Again we will soon halve our latency on our next demo.

We also talked about Cloud VR/AR/MR. Many service providers see this as a great opportunity for Fibre+WiFi to the home and in the future 5G. More demos on this to come. BTW in my hotel in las Vegas (Luck Dragon) I had a 3ms ping to the local internet server in Overton, NV. Clearly this hotel has a fibre connections and could deliver Cloud VR today.

P.S. I especially want to thank Neel our senior software enginner who prepared the demo and drove and set it up in Las Vegas.

P.P.S Our next demo & meetings will be at MWC and GDC.

Comment

Comment

Amazon Webinar “Announcing Amazon EC2 F1 Instances with Custom FPGAs”

Below are the video and slides for the recent Amazon webinar. It's great that they highlighted our F1 demo, which we conducted at AWS re:Invent, on slide number #20 (29 minutes, 52 seconds into the video).

Amazon EC2 F1 is a new compute instance with programmable hardware for application acceleration. With F1, you can directly access custom FPGA hardware on the instance in a few clicks.

Comment

Comment

NGCodec Video from AWS on the worlds first F1 demo is now live

NGCodec 15m presentation starts at 28 minutes and 50 seconds into the video.

Accelerated computing is on the rise because of massively parallel, compute-intensive workloads such as deep learning, 3D content rendering, financial computing, and engineering simulations. In this session, we provide an overview of our accelerated computing instances, including how to choose instances based on your application needs, best practices and tips to optimize performance, and specific examples of accelerated computing in real-world applications.

Comment

2 Comments

Delivering a Moonshot - Being the first company to use the new Amazon F1 Instance

Last Thursday 12/1 NGCodec presented at the AWS:ReInvent 2016 conference in Las Vegas and gave a live demo using the new AWS F1 Instance type.  We also provided a quote for an Amazon press release and issued our own press release. The session was sold out and had around 350 seats, although some people did leave before I presented.

Four weeks ago Amazon told us confidentially they were going to launch an FPGA Instance and asked us if we would like to create a demo. The challenge was we had little time and we would be the first company to try to use the F1 Instance. We were told we could not tell our investors, including Xilinx, only employees who 'need to know'. 

We came up with this 'Moonshot' demo idea:

Live in Las Vegas we would stream HD video in H.264 from an iPhone to the AWS EC2 F1 Instance in the 'us-east-1' data center. In the AWS data center we would decode the H.264 video and then send it to our RealityCodec running on the Xilinx FPGA where it would be re-encoded in H.265. We would then send the H.265 video back to a laptop on stage in Las Vegas. The laptop on stage would then use a software decoder to display in real time. The goal is to show how we could live transcode H.264 to H.265 and halve the bit rate at the same quality.

To deliver this demo was an amazing achievement by 3 different teams working together:

  • The AWS F1 Infrastructure team primary located in Austin. They provided the Hardware Development Kit (HDK) and supported the NGCodec team.
  • The NGCodec team ported our H.265/HEVC RealityCodec to the AWS F1 Instance and created a plug in for FFMPEG.
  • The Cogniance professional services team who we hired to build the supporting cloud software infrastructure for the live transcoding demo. Initially they prototyped with x265 software encoding.

Below you can see a picture from the live demo. When the video from the session is posted I will add the session video below.

NGCodec 15m presentation starts at 28 minutes and 50 seconds into the video.

Just visible is my iPhone 7+ on the podium in front of the speaker. It is using the front facing camera to capture the speaker in HD. We used the BroadcastMe App to stream the HD H.264 footage to AWS. On the screen on the right you can see the output from the laptop. The video is using H.265 and if you look very closely you can see the NGCodec water mark below my head which is inserted by the FPGA RealityCodec.

A few hours before the scheduled demo time of 11.30am the demo was still not working. Our collective teams had been working night and day, Including Thanksgiving weekend to try to get it to work. Luckily in the final hours it started to work. Initially I got a single frame and then video a few hours later. It was an amazing achievement given the short amount of time and the early access to the F1 Instance type that we succeeded just in time.

Below you can see a picture of part of the NGCodec team working in our Lab, which was taken on the Thanksgiving weekend. If you look closely you can see an F1 Instance hardware board. Amazon Austin team was also working as we had many support calls with them over the weekend.

I would like to conclude by thanking the three teams for their incredible amount of passion and effort to ultimately deliver the worlds first demo on an Amazon F1 Instance.

P.S. If you want to learn more about the technical challenges then please leave a comment and we will try to write a new technical blog post.

2 Comments

Comment

Video of AWS CEO announcing the new F1 FPGA Instance

Amazon EC2 F1 is a compute instance with field programmable gate arrays (FPGAs) that you can program to create custom hardware accelerations for your application. F1 instances are easy to program and come with everything you need to develop, simulate, debug, and compile your hardware acceleration code, including an FPGA Developer AMI and Hardware Developer Kit (HDK).

Comment

Comment

NGCodec Quote in Amazon Press Release on the new F1 instance announcement

“Cloud video encoding is exploding, but traditional software approaches need massive CPU resources and cannot deliver the video quality and latency necessitated by new applications. We have developed a next generation video encoder that, using FPGA hardware acceleration, offers low latency and low cost while maximizing video quality to keep up with today’s broadcasting standards,” said Oliver Gunasekara, Chief Executive Officer and co-founder, NGCodec. “AWS is the first cloud provider to add FPGA instances, which provide massive acceleration. In just three weeks, we ported our RealityCodec™ H.265/HEVC encoder to the new F1 instance type to enable AWS Cloud customers to leverage the benefits of higher video quality, lower latency, and lower cost for live H.265/HEVC video encoding. Updating our install base globally will be as simple as publishing a new Amazon FPGA image and re-launching F1 instances.”

The full Amazon Press Release is here: AWS Announces Seven New Compute Services and Capabilities to Support an Even Wider Range of Workloads

 

Comment

Comment

NGCodec engineers use Elecard StreamEye

The ultimate goal of a Video codec is to produce video bitstreams, of high visual quality, at low bit-rates. The current HEVC standard provides many tools toward that goal. But, which tools are implemented and when they are used is part of the art of codec development. There are closed-loop trade-offs of performance, complexity, computational cycles, memory access, etc. There are open-loop decisions of picture types, rate-control, bit distribution, etc. And, as a developer, there are the inevitable sessions of debugging.

As a video codec architect and developer, the Elecard StreamEye tool is indispensable. It provides a clear visualization of the bitstream in multiple perspectives. I can examine the syntax elements to be sure the encoding context is being established properly. I can view the predicted and residual representations of a picture to see if the basic mode decisions can be refined. Then, I can view the QP and bit "heat-map" to see if the bit allocation is reasonable for preserving the video quality. Additional views allow examination of stream and picture statistics, detailed block coding, prediction and transform elements, etc. 

Figure 1 Bit distribution heat map.   http://www.Elecard.com

Figure 1 Bit distribution heat map. http://www.Elecard.com

Figure 2 Qp distribution and picture stats.   http://www.Elecard.com

Figure 2 Qp distribution and picture stats. http://www.Elecard.com

As a team, we have a mix of Windows and OSX machines.  StreamEye supplies versions that run natively on either, so our team is covered.

Elecard tools have been widely deployed for years. StreamEye has been widely used and valued for AVC development. The recent expansion to support HEVC is well done. In short, it would be inconceivable to develop a capable HEVC codec without a full featured HEVC analysis tool, and Elecard StreamEye fits the role. You can learn more about Elecard StreamEye analyzer including downloading a free 30-day trial from here: http://www.Elecard.com

Comment