X-Silicon Inc. (XSi) has created a new RISC-V microprocessor chip architecture that combines RISC-V CPU cores with vector capabilities and GPU acceleration into a single chip. Jon Peddie Research reports that the CPU/GPU hybrid chip is open-source, and it's designed to handle a variety of different functions, including artificial intelligence, which is typically handled by dedicated CPUs and GPUs. The problem is that it should do all this in a more efficient way.
The new CPU/GPU hybrid processor is designed to be a "one-size-fits-all" processor. According to JPR, the industry has been looking for an open-standard GPU that is flexible and scalable enough to support a variety of markets, including virtual reality, automotive, and IoT devices. This new RISC-V CPU/GPU is designed to solve this problem by providing manufacturers with a single, open-chip design that can handle any desired workload.
X-Silicon's chips differ from other architectures in that they are designed to combine the capabilities of the CPU and GPU into a single-core architecture. This is different from the typical designs of Intel and AMD, which have separate CPU cores and GPU cores. Instead, the cores themselves are designed to handle both CPU and GPU tasks. In that sense, it sounds a bit like the Larabee project that Intel abandoned that tried to use x86 for graphics and other workloads.
The chip uses X-Silicon's C-GPU architecture, which incorporates GPU acceleration into RISC-V vector CPU cores. The architecture features a RISC-V vector core with a 32-bit FPU and scaler ALU. It has a thread scheduler, a clipping engine, a rasterizer, a texture unit, a neural engine, and a pixel processor. The chip is designed to handle applications including artificial intelligence, high-performance computing (HPC), geometric computing, and 2D and 3D graphics.
Theoretically, X-Silicon's hybrid chips are capable of processing both CPU and GPU code in the same core, which gives it a number of advantages. The chip uses the open-source RISC-V ISA as the CPU and GPU, running a single instruction stream. This provides low-memory footprint execution and greater efficiency because there is no need to copy data between CPU memory space and GPU memory space.
CPU/GPU cores can be combined to form a multi-core design, allowing manufacturers to scale processing power as needed. In a multicore format, multiple cores are tiled on a single chip and connected using a fast fabric. Fast on-chip SRAM or eDRAM caches are also implemented in this design, which act as L2 caches that can aggregate data from multiple cores. Each core can be scheduled to run graphics, AI, video, physical, HPC, or other workloads independently of the others as needed.
With this design, X-Silicon's C-GPU architecture can run any type of CPU or GPU workload. X-Silicon claims to already have a Vulkan graphics API for use with "converged GPU acceleration". This will greatly help in its development and adoption on Android devices.
Since the new design is based on RISC-V, anyone can use the architecture without having to pay instruction set royalties – unlike x86 and ARM. If it works as intended, these chips could shake up the microprocessor industry. Theoretically, the standard designs currently in use are not as flexible or robust as X-Silicon claims.
Although we may not have to wait long to know, it remains to be seen whether it will be as effective in practice as it is on paper. The SDK will reportedly be released to early partners sometime this year.
X-Silicon's low-power, open-standard, Vulkan-enabled C-GPU
X-Silicon Inc. (XSi) showcased its open-standard, low-power C-GPU architecture, which combines GPU acceleration with RISC-V vector CPU cores and tightly coupled memory to form a low-power, single-processor solution. It is an open-source version of its unified RISC-V vector CPU-with-GPU ISA and provides register-level hardware access through the Hardware Abstraction Layer (HAL). The NanoTile architecture in the XSi C-GPU is designed to handle real-time data processing and motion graphics rendering, overcoming the limitations of traditional GPU architectures.
For more than 20 years, the industry has sought an open-standard GPU that is flexible and scalable enough to support a variety of markets, such as AR/VR, automotive, connected IoT, and vast embedded verticals, including robotics. X-Silicon's dynamic and highly scalable C-GPU NanoTile architecture and its ability to process multiple tasks simultaneously or sequentially are unique in this regard.
X-Silicon Inc (XSi), a San Diego-based startup founded in March 2022, unveiled its latest innovation: an open-standard, low-power C-GPU architecture that incorporates GPU acceleration into RISC-V vector CPU cores, tightly coupled memory, and delivers a low-power, single-processor solution. XSi's approach introduces the open-source of its unified RISC-V vector CPU-with-GPU ISA and provides register-level hardware access through the Hardware Abstraction Layer (HAL). This allows OEMs and content providers to tailor drivers and applications with unusual customization, unlike competitors' closed solutions, the company said. XSi believes that it pioneered the adoption of Vulkan with converged GPU acceleration on RISC-V to facilitate development on Android devices, which is critical to effectively addressing the diverse embedded vertical market.
XSi's C-GPU uses the NanoTile architecture, which the company says can meet the needs of real-time data processing and motion graphics rendering. Traditional GPU architectures struggle to handle data in motion, so a novel approach is needed. XSi says its processors are designed for optimal management and rendering of dynamic content, surpassing the limitations of traditional GPUs. By integrating AI/ML computing and GPU rendering, NanoTile ensures efficiency and adaptability, making it ideal for implementing advanced graphics algorithms.
The company, which holds 14 patents, claims that its NanoTile architecture has revolutionized AI/ML computing and GPU rendering. These patents focus on optimizing the flow of data between processing cores and memory, reducing latency, and improving computational efficiency. In addition, NanoTile's patented IP can be deployed in both edge and cloud configurations, nurturing federated models of GPU computing and providing flexibility and scalability.
XSi's open standards, low-power C-GPU architecture, and NanoTile platform herald a paradigm shift in GPU technology. With support for open standards, customizable hardware access, and dynamic content rendering methods, XSi believes it will set a new standard for GPU architectures, enabling developers and OEMs to unlock unprecedented levels of performance and efficiency in graphics rendering and AI/ML. Computing applications.
The company reports that the RISC-V ecosystem has responded positively to the launch of a new computational graphics company that is fully committed to advancing the open standards ecosystem.
The company plans to make its software development kit available to a select group of early-stage development partners later this year.
Startups bring MIMD to the world of graphics and computing
Formed by former Silicon Valley experts, X-Silicon aims to revolutionize GPU design with a RISC-V vector-based unified graphics computing engine (C-GPU) capable of performing AI, HPC, and 2D/3D graphics tasks. Its MIMD architecture executes CPU and GPU code independently within a single core, reducing memory usage and improving performance. The company's multi-core layout features a fast synthesizer structure that enhances data aggregation for different applications. X-Silicon seeks to reduce GPU latency through near-memory computing and novel hardware setups, backed by 14 patents. X-Silicon supports standard APIs and open programming for AR/VR, embedded devices, automotive, and more, enabling rapid development. Initial IP sales are aimed at OEMs and hyperscalers.
What do we think? The market for some new architectures is ripe. The Big Three have squeezed everything out of traditional SIMDs and added extra cores such as matrix math (sometimes referred to as tensor cores), ray-traced interference test engines, codecs, and audio processors. X-Silicon believes it can meet many, if not all, of these purpose-built processor needs with a tightly coupled block communication MIMD architecture. To do this, they will make use of the RISC-V ISA.
In 2022, a group of engineers, architects, programmers, and business developers from AMD, Qualcomm, Intel, ATI Technologies, Dell, and others are hard at work. A new GPU design was developed and realized that they should make a name for their company. As newbies, they know that startups waste a lot of time trying to get a smart, memorable name and logo. So I chose X. It can be experimental, sexy, or unknown – hopefully it's all true.
To revolutionize GPU shader cores, X-Silicon says it is creating a new scalable RISC-V vector-based Unified Compute Graphics Engine (C-GPU) that can efficiently compute next-generation workload types that traditional GPUs can't.
Such applications include artificial intelligence, high-performance computing, vision, geometric computing, and 2D and 3D graphics. The company says its MIMD architecture has the unique ability to run CPU and GPU code independently in the same core, providing features such as low-memory footprint execution, hardware register bare-metal programming, high performance, low-power operation and substitution using a single instruction stream, combining traditional shader programs with the open-source RISC-V ISA for CPUs and GPUs.
In the company's multi-core design, multiple C-GPU cores are tiled on a single chip and connected using an on-chip fast synthesizer structure that dynamically aggregates the output of each core into a common buffer, either a frame buffer for graphics use cases or a pipeline buffer for codecs, video effects processing, and AI processing, as shown in the figure below.
In this design, a fast on-chip SRAM or eDRAM cache will be used as a L2 cache that can aggregate data from multiple cores. Common operations that compute RAM (C-RAM) are done near memory, further reducing bandwidth and further improving performance. The company claims that each core can be software programmed to compute graphics, artificial intelligence, video, physics, high-performance computing, or other workloads independently of all other cores.
As a result, workloads can be implemented in parallel or pipelined, and run simultaneously on cores, rather than sequentially on traditional GPUs. X-Silicon says it can also run the operating system on the core.
The company claims that it can also accelerate computing with near-memory computing, unified memory architectures, and other novel hardware configurations that reduce the latency inherent in GPUs. They have filed 14 patents for this.
Eric Powers, former Disney/Applied Minds/Giant AI animation industry expert, commented, "For decades, high-end animations and effects have not been able to switch to GPUs to generate final images. The sheer complexity and scale of professional rendering pipeline software, combined with the huge cost of a dedicated GPU design to cross memory and platform barriers, has completely hindered the mass adoption of GPUs in our state-of-the-art technology. Integrated design gives HPC developers in-place access to direct hardware acceleration, such as X-Silicon's C-GPU architecture, which is the only future that will lead us across this boundary. ”
X-Silicon's C-GPUs offer market opportunities for emerging use cases for OEMs looking to control their compute and GPU fortunes. It allows for new APIs (including custom APIs) as well as an ecosystem tailored for the application, rather than bootstrapping the API to perform tasks it was never intended to do. It no longer requires OEMs and manufacturers to succumb to the Big Five in a stagnant ecosystem. The company sees market opportunities in emerging markets such as entertainment and enterprise AR/VR with longer battery life, smart embedded devices with display needs, low-cost automotive displays and modules that require predictable dedicated processing, wearables, custom animation and others.
The company plans to support direct hardware and pixel access, so for low-memory applications, there is no need for bulky drivers. The company says X-Silicon's open standards approach and its own open programming model will help develop new use cases quickly and easily, as well as improve existing products. Of course, X-Silicon also plans to support all APIs in the traditional software ecosystem, including OpenGL ES, Vulkan, Mesa, and OpenCL, but the company will also provide a Hardware Abstraction Layer (HAL) that will allow direct access to others who are optimizing open source or creating their own drivers and custom APIs. This is particularly interesting because the architecture supports emerging technologies, including new rendering models that are not supported by traditional architectures, such as Neural Radiance Fields (NeRF) and non-triangular primitives.
The company plans to sell IP to OEMs, hyperscalers and other system integrators first. The launch date of the first wafer has not yet been given.
By integrating CPUs and GPUs (c-GPUs) with a single ISA and open graphics operating system (GOS) platform, X-Silicon can have a profound impact on the overall software development, support, and maintenance of next-generation graphics rendering. This could usher in an exciting era of innovation in the future of graphics, providing new approaches to graphics algorithms, performance, power, flexibility, and cost for emerging and traditional market segments. This approach has revolutionized the graphics world, freeing OEMs from the shackles of traditional GPU vendors who offer black box drivers with complex APIs and expensive legacy support.
Traditional GPU SIMD architectures are constrained by host CPUs, operating systems, and graphics services, limiting innovation and helping to maintain incumbent enterprise control of their markets. New, emerging, smaller vertical markets, often unserved by these traditional graphics vendors, can develop and support compelling graphics solutions that can be upgraded and maintained throughout the product lifecycle. Next-generation products that use new display technologies, new formats, and usage paradigms (VR/AR, 360, depth, stereoscopic, multiplane holography) often require a new approach to rendering. New development and deployment paradigms also require consistent architectures, such as those that scale from the edge to the cloud.
More than just an open-source alternative to traditional GPU vendors, X-Silicon is poised to offer a new technology graphics processing framework that incorporates the latest artificial intelligence and beyond triangle-based rendering to deliver an innovative platform not seen since the introduction of 3D graphics on mobile devices at the beginning of this century.
Reference Links
https://www.tomshardware.com/pc-components/cpus/former-silicon-valley-vets-create-risc-v-microprocessor-that-can-run-cpu-gpu-and-npu-workloads-simultaneously
https://www.x-silicon.com/news/
https://www.zonepeddy.com/news/s-silicones-low-power-open-standard-vulcan-enabled-c-gpu/
来 源 | 半导体行业观察（ID：icbank）编译自Tomshardware
☞ Business Cooperation: ☏ Please call 010-82306118 / ✐ or to [email protected]