Moore Threads released software and hardware technology solutions with MUSA as the core, and meta-computing drives infinite innovation.

Beijing, November 3, 2022 – The Moore Threads 2022 Autumn Press Conference was held today in Beijing’s Zhongguancun National Independent Innovation Demonstration Zone. Moore Threads announced a new multi-functional GPU chip called “Chunxiao,” the industry’s first national trend graphics card MTT S80 based on the MUSA architecture, the MTT S3000 for server applications, and the meta-computing all-in-one machine MCCX at the press conference. After a 7-month hiatus, Moore Thread’s multi-functional GPU products have taken another step forward in iterative innovation.

Moore Threads has released a series of GPU software stacks and application tools based on MUSA, including MUSA developer kits, cloud native sGPU technology, and Metaverse platform MTVERSE, among others, with the goal of creating an overall solution from the underlying chip to upper-level development and application. Moore Thread’s multi-functional GPU software and hardware integration innovation model is completely upgraded with this solution.

AOC, Dell, Philips, Gujia, Tsinghua Tongfang, Inspur, Great Wall, Haier, and other partners provided equipment support for the application display, fully demonstrating the powerful product ap

MUSA is more than just a building; it is an ecosystem.

GPU is a systematic project with high R&D barriers and a long industrial chain that includes hardware architecture, driver development, software ecology, sales applications, and so on. After decades of evolution, the current GPU ecosystem has grown enormously large and complex. To complete the commercial application of a GPU from research and development to market, it is inextricably linked to ongoing investment in software and hardware, as well as strong environmental support.

Moore Threads officially released the first multi-function GPU chip “Sudi” in March of this year, which has been recognized by many markets and ecosystems. Moore Threads PES Perfect Experience System Alliance partners include CPU, operating system, OEM manufacturers, software service providers, cloud service providers, and system software developers, and the list is growing. Moore Threads and OEM partners have successfully launched a variety of personal computers, workstations, and data center server products based on the “Sudi” chip, which are used in daily office, digital twin, artificial intelligence training and reasoning, and other business scenarios; it also collaborates with cloud services. Manufacturers offer GPU cloud computing capabilities to users in various industries, paving the way for Moore Threads GPUs to be used in a variety of applications. China Mobile Cloud Competence Center has signed a strategic cooperation memorandum with Moore Threads to conduct extensive collaboration in the fields of cloud computing, cloud applications, and accelerated computing; China Telecom Research Institute will collaborate with Moore Threads on intelligent computing power, technology research and development, application scenarios, and so on. Explore the new Metaverse infrastructure and localized ecology together, and promote the use of Metaverse technology in key business scenarios.

Zhang Jianzhong, founder and CEO of Moore Threads, expressed his appreciation to all partners and users present, and added: “GPU entrepreneurship is a long-term career full of challenges, and we understand the value of ecology. Moore Thread’s multi-purpose GPU is built on The advanced MUSA architecture is still constructing a complete software stack and application ecosystem, with the goal of providing friendly support and experience for the open ecosystem. Only by working together with environmental partners and industry users will we be able to fully utilize Moore Thread’s computing power. Give the metaverse and digital economy core power.”

MTT S80, the first national tide game graphics card, and the “Chunxiao” chip

Moore Threads officially released the second multi-function GPU chip “Chunxiao,” integrating 22 billion transistors, a built-in MUSA architecture general computing core, and a tensor computing core, which can support FP32, FP16, and INT8 and other computing precision. The four built-in computing engines of “Chunxiao” have been comprehensively upgraded in comparison to the previously released “Sudi” chip, resulting in significant performance improvement: The graphics rendering ability is increased by an average of three times; the encoding ability is increased by four times, and the decoding ability is improved by two times; the AI calculation acceleration is increased by four times, and physical simulation calculation is increased by two times.

The Moore Threads MTT S80 is the first national trend graphics card for gamers, and it is based on the “Chunxiao” GPU chip. It has 4096 programmable MUSA cores and can deliver 14.4TFLOPS of single-precision floating-point computing power at 1.8GHz. At the same time, the MTT S80 is the industry’s first graphics card with a PCIe Gen5 interface. With 16GB of GDDR6 high-speed video memory and 8K UHD and 1080P 360Hz high refresh rate display output capability, it can provide gamers with the ultimate game vision and operation experience.

Moore Threads is the first GPU company in China to support the Windows environment and the DirectX graphics interface, thanks to the successful launch of MTT S80. Its powerful 3D graphics rendering capabilities will provide users with a smooth operating experience in Windows DirectX games at 4K resolution. At the moment, the MTT S80 Windows driver includes the MUSA DirectX Driver module and has completed the adaptation of dozens of popular games such as “Diablo 3,” “League of Legends,” and “Cross Fire.”

Moore Threads demonstrated the smooth game running effect brought by MTT S80 during the press conference. Moore Threads is currently working closely with game engine developers like Unreal and Unity, as well as top domestic game developers like Tencent Games, NetEase Games, Xishanju, Perfect World, and 360 Games (in no particular order). Game products provide better and faster support, allowing players to enjoy a AAA gaming experience that is constantly updated. Moore Threads will continue to update the Windows driver and MUSA DirectX version in the future to improve game compatibility and performance.

Moore Threads announced on the spot that the MTT S80 graphics card has finished the first batch of production and stocking, and that limited sales will begin on the Jingdong e-commerce platform on November 11, 2022.

MTT S3000 is a new multi-function server GPU product.

The newly released MTT S3000 is based on Moore Threads MUSA architecture and is also the first “Chunxiao”-based multi-function server GPU product. Moore Threads MTT S3000’s diverse computing power, combined with the full MUSA software stack covering graphics rendering, video processing, and deep learning, can provide AI inference and training, cloud gaming, cloud rendering, video cloud, digital twin, digital content creation, and other scenarios. The goal of general intelligent computing power support is to lay a solid computing power foundation for the construction of data centers, intelligent computing centers, and meta-computing centers, as well as to aid in the development and implementation of multiple applications in the Metaverse.

MTT S3000 has 4096 MUSA stream processing cores and 128 dedicated tensor computing cores, with a transistor scale of 22 billion, an operating frequency of 1.9GHz, and a memory bit width of 256bit; the bandwidth is 448GB/s with 32GB GDDR6 memory; it supports FP32, FP16, INT8 and other computing precision, with FP32 computing power reaching 15.2TFLOPS.

Moore Threads MTT S3000 and its supporting software and hardware products can cover the entire process from algorithm model to application deployment, providing AI users with a friendly and comprehensive package of solutions. MTT S3000 has multi-dimensional advantages in terms of deep learning training with the support of the MUSA computing platform, such as ease of use, scalability, and compatibility. MTT S3000 supports mainstream AI models in vision, speech, natural speech understanding, and multimodality for deep learning inference. At the same time, Moore Thread’s CUDA ON MUSA compatible solution allows users to seamlessly migrate CUDA-developed code to MTT S3000. Moore Threads also performed extensive performance optimization on the MUSA software stack and launched TensorX, a self-developed AI inference engine.

The advancement of AI applications requires ecological collaboration. MTT S3000 is currently compatible with a variety of mainstream deep learning frameworks, including PyTorch, TensorFlow, Baidu PaddlePaddle, Jittor, and others, and has enabled the acceleration of dozens of AI models, including Transformer, CNN, and RNN. Moore Threads also announced strategic partnerships with Baidu PaddlePaddle, Luchen Technology, Jittor, OpenMMLab, and Zhiyuan Research Institute (in no particular order) to help the AI ecosystem thrive.

Accelerates GPU application development while also providing extensive software stacks and tools.

The importance of software ecology in promoting GPU computing cannot be overstated. Moore Threads has released a complete MUSA software stack to serve a large number of developers and end users, based on the MUSA architecture. Graphics rendering, multimedia, artificial intelligence, physical simulation, and general computing are among the functional modules included in the MUSA software stack. It will cover everything from underlying drivers to GPU acceleration libraries to application development kits tailored to specific industries, and it is committed to meeting a wide range of industry scenarios. requirements for application development

MUSA development kit: contains the MUSA general computing driver, the MUSA compiler, the AI operator library, the general computing library, performance analysis tools, and other components. MUSA developers can use the MUSA development kit to create, optimize, and deploy applications on a variety of platforms, including desktop PCs, workstations, enterprise data centers, cloud platforms, and distributed computer clusters based on the Moore Threads multi-function GPU.

CUDA ON MUSA: Moore Threads has created a CUDA ON MUSA compatible solution for CUDA language users. The CUDA source code can be run on the Moore Threads MUSA architecture GPU by compiling and running two steps.

MUSA user software: It provides a multi-level series of software for the majority of users, including: drivers for different system platforms, PES and other platform software series; optimization tools, containerized deployment runtime libraries, elastic segmentation scheduling, and so on; as well as cloud computing and application software series such as Digital Human, MT OCR, and MT Smart Stream.

The Moore Threads developer website (https://developer.mthreads.com/) is officially launched to make it easier for developers to obtain the Moore Threads series software stack, application solutions, and technical support. This will be the release and download site for MUSA software products. The website also serves as the MUSA technology exchange community. Moore Threads will create an open application and developer ecosystem centered on the MUSA platform to promote the growth and prosperity of the GPU ecosystem.

 

A cloud-native GPU will be used to accelerate intelligent cloud applications.

GPU is the primary computing power required by today’s data centers, and cloud-native technology applications in cloud computing, cloud desktop, cloud games, and other fields are becoming more common. Moore Threads has released a series of GPU cloud-native solutions based on Moore Thread’s innovative MT Mesh 2.0 technology to that end.

MT Mesh 2.0 can automatically allocate GPU computing and video memory resources based on the cloud center’s application load, enabling elastic scaling of GPU computing power. It can divide a GPU arbitrarily into multiple containers or virtual machines, or it can support only one container or virtual machine. Schedule several GPUs.

Elastic Containerized GPU (sGPU): MT Mesh 2.0, which is based on the Kubernetes ecosystem, is used to achieve flexible and efficient allocation of containerized GPU resources.

Elastic virtualized GPU (vGPU): It pioneered resource elastic segmentation technology in the industry by utilizing MT Mesh 2.0. It is capable of dynamically allocating and modifying GPU virtualization resources without restarting, as well as achieving computing power on-demand, dynamic scaling, and exhaustion. Release; support hardware virtualization (SR-IOV), secure physical segmentation, up to 32-way virtualization, Windows cloud desktop GPU virtualization, and GPU pass-through of Tongxin and Kylin operating systems.

Cloud acceleration technology for Android containers ACX: It can accelerate Android cloud mobile phone solutions, reduce application delay, and increase the number of concurrent channels by utilizing GPU Android container transparent transmission technology and rendering and coding integration technology. It can provide better Android cloud game rendering effects and compatibility features by natively supporting OpenGL ES, OpenGL rendering framework, and hardware acceleration of Android game material compression algorithms such as ETC/ETC2.

Cloud desktop and cloud games are important GPU in the cloud application scenarios. Moore Threads has collaborated with many cloud desktop ecosystem partners to provide low-cost cloud desktop solutions that serve the majority of users in education, office, finance, geographic information, and other industries. Moore Threads and Tencent Pioneer have collaborated to create a leading Android cloud game solution, as well as an excellent PC cloud game solution with Weiling Times, and are committed to bringing high-quality and smooth cloud games to the majority of game enthusiasts. gaming expertise

 

Metaverse platform MTVERSE is developing a full-scene Metaverse application scheme.

Moore Threads also unveiled the MTVERSE metaverse platform, as well as a slew of software and hardware products, including the MCCX metacomputer all-in-one based on MTT S3000, at the press conference.

MTVERSE’s computing power base is the Moore Threads MUSA GPU cluster, which provides users with computing infrastructure services such as three big data platforms, AI training and reasoning, graphics rendering, and physical simulation, as well as hardware clusters, software infrastructure, and SDK toolchains. Metaverse’s full-stack solution addresses multiple core elements in the metaverse, including people, scenes, and content. The vast SDK tools at the upper layer can assist developers and applications in calling these capabilities and realizing a variety of functions such as digital human, document understanding, speech recognition, visual recognition, natural semantic understanding, dialogue interaction, physical simulation, AIGC content generation, and so on, further simplifying the development cycle and difficulty of applications and solutions.

AlphaCore physical simulation engine upgrade: two new physical simulation products, Flood Dynamics and Storm System, have been added, as have the effects of the gas fluid simulation tool CatalystFX and the cloth and hair production tool VeraFiber. Moore Threads has announced that it is deeply integrated with the cloud-native rendering engine RaysEngine to enable digital twin real-time 3D rendering and physical simulation, as well as the creation of more realistic digital scenes.

Upgrade the DIGITALME digital human solution to include the “Nuwa” digital human generator, the “Painted Skin” expression-driven engine, the “Suiying” digital human action-driven engine, and the “SuiAn” digital human dialogue system.

AIGC Content Creator Ma Liang: Provides users with a zero-threshold AIGC creation platform and supports bilingual graphic generation and graphic editing in both Chinese and English. Moore Threads also collaborated with Yunnan Art Institute to provide artists with end-to-end “computing power and algorithm” solutions, co-founded the “Design College AI Art Creation Laboratory,” and launched the Ma Liang course to train design students and aid in the creation of art.

MCCX All-in-One Metacomputing Machine: Its goal is to provide powerful and user-friendly meta-computing computing power. MCCX has designed a reasonable system architecture and resource ratio to provide users with high-end computing power for AI, rendering, encoding and decoding, and other meta-computing power application scenarios. A cost-effective meta-computing hardware solution; and using an integrated software and hardware delivery method, pre-installing a complete basic software stack, development environment, AI and rendering framework, supporting one-click application deployment and upgrade, and realizing meta-computing computing power out of the box.

About Moore Threads

Moore Threads is a high-tech integrated circuit company that specializes in GPU chip design. Its primary focus is the development and design of full-featured GPU chips and related products that can provide powerful computing acceleration capabilities to Chinese technology partners. Founded in October 2020, the company is dedicated to developing a new generation of GPUs for meta-computing applications, constructing a comprehensive computing platform that integrates visual computing, 3D graphics computing, scientific computing, and artificial intelligence computing, and establishing an ecosystem based on cloud-native GPU computing to help drive the digital economy’s development.

About UnixCloud

Moore Threads is distributed by Unixcloud Technology (Shenzhen) Co., Ltd. Unixcloud has a long history of product development and manufacturing. Moore Threads’ official authorized distributor is UnixCloud. It focuses on the field of edge computing in response to the computing power requirements of artificial intelligence development, and it offers the industry edge computing products and solutions that meet a variety of AI requirements. UnixCloud launched the 10G network card business at the same time, offering four-port and two-port 10G optical fiber network cards based on the Mucse-based network controller N10.