SIMA.ai Unveils MLSoC Modalix: Bringing Generative AI to the Edge

SiMa.ai Expands ONE Platform for Edge AI with MLSoC™ Modalix, a New Product Family for Generative AI

Industry's first multi-modal, software-centric edge AI platform supports any edge AI model from CNNs to multimodal GenAI and everything in between with scalable performance per watt SiMa.ai, the software-centric, embedded edge machine learning system-on-chip (MLSoC) company, today announced MLSoC™ Modalix, the industry's first multi-modal edge AI product family. SiMa.ai MLSoC Modalix supports CNNs, Transformers, LLMs, LMMs and Gen AI at the edge, and delivers industry leading performance - more than 10X the performance per watt of alternatives. The rise of generative AI is changing the way humans and machines work together. The next wave of the AI technology revolution will advance multi-modal machines with the ability to understand and process multiple forms of inputs across text, image, audio and visual. This shift will ripple across every industry, from agriculture and logistics, to medicine, defense, transportation and more. SiMa.ai is the breakthrough multi-modal edge AI platform that organizations can now rely on for the ultra-responsive, power-efficient, reliable and secure insights their critical innovations demand - in any modality they prefer. MLSoC Modalix will enable developers to push the envelope on performance with a small power envelope and physical footprint across a wide breadth of applications. They will be able to run end to end GenAI application-pipelines on just a single chip while delivering measurable contributions to business top and bottom lines. SiMa.ai MLSoC Modalix is the second generation of the successful, commercially deployed first generation MLSoC. MLSoC Modalix is offered in 25 (Modalix 25 or "M25"), 50 (Modalix 50 or "M50"), 100 (Modalix 100 or "M100") and 200 (Modalix 200 or "M200") TOPS configurations, in multiple form factors, and is purpose-built to provide effortless deployment of Generative AI for the embedded edge ML market. Fully software compatible with first generation MLSoC, the MLSoC Modalix product family was designed to enable the capability to run DNNs, as well as advanced Transformer models, including LLMs, LMMs and Generative AI. Samples of MLSoC Modalix will be available to customers in Q4 of 2024. "SiMa.ai is leading the edge AI market and building increasing momentum as customers around the world continue to adopt our (first generation) MLSoC. We are excited to introduce the MLSoC Modalix family to give customers even greater choice and flexibility to embrace multi modal AI," said Krishna Rangasayee, Founder and CEO at SiMa.ai. "With the introduction of MLSoC Modalix, SiMa.ai's ONE Platform for Edge AI now provides coverage from CNNs to Gen AI and everything in between with industry leading performance and power efficiency." SiMa.ai Broadens ONE Platform for Edge AI with New MLSoC Modalix Family SiMa.ai ONE Platform for Edge AI is a software-centric framework offering compatibility across the entire MLSoC platform family to provide a seamless experience for upgrades, transitions and mix-and-match, minimizing TCO for adopters of SiMa.ai at the edge. First generation SiMa.ai MLSoC is a market leader in performance and power efficiency, validated most recently by MLCommons® MLPerf Inference benchmark (4.0) results in March 2024. First generation SiMa.ai MLSoC will continue to support CNN-based models across Industrial Inspection, Retail, Aerospace and Defense, and Smart Vision (Ethernet-based cameras only). Customers leveraging MLSoC Modalix will be able to accelerate deep learning, transformer, and Generative AI models - pushing the envelope on models like Llama2 7B to beyond 10 tokens per second. With an on-chip application processor, integrated ISP, vision and digital signal processors, and innovations like BF16 in the hardware, MLSoC Modalix will accelerate all known CNN and transformer models: language, vision and speech, delivering inference-accuracy close to fp32 at significantly lower cost and power. Generative AI applications running on MLSoC Modalix will: ● Benefit from the SiMa.ai Palette software patented Compiler with layered direct proactive data prefetch, ensuring each data layer is always available, on time, when it is needed; ● Leverage the improved Machine Learning Accelerator to run DNNs and Large Multi-Modal Models at high performance without sacrificing power. In addition new hardware components of MLSoC Modalix include: ● An integrated ISP module supporting 4 x 4 lanes of MIPI CSI 2, allowing the chip to interface with MIPI cameras as well as Ethernet-based cameras via 4 x 10G Ethernet ports on-chip; ● Eight lanes of PCIe Gen 5 facilitating quicker ingress and egress of data to and from the chip; ● 8 Arm® Cortex®-A65 dual-threaded CPUs, further extending SiMa.ai's reach to customers with specialized use cases through Arm's unrivaled and diverse software partner ecosystem; ● TSMC's N6 technology. Industry Validation for SiMa.ai "The launch of the SiMa.ai ONE Platform with the MLSoC Modalix family has the potential to be a great enabler for Elementary's applications," said Arye Barnehama, CEO of Elementary. "Elementary is at the forefront of deploying AI-based vision inspection systems for global manufacturers dedicated to leading-edge innovation. The Modalix family's high performance and energy efficiency, coupled with the ONE Platform approach, aligns with our commitment to delivering modular, uncompromising solutions to both our partners and end users. As Elementary continues to scale in the industrial sector, the SiMa.ai MLSoC Modalix product family has the potential to power solutions that are integral to the high-speed production processes of global enterprises." "The SiMa.ai MLSoC Modalix family has the ability to accelerate a wide variety of architectures ranging from CNNs to LMMs and in turn making multimodal edge AI on power-constrained platforms a reality. Hayden AI is excited at the prospect of evaluating the SiMa.ai Modalix family for potential inclusion in our future products," said Vaibhav Ghadiok, CTO, Hayden AI. "The availability of SiMa.ai MLSoC Modalix product family is a game changer for us and our edge AI customers. Advantech has been building and deploying industrial PCs, smart cameras, and edge gateways for healthcare, industrial automation, and smart cities for many years," said Emily Teng, Associate Director of Product Management, Advantech. "Rich peripherals including MIPI and advanced ISP combined with the high-performance and low-power envelope of the SiMa.ai MLSoC Modalix family will enable a wide range of smart devices at the edge. SiMa.ai will allow us to adapt and scale to embrace new modalities and innovation inherent in GenAI." "The availability of SiMa.ai ONE Platform with the MLSoC Modalix family is a potential game changer for Edge-AI applications that LIPS Corporation has been building and deploying to scale multi-camera 3D sensing solutions in smart factory and smart vision projects," said Luke Liu, CEO of LIPS Corporation. "The high-performance and low-power envelope of the MLSoC Modalix family, combined with their ONE Platform approach, can enable us to overcome the performance and bandwidth limitations often associated with multi-camera 3D sensing projects. SiMa.ai MLSoC Modalix can enable us to adapt and scale new AI modalities and innovations in real-time pose detection, precise spatial mapping, and advanced AI detection across various industrial and automation settings, ultimately accelerating our ability to deliver smarter, more efficient, and safer solutions for our clients." "In the age of AI, hardware and software capabilities need to advance at a faster pace than ever before," said Paul Williamson, SVP and GM, IoT Line of Business, Arm. "By building solutions like MLSoC Modalix on the high performance, efficient Arm platform, partners like SiMa.ai can access a powerful software ecosystem to create the next generation of secure and innovative user experiences across applications ranging from retail to industrial." "TSMC has long-partnered with industry innovators like SiMa.ai, helping them achieve groundbreaking next-generation semiconductor designs with our industry-leading technologies and manufacturing excellence," said Lucas Tsai, Senior Director of Market Development and Emerging Business Management, TSMC North America. "We are excited to continue our collaboration in accelerating SiMa.ai's powerful and energy-efficient chip innovation to meet the rapidly growing demand for GenAI at the edge." "Synopsys continues to help companies at all levels accelerate and scale their advanced AI chip development process," said Ravi Subramanian, GM of the Systems Design Group at Synopsys. "By leveraging Synopsys' leading AI-driven EDA suite, broad interface, security, and processor IP portfolio, architecture design solution and emulation systems, SiMa.ai is speeding development cycles and unlocking step-function gains in performance and power for their next-generation MLSoC Modalix family. Together, we are pushing the boundaries of what is possible with GenAI applications, empowering engineers to be more productive and proficient, and shaping the future of AI at the edge." About SiMa.ai SiMa.ai is the software-centric, embedded edge machine learning system-on-chip (MLSoC) company. SiMa.ai's hardware to software stack flexibly adjusts to any framework, network, model, sensor, or modality all in one platform. Edge ML applications that run completely on the SiMa.ai MLSoC see a tenfold increase in performance and energy efficiency, bringing higher fidelity intelligence to ML use cases spanning computer vision to generative AI, in minutes. With SiMa.ai, customers unlock new paths to revenue and significant cost savings to innovate at the edge across industrial manufacturing, retail, aerospace, defense, agriculture, and healthcare. SiMa.ai was founded in 2018, has raised $270M and is backed by Fidelity Management & Research Company, Maverick Capital, Point72, MSD Partners, VentureTech Alliance and more.

SiMa.ai expands ONE platform for edge AI with new product family for GenAI

SiMa.ai, a software-centric, embedded edge machine learning system-on-chip (MLSoC) company, has announced MLSoC Modalix, which it claims is the industry's first multi-modal edge AI product family. The MLSoC Modalix supports CNNs, Transformers, LLMs, LMMs, and Gen AI at the edge. Krishna Rangasayee, Founder and CEO at SiMa.ai, said, "Given the amount of change the AI industry goes through, being software-centric is key. Over the last year, we have doubled the number of customers, so we're growing quite well. Multimodality has become big and we will see that shift in the edge." "Large language models and generative AI have also really taken off and they're happening not only at the cloud but also at the edge in certain applications. We have introduced an extension to our ONE platform for edge AI with a product line called MLSoC Modalix." Also read: Oracle announces AWS partnership and multi-cloud alliance with Google Cloud MLSoC Modalix is supposed to enable developers to push the envelope on performance with a small power envelope and physical footprint across applications. They will be able to run end-to-end GenAI application pipelines on just a single chip while delivering measurable contributions to the business's top and bottom lines. Rangasayee said that while the company's primary competition ends up being Nvidia. For the edge market, both performance and power matter. "A key metric our customers care about is performance per watt, as in how much to compute for a given power budget. We are consistently 10x of Nvidia. Doing something 10x of the market leader is not easy and as a startup, we are doing this quite well globally." He added that the company has customers in India and the rest of the world. SiMa.ai MLSoC Modalix is the second generation of the commercially deployed first generation MLSoC. MLSoC Modalix is offered in 25 (Modalix 25 or "M25"), 50 (Modalix 50 or "M50"), 100 (Modalix 100 or "M100") and 200 (Modalix 200 or "M200") TOPS configurations, in multiple form factors, and is purpose-built to provide deployment of GenAI for the embedded edge ML market. Fully software compatible with first-generation MLSoC, the MLSoC Modalix product family was designed to enable the capability to run DNNs, as well as advanced transformer models, including LLMs, LMMs and Generative AI. Samples of MLSoC Modalix will be available to customers in Q4 of 2024. The founder also added that SiMa.AI is scaling globally and doubling its customer footprint. The company also established its R&D center in Bangalore in 2020 and currently has around 70 people. "We are engaged with some significant large market leaders and smaller companies in India. We are in multiple markets like robotics, automotive, aerospace defense, smart vision systems, medical and industrial automation. Given our footprint in India, we're excited about supporting the customers locally from our India center," he said. SiMa.ai was founded in 2018, has raised $270M, and is backed by Fidelity Management & Research Company, Maverick Capital, Point72, MSD Partners, VentureTech Alliance, and more. Also read: Xiaomi India appoints Sudhin Mathur as COOSHARE Copy linkEmailFacebookTwitterTelegramLinkedInWhatsAppRedditPublished on September 11, 2024

Edge chip maker SiMa.ai launches Modalix to bring multimodal gen AI everywhere

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Edge computer chip and software startup SiMa.ai, fresh off a $70 million funding round from industry heavyweights including Dell Technologies Capital, is expanding its foothold in the edge AI market with the release of a new, smaller, lower power chip: MLSoC Modalix. At 6 nanometers, it comes in way smaller than the San Jose, California-based startup's prior MLSoC chip of 16 nanometers. Building on the company's ONE Platform for Edge AI, this new offering is designed to support advanced AI models such as Convolutional Neural Networks (CNNs), Transformers, and Generative AI, all while delivering industry-leading energy efficiency and scalable performance. In a video call interview, SiMa.ai CEO Krishna Rangasayee shared the excitement surrounding the launch. "We're extending the momentum we have as being the one platform for AI and introducing a capability not only in silicon but also the software that goes along with it," he told VentureBeat. Where will the Modalix family end up? Rangasayee says it's perfect for "industrial automation, healthcare, smart vision systems, aerospace and defense, and anywhere there's a need for multimodal elements." As Rangasayee pointed out, "Robotics, embodied AI, and sensory information are the future. Modalix is perfect for that. It's about generative AI-centric architecture driving new applications, like human-robot interactions." But, the ultimate vision is even more ambitious. Rangasayee says the chip could help usher in an age where "every appliance, every device, is going to be capable of human-like capacity. So it'll be able to talk, express, and visualize." Pushing the boundaries of edge AI Generative AI is rapidly transforming industries, but has largely been confined so far to desktop PCs and mobile devices. Now, thanks to Sima.AI and its competition -- namely GPU leader Nvidia, which also offers edge chips in its Orin and Xavier families -- the technology is advancing to allow for powerful AI models to be deployed in dedicated, specialized devices out in the field and the factory floor, such as robotic arms and drones. "We are consistently in real-life applications 10x better than an immediate competitor, and now this further extends where it will be more than 10x of anybody else," Rangasayee claimed. SiMa.ai's MLSoC Modalix platform is designed to handle multimodal AI processing, integrating inputs such as text, images, and audio. It can run variants of Meta's Llama 2-7B parameter model right on it, a huge potential unlock for reasoning at the edge. As Rangasayee noted, "People are combining reality. So you could get audio with video with text, and the input could be any of these, and the output could be a combination. That's the second big shift we're addressing." The MLSoC Modalix family introduces several configurations ranging from 25 to 200 TOPS, each engineered to handle demanding AI workloads while minimizing power consumption. According to Rangasayee, this new platform represents a leap in capability: "Modalix bridges the evolutions that have happened in the last two years. Now you can run everything from CNNs to the latest cutting-edge models on a single chip." SiMa.ai's technology, designed specifically for edge applications, addresses key challenges in the field, including performance-per-watt. "One key technical merit is frames per second per watt, or inferences per second per watt. In real-life applications, we're 10x better than our immediate competitors, and Modalix extends that lead," Rangasayee explained. The goal, he added, is for customers to no longer worry about power and cooling constraints. "With Modalix, it's a checkbox: low power, high performance - reshaping what's possible at the edge." Endorsed by industry players The potential of SiMa.ai's MLSoC Modalix family has not gone unnoticed by industry leaders. Arye Barnehama, CEO of Elementary, expressed enthusiasm for the platform's energy efficiency and high performance, which aligns with Elementary's vision inspection systems. Similarly, Vaibhav Ghadiok, CTO of Hayden AI, highlighted the platform's ability to enable multimodal AI on power-constrained edge devices. The MLSoC Modalix family also benefits from SiMa.ai's Palette Edgematic software stack, a no-code, drag-and-drop platform designed to make AI deployment accessible to non-specialist developers. Incorporating innovations such as an integrated Image Signal Processor (ISP), PCIe Gen 5 support, and eight Arm Cortex-A65 CPUs, the platform is built to handle a range of AI workloads. SiMa.ai's approach allows for seamless integration of AI into existing workflows. A growing edge AI market SiMa.ai's latest product launch also signals its ambition to compete with industry giants like Nvidia. While Nvidia dominates in cloud-based AI applications, SiMa.ai is focusing on a niche where real-time, on-device processing is critical. Last year, Rangasayee emphasized that SiMa.ai's chips outperformed Nvidia's in terms of both performance and power efficiency for edge AI applications. As edge AI continues to grow, analysts predict the global edge computing market will double in the coming years, driven by advances in AI and increased demand for real-time decision-making at the edge. SiMa.ai's MLSoC Modalix family is well-positioned to meet this demand, offering a platform capable of processing multimodal AI models on a single chip. With the launch of the MLSoC Modalix family, SiMa.ai is extending its leadership in edge AI. The platform's high performance, energy efficiency, and ease of deployment make it a compelling option for industries seeking to harness the power of AI at the edge. With strong backing from investors like Dell Technologies Capital and a growing list of industry partners, SiMa.ai is poised to lead the next wave of AI innovation at the edge.

SiMa.ai Announces MLSoC Modalix to Run GenAI on the Edge

Samples of MLSoC Modalix will be available to customers in Q4 of 2024 SiMa.ai, the software-centric, embedded edge machine learning system-on-chip (MLSoC) company, today announced MLSoC™ Modalix, the industry's first multi-modal edge AI product family. SiMa.ai MLSoC Modalix supports CNNs, Transformers, LLMs, LMMs and Gen AI at the edge, and delivers industry leading performance - more than 10X the performance per watt of alternatives. MLSoC Modalix will enable developers to push the envelope on performance with a small power envelope and physical footprint across a wide breadth of applications. They will be able to run end to end GenAI application-pipelines on just a single chip while delivering measurable contributions to business top and bottom lines. SiMa.ai MLSoC Modalix is the second generation of the successful, commercially deployed first generation MLSoC. MLSoC Modalix is offered in 25 (Modalix 25 or "M25"), 50 (Modalix 50 or "M50"), 100 (Modalix 100 or "M100") and 200 (Modalix 200 or "M200") TOPS configurations, in multiple form factors, and is purpose-built to provide effortless deployment of Generative AI for the embedded edge ML market. The MLSoC Modalix product family is fully software compatible with the first-generation MLSoC and was designed to enable the capability to run DNNs and advanced Transformer models, including LLMs, LMMs, and Generative AI. Samples of MLSoC Modalix will be available to customers in Q4 of 2024 "With the introduction of MLSoC Modalix, SiMa.ai's ONE Platform for Edge AI now provides coverage from CNNs to Gen AI and everything in between with industry-leading performance and power efficiency," Krishna Rangasayee, Founder and CEO at SiMa.ai, said.

SiMa.ai debuts new system-on-chip for multimodal AI at the network edge - SiliconANGLE

SiMa.ai debuts new system-on-chip for multimodal AI at the network edge Artificial intelligence chip startup SiMa Technologies Inc. has unveiled what it says is the first system-on-chip platform that's specifically designed to handle multimodal AI workloads at the embedded edge. With today's launch, the startup reckons it can power a wide range of AI workloads in industrial robots and other edge computing devices with much lower latency than was previously possible. The new chip, called the MLSoC Modalix, supports various kinds of AI models, including large language models such as OpenAI's GPT family, convolutional neural networks, transformer models and more. According to the startup, AI has already had a huge impact on the way humans work together with machines, powering edge applications in warehouses, factory floors, weather stations, autonomous cars, drones and other areas. However, most of these edge applications run much smaller models than the ones it's targeting with its latest chip. The company is specifically aiming at so-called multimodal AI applications, which can understand and process multiple inputs in the shape of text, images, audio and more. Such models typically require much more processing power. Until now, they've always relied on cloud-based processors, resulting in lower latency when used in edge applications. That will change with the MLSoC Modalix chip, as it packs much more processing power than the company's original MLSoC platform. According to the startup, the MLSoC Modalix will be made available in a number of configurations, including 25, 50, 100 and 200 TOPS - a metric for performance, where one TOP corresponds to 1 trillion computations per second. The chips are integrated with a 25-megabyte memory pool, which allows them to store data directly on the chip without using external random access memory. SiMa.ai said the new chip is built on Taiwan Semiconductor Manufacturing Co.'s six-nanometer process and packages four vector processing units for running computer vision algorithms, as well as modules for video encoding and computer vision tasks. There's also an integrated ISP module that allows the chip to interface with Mobile Industry Processor Interface-based cameras and Ethernet-connected cameras, as well as four lanes of PCIe Gen 5 for data ingress and egress to and from the chip. Meanwhile, general-purpose computing tasks are offloaded to an Arm-designed eight-core Cortex A65 central processing unit that runs Yocto Linux, an operating system specifically designed to run connected edge devices. SiMa.ai says MLSoC Modalix will enable developers to "push the envelope" in terms of edge AI performance, running end-to-end generative AI application pipelines on just a single chip. The result will be far more powerful edge AI applications, including drones that can operate autonomously to perform tasks in agriculture, industrial maintenance and building inspections, and autonomous vehicles that can drive themselves while performing routine maintenance checks on themselves. Other use cases include things such as medical imaging and robot-assisted surgery in healthcare, and smart retail applications, the company said. Besides the new chip, developers also get access to SiMa.ai's suite of low-code AI development tools, known as Pallet, which includes a compiler that enables AI models to run more efficiently on the hardware, plus a layered, direct proactive data prefetch capability that ensures the underlying models have access to data in real time. SiMa.ai founder and Chief Executive Krishna Rangasayee said his company has already seen significant uptake of its first-generation MLSoC chip. "With the introduction of MLSoC Modalix, SiMa.ai now provides coverage from convolutional neural networks to generative AI and everything in between, with industry-leading performance and power efficiency," he said.