Panmnesia has received the CES 2024 Innovation Award

Panmensia, a CXL Intellectual Property (IP) leader, announced that it has been bestowed with the CES 2024 Innovation Award for its pioneering CXL-enabled AI accelerator. This recognition came ahead of CES 2024, the world’s premier technology event, set to unfold in Las Vegas. The CES Innovation Awards program is the pinnacle of recognition at CES, bestowed by the Consumer Technology Association (CTA), the event’s owner and producer. Each year, the program curates products from across the globe, seeking out the very best in technology, design, and innovation. Panmnesia won the award for its CXL-enabled AI Accelerator, which will be exhibited at the upcoming CES 2024 event. Panmnesia’s CXL-enabled AI accelerator, the game changer for computing in the large-scale AI era The award-winning accelerator features virtually limitless memory capacity that brings significant performance boost for large-scale AI-driven services. This innovative feature empowers AI-driven services to handle increasing amount of data, leading to significant enhancements in accuracy and quality of the services. Unlike conventional AI accelerators constrained by their limited memory capacity, Panmnesia’s accelerator shatters this barrier, enabling AI services to operate at an unprecedented level of sophistication. This holds a great potential for the AI industry, where the accuracy and quality of AI-driven services are a paramount of success in the competitive landscape. The backbone of this innovation is Compute eXpress Link (CXL), an interface technology that connects an array of system devices, such as accelerators, memory expanders, processors, and switches. By connecting multiple accelerators and memory expanders through CXL switches, CXL can provide memory capacity to the system that satisfies the demands of AI applications. Panmnesia's AI accelerator design allows the accelerator to fully leverage CXL by incorporating a CXL 3.0 controller into the accelerator chip. Furthermore, Panmnesia maximized application performance by tailoring the accelerator's hardware architecture to the specific application. Notably, they chose AI-powered search, a representative large-scale AI application, as their primary acceleration target. Considering that AI-powered search necessitates frequent vector processing, Panmnesia optimized their acceleration architecture to maximize parallelism for vector processing while designing a low-overhead host interface. By combining CXL technology with specialized hardware, Panmnesia has achieved a 101x speedup over existing solutions that utilize SSDs to accommodate large volumes of data. Additionally, Panmnesia emphasized that the accelerator offers significant operating expenses savings for the system, as their specialized hardware reduces energy consumption for data movement and calculation. "We are excited to share that our technological leadership and innovation have been recognized by experts worldwide," said Myoungsoo Jung, CEO and founder of Panmnesia. "We believe that our CXL technology will be a cornerstone for next-generation AI acceleration system. We remain committed to our endeavor revolutionizing not only for AI acceleration system, but also other general-purpose environments such as data centers, cloud computing, and high-performance computing," he added. More details about Panmnesia's award-winning technology will be revealed at the upcoming CES 2024 event in Las Vegas on January 9-12 next year. CES 2024 Innovation Award

Panmnesia on SC23

At the Supercomputing Conference 2023 (SC23), Panmnesia, a leader in CXL Intellectual Property, unveiled the pioneering CXL 3.0 All-in-One Framework. SC23, a pivotal event in the HPC industry, was hosted at the Colorado Convention Center from November 14-16, drawing key players like Google, AWS, and Microsoft. Panmnesia's exhibit 1 Panmnesia's exhibit 2 Panmnesia's exhibit 3 Panmnesia's exhibit 4 Compute eXpress Link (CXL), a cornerstone of Panmnesia’s expertise, is an interface technology that connects an array of system devices, enabling on-demand memory utilization and cost-efficient memory management for HPC and data center environments. Panmnesia's booth 1 Panmnesia's booth 2 Panmnesia's booth 3 Panmnesia’s CXL 3.0 All-in-One Framework Panmnesia’s latest endeavor, the CXL 3.0 All-in-One Framework, was a highlight at SC23, showcasing their ongoing dedication to CXL technology innovation. Dr. Myoungsoo Jung, the CEO and founder of Panmnesia, expressed his enthusiasm for the framework’s launch. “We’re thrilled to introduce our renewed framework,” he said, highlighting the framework’s enhanced internal architecture and advanced CXL 3.0 IPs. This development aims to provide an efficient and comprehensive environment for customers to develop custom CXL software and hardware solutions. This framework features a full array of both hardware and software components, meticulously designed for robust development processes. The hardware suite boasts state-of-the-art CXL 3.0 devices, including a CXLenabled CPU, a CXL switch, and various CXL endpoints. These devices are intricately constructed using Panmnesia’s proprietary CXL 3.0 IPs and are fully compatible with all CXL protocols, such as CXL.io, CXL.cache, and CXL.mem. Panmnesia underscores the significance of these devices, particularly their ability to leverage unique CXL 3.0 features, including memory sharing capabilities. A pivotal component of the framework is the CXL switch, tasked with the crucial function of connecting devices according to user specifications. To enable configurable connectivity, Panmnesia has meticulously designed their CXL switch to incorporate a sophisticated software known as the fabric manager. This software is adept at managing internal routing rules within the switches. Moreover, the switch’s design is not limited to interconnecting devices within the framework; it can also extend its connectivity to include external devices or even another switch. This versatile feature allows users to effortlessly scale their framework, facilitating the construction of systems that closely resemble actual production environments. Panmnesia’s CXL 3.0 All-in-One Framework not only provides advanced hardware but also delivers a comprehensive software stack crucial for executing user-level applications with CXL enhancements. This includes a specialized OS based on Linux, complete with CXL device drivers and a CXL-aware virtual memory subsystem, enabling applications in AI and computational science to fully utilize CXL without requiring code changes. The framework garnered significant attention at SC23, appealing to those looking to incorporate CXL into their solutions. It received positive feedback for its holistic approach, offering both hardware and software capabilities. Software developers were particularly impressed with its readiness for user-level applications, while hardware developers valued its support for interoperability testing, a key aspect in CXL hardware development and testing. This dual appeal underscores Panmnesia’s success in providing a comprehensive, ready-to-use solution for CXL technology integration. Panmnesia's CXL3.0 All-in-One Framework Memory sharing demonstration CXL 3.0 memory sharing demonstration At the booth, Panmnesia showcased a demonstration of memory sharing using CXL 3.0. This was conducted using their CXL 3.0 All-in-One Framework. The demonstration focused on a plasma simulation, an HPC application, to illustrate the benefits of CXL memory sharing. This simulation involves modeling the movement of charged particles by calculating the forces between them. For efficiency, the simulation is typically parallelized by dividing the space into smaller sections and assigning these to different hosts. In order to accurately compute the forces on a particle, the host responsible for that particle must gather information like charge and velocity from neighboring particles. These neighboring particles might be located on different hosts, necessitating data exchange among hosts. Traditional methods for this exchange rely on network communication, which can be slow due to network latency, a known issue in HPC scientific applications. During SC23, Panmnesia introduced a solution based on CXL 3.0 that reduces the overhead of inter-host data exchange. This solution utilizes the memory sharing feature of CXL 3.0 to facilitate data exchange. It enables multiple hosts to access data stored in CXL-attached memory. Panmnesia’s approach, which automates key memory sharing functions, enhances performance significantly. They also shared details of their software implementation, which is built on a popular data exchange library. Panmnesia highlighted the superiority of their CXL 3.0-based solution over traditional network-based methods by demonstrating a notable 1.8x speed increase. They pointed out that the utility of their solution extends beyond plasma simulation, encompassing a broad spectrum of applications such as molecular dynamics, bioinformatics, and computational fluid dynamics, making it broadly applicable in various scientific domains within HPC environments. Plasma simulation demonstration 1 Plasma simulation demonstration 2 Panmnesia’s introduction of this framework marks a pivotal moment for the CXL ecosystem, facilitating accelerated development for both software and hardware enterprises. Furthermore, Panmnesia has solidified its position as a leader in CXL technology with this industry-first public demonstration of CXL 3.0 memory sharing in a complete system, inclusive of a CXL switch. For those interested in exploring Panmnesia’s All-in-One framework and related demonstrations further, videos are available on Youtube (refer to the link below). Demo Video (Youtube)

Panmnesia will participate in SC23 exhibition

Panmnesia is excited to announce the participation in Supercomputer (SC) 2023, which will take place in Denver, Colorado, US from November 12th to November 17th. During this event, Panmnesia will unveil the groundbreaking CXL 3.0 all-in-one framework to the global audience. Panmnesia’s CXL framework solution, featuring the latest CXL 3.0 technologies, enables an infinite memory pool shared by multiple hosts through Port-based Routing(PBR) functionality. With our innovative solution, users can experience efficient data sharing across infinite memories. Panmnesia’s CXL all-in-one framework provides both CXL 3.0 hardware stack (CXL CPUs, switches, and DIMM pools) and its corresponding Linux-based CXL 3.0 software stack. Users can seamlessly apply their applications without modification and benefit from the advanced CXL 3.0 features within Panmnesia’s all-in-one framework. Therefore, for companies who are interested in CXL technology, this framework would be a best option to develop their own CXL hardware/software solutions. Panmnesia will also showcase demonstration accelerating practical HPC applications via memory sharing feature of CXL 3.0. Join us to explore our framework and demo at booth 2253. You can find more detailed information about SC 23 exhibition at the link below. Link - SC23

Panmnesia Has Been Featured in the CXL Consortium Member Spotlight Blog

CXL Consoritum operates a blog that highlights leading companies in the field of CXL technology. Recently, the CXL Consortium’s member spotlight blog featured Panmnesia. The post provides a brief introduction to Panmnesia and outlines Panmnesia’s expertise. It also delves into Panmnesia’s perspective on CXL technology. Specifically, Panmnesia shares insights into the ideal use cases for CXL technology, the market segments that will benefit the most from it, and how CXL will reshape data center configurations. The post concludes by explaining how Panmnesia’s CXL technology can assist other companies in adapting to this emerging CXL landscape. For the complete blog post, you can follow the link provided below. Link - CXL Consortium Member Spotlight Blog

Panmnesia's Remarkable Achievement - Raising $12.5 Million in Seed Funding with a Valuation Exceeding $81.4 Million in the CXL Semiconductor Arena

Recently, Panmnesia, a fabless semiconductor company focused on CXL (Compute Express Link) technology, proudly announced its successful completion of a seed funding round amassing $12.5 million (16 Billion KRW). This significant financial feat led to the company’s post-money valuation soaring to an impressive $81.4 million (103.4 Billion KRW). In the context of South Korea’s semiconductor sector, achieving such a hefty valuation in the seed round stage is indeed exceptional. Leading the investment charge was Daekyo Investment, and the round witnessed contributions from a consortium of seven esteemed investment entities including SL Investment, Smilegate Investment, GNTech Venture Capital, Time Works Investment, Yuanta Investment, and Quantum Ventures Korea. Behind Panmnesia's Evolution Deeply rooted in academic excellence, Panmnesia was born out of the collaborative efforts of a talented group of Ph.D. graduates from the Korea Advanced Institute of Science and Technology (KAIST). They were guided under the expert mentorship of Professor Myoungsoo Jung. With its foundation firmly set on groundbreaking research and innovation, Panmnesia is committed to offering a comprehensive suite of solutions encompassing CXL semiconductor IP, leading-edge data processing acceleration solutions, and holistic CXL-driven technologies. CXL Interface: The Next Focus for Semiconductor Companies CXL is rapidly gaining traction as a transformative interface technology designed to seamlessly connect diverse system devices, including memory expansion units, accelerators, processors, and switches. Its promise lies in the ease it offers in memory capacity augmentation and its prowess in streamlining data processing. One standout feature of CXL, memory pooling, stands as a beacon for data center efficiency, presenting avenues for substantial cost savings in memory-related operations. Given these advantages, tech giants such as Meta, Microsoft, and Google are demonstrating heightened interest in CXL-centric solutions. Aligning with this trend, global semiconductor powerhouses including Intel, AMD, Samsung Electronics, and SK Hynix have laid out their strategic blueprints to introduce CXL-compliant devices. Echoing this widespread enthusiasm, market projections hint at the CXL domain expanding to a robust $20 billion by the year 2030, placing Panmnesia in an enviable position to capitalize on these unfolding opportunities. Panmnesia: Revolutionizing the CXL Landscape In the fiercely competitive CXL market, where numerous global semiconductor companies are contending, Panmnesia stands alone as a domestic firm making unparalleled advancements. Panmnesia's efforts came into the limelight when they presented the world's first CXL 2.0-based full-system memory pooling framework at the USENIX ATC conference in Boston last year. Extending its technical lead, the company unveiled a solution this May that accelerates data center-oriented Artificial Intelligence (AI) applications, using CXL 3.0 technology. In a more recent development, Panmnesia made another world's first at the Flash Memory Summit 2023 in Santa Clara, California. The company disclosed a full system, from devices to operating systems, connected through a multi-level switch architecture supporting CXL 3.0. As the only enterprise that has publicly demonstrated CXL 3.0 hardware and software technologies, Panmnesia has garnered attention from leading global IT firms such as AMD and Meta, affirming its front-runner status in CXL technologies worldwide. Myoungsoo Jung, CEO of Panmnesia, elaborated, "CXL 3.0 takes a fundamentally different viewpoint from CXL 2.0 in its approach to various processors and memory semiconductor devices to support high scalability. Regrettably, most companies are still only discussing certain features of CXL 1.1 or 2.0." Jung added that Panmnesia has already completed the development of hardware and software to support CXL 3.0, which will be of significant assistance to a diverse range of companies utilizing CXL. He continued, "The CXL ecosystem has yet to mature fully. Without a comprehensive support framework spanning hardware, software, and the full stack, it's difficult to harness the proper performance capabilities that CXL offers." Jung emphasized Panmnesia's distinct advantage: the company's CXL solution covers everything, including CXL IP, hardware, system solutions to enable these, and software essential for accelerating AI and data processing at the server and data center levels. These solutions are planned to be available to a broader customer base following the completion of state-of-the-art silicon tape-out later this month. Looking forward, Panmnesia aims to continue leading the CXL industry. The company will further refine its CXL semiconductor technology and develop additional solutions to speed up AI applications and massive parallel data processing through CXL. "Although many companies have shown interest in CXL, products that effectively employ CXL technology are still hard to find," said Jung. "Panmnesia will not stop at technological development; we will play a pivotal role in expanding the CXL ecosystem." English Articles Link - Article (Reuters) Link - Article (Nasdaq) Link - Article (Yahoo Finance) Link - Article (Design & Reuse) Link - Article (Electronics Weekly) Link - Article (eeNews Europe) Link - Article (Technode Global) Link - Article (Tech in Asia) Link - Article (KrASIA) Link - Article (Bridge Group) Link - Article (Metaverse Post) Link - Article (PitchBook) Link - Article (ETCIO) Korean Articles Link - Article (ChosunBiz) Link - Article (ZDNet) Link - Article (Sedaily) Link - Article (IT BizNews) Link - Article (Digital Today) Link - Article (KOIT) Link - Article (Newswave) Link - Article (CEO Score Daily) Link - Article (Wow Tale) Link - Article (The Stock) Link - Article (Metro) Link - Article (eNewsToday)

Panmnesia's ML-driven Prefetching Technology for CXL-SSD Has Been Selected as KAIST Breakthroughs

Panmnesia’s machine learning-driven prefetching technology for CXL-SSD has been selected as KAIST Breakthroughs this year. KAIST Breakthroughs is a webzine that selectively highlights innovative technologies across various engineering disciplines, encompassing electrical engineering, biotechnology, environmental engineering, and industrial engineering. In this year, KAIST Breakthroughs has highlighted 31 technologies, including Panmnesia’s pioneering work. Panmnesia’s featured technology focuses on enhancing performance of systems that utilize CXL-SSDs. By utilizing the CXL interface, we can use SSD, which is commonly used as a block storage, as a huge memory device. The CXL-SSD harnesses host-side caching advantages while providing tremendous memory capacity. Therefore, it is garnering considerable attention from both industry and academia. However, systems employing CXL-SSDs may encounter long latency of SSD, when cache miss occurs. Panmnesia has addressed this challenge by developing machine learning-based prefetching techniques capable of predicting future memory access information and reading it prior to the actual request. With an in-depth understanding of CXL-based system hierarchies, Panmnesia has crafted a solution that accurately forecasts future access addresses and determines the optimal prefetching timing. If you are intrigued by this innovative work, you can access the original article featured in KAIST Breakthroughs through the following link. KAIST Breakthroughs

Panmnesia on Flash Memory Summit 2023

Panmnesia, the global frontrunner in CXL IP (Intellectual Property) advancement, is proud to unveil its groundbreaking, self-developed multi-terabyte, full-system CXL framework. This exciting development was launched during the 2023 Flash Memory Summit, held August 8-10 at the Santa Clara Convention Center, California. The Summit, a major annual event attracting over 3,000 participants, unites leading companies from around the globe, all specializing in memory and storage solutions. A brief on CXL Interface: The emergence of large-scale applications such as generative AI, Chat-GPT, and recommendation systems has fueled the demand for efficient large-capacity memory technologies and accelerators like GPUs in data centers. Amid this technological shift, the CXL (Compute eXpress Link, CXL) interface technology has gained prominence. CXL allows data centers to drastically expand their memory capacity and enables computing devices such as CPUs and accelerators to utilize this enlarged memory space efficiently, putting it in the industry spotlight. Panmnesia and CXL Technology: As the pioneer of the world's first CXL interface-based full-system framework, Panmnesia stands at the forefront of CXL IP development. Panmnesia design and supply all necessary CXL IPs to implement CXL solutions for data centers. Panmnesia's portfolio includes CXL Core IP and CXL IP for various CXL devices like CPU, NPU, switch, and device. The CXL Core IP supports all CXL devices to facilitate CXL cache communication (CXL.cache), CXL memory communication (CXL.mem), and CXL I/O communication (CXL.io), the primary communication protocols of the CXL interface. Additionally, Panmnesia provides CXL IPs tailored to specific hardware automation functions for each CXL device. Clients of Panmnesia can efficiently develop CXL devices by applying the company’s verified CXL IP to existing devices, thereby reducing additional costs and time associated with IP development and validation. "We're excited to introduce our innovative multi- terabyte full-system CXL framework at this year's Flash Memory Summit," says Dr. Myoungsoo Jung, CEO at Panmnesia. "Our commitment to pioneering cutting-edge solutions is demonstrated through our CXL IP, which we believe will significantly enhance memory and storage capabilities within data centers." At the 2023 Flash Memory Summit, Panmnesia exhibited a multi-terabyte memory pool framework powered by its proprietary CXL IP. Panmnesia also showcased a demonstration video of Meta’s large-scale AI application running within the memory pool framework. The Panmnesia Solution: Panmnesia showcased its multi-terabyte memory pool at the Flash Memory Summit 2023. This pool, consisting of CXL CPU, CXL Switch, and CXL Memory Expander, sets new benchmarks in performance, scalability, adaptability, and efficiency, distinguishing it from other existing memory expansion solutions. Panmnesia’s cutting-edge multi-terabyte memory pool framework encompasses both CXL hardware IP and corresponding software, ensuring high performance and substantial memory capacity expansion. By leveraging multiple CXL switches and CXL memory expanders, this complete solution can seamlessly integrate with existing data center ecosystems. Panmnesia’s innovative CXL Switch technology enables the memory pool framework to efficiently allocate expanded memory space based on user demand. Additionally, by employing a DIMM pooling method for memory module replacement, memory capacity can be increased without procuring extra memory expanders. This makes Panmnesia's memory pool framework a cost-effective memory expansion solution, balancing memory resource management with data center operational cost efficiency. Panmnesia’s CXL Demo - A Real-World Application Success: Panmnesia showcased a demonstration video featuring Meta’s recommendation system, a notable large-scale AI application, running on the implemented memory pool framework. This demonstration showed Panmnesia's solution could reduce the execution time of Meta’s recommendation system by 3.32 times compared to the existing RDMA-based memory expansion solutions used in data centers. Looking forward, Panmnesia is committed to continue pushing the boundaries of CXL technology. Panmnesia aim to further improve data center performance and efficiency through ongoing research, development, and implementation of advanced CXL solutions. The CXL IP and multi-terabyte memory pool framework presented by Panmnesia at the Flash Memory Summit 2023 embody a forward-thinking data center solution. It offers system-side benefits along with operational cost efficiency. Uniquely, Panmnesia is the only company to display the comprehensive use of CXL IP technology from the operating systems to real-world applications, giving them a significant edge over other data center solutions. To view Panmnesia’s CXL memory pool demonstration video again, visit the link below. Link - Demo Video (Youtube) Link - Article (Blocks and Files)

Demo - Panmnesia's Authentic, Multi-Terabyte CXL Memory Disaggregation Framework

All set to explore the horizon of technological advancement at this year’s Flash Memory Summit (FMS)? Make sure you’ve got Panmnesia on your must-see list! This year at FMS 23, we’re beyond excited to present our live demo, entitled “Experience Reality, Not Blueprints: Panmnesia’s Authentic, Multi-Terabyte CXL Memory Disaggregation Framework.” We believe in the power of action over words, which is why we’re bringing you a fully-operational CXL solution. Panmnesia’s groundbreaking CXL system includes all the requisite hardware components, ranging from custom-designed CXL CPUs and switches, to our state-of-the-art DIMM pool devices. Our robust framework enables multi-terabyte capacity by integrating a myriad of memory devices via multi-tiered CXL switches. We can’t wait to showcase how CXL technology can supercharge real-world applications. As a part of our demo at FMS 23, we’ll demonstrate a deep learning recommendation system operating in harmony with LinuxOS and Panmnesia’s innovative CXL framework. Plus, you can gain deeper insights from the other demo videos we’ll be sharing. Intrigued by what real innovation looks like? Make sure to visit us at booth #655 during the Flash Memory Summit, and get up close with the future of technology, brought to life by Panmnesia! Link - Demo Video (Youtube)

Panmnesia will participate in Flash Memory Summit 2023

Panmnesia is excited to announce our participation in the Flash Memory Summit (FMS) 2023, taking place in Santa Clara, California, US, from August 8th to August 10th. This premier event is a global platform for industry leaders and innovative thinkers to showcase and discuss the latest advancements in flash memory technology. At this year’s FMS, Panmnesia will be in the spotlight with our groundbreaking full-system CXL (Compute Express Link) framework. This pioneering infrastructure consists of customized CXL CPUs, switches, and DIMM pools. Our innovation doesn’t stop there, though; we’ve designed the framework to offer multi-terabyte capacity by connecting a plethora of memory devices through multi-level CXL switches. One of the standout features of our system is the incorporation of DIMM pooling technology. By strategically utilizing this technology, we’ve made it possible to achieve substantial cost reductions, all the while maintaining the system’s high performance and reliability. Our team will demonstrate exactly how CXL technology can revolutionize real-world applications. Visitors will be treated to a live and captivating demo of recommendation systems running on a real deep learning framework with Linux OS. This example will provide a tangible glimpse into how our technologies can enhance and streamline systems in various sectors, from cloud computing to high-performance computing (HPC) to data centers and beyond. For everyone interested in learning more about Panmnesia’s innovative technologies, we extend a warm invitation to visit us at booth 655 at the Flash Memory Summit. Our team will be on hand to share insights, answer questions, and explore potential collaborations. We look forward to meeting fellow industry enthusiasts and exploring the future of memory technology together at FMS 2023! Link - Flash Memory Summit 2023