MSI, a leading global provider of high-performance server solutions, is excited to showcase its comprehensive lineup of motherboards and servers based on the OCP Modular Hardware System (DC-MHS) architecture at the OCP Global Summit from October 15-17 at booth A6. These cutting-edge solutions represent a breakthrough in server designs, enabling flexible deployments for cloud and high-density data centers. Featured innovations include CXL memory expansion servers and AI-optimized servers, demonstrating MSI's leadership in pushing the boundaries of AI performance and computing power.

DC-MHS Series Motherboards and Servers: Enabling Flexible Deployment in Data Centers
"The rapidly evolving IT landscape requires cloud service providers, large-scale data center operators, and enterprises to handle expanding workloads and future growth with more flexible and powerful infrastructure. MSI's new rage of DC-MHS-based solutions provides the needed flexibility and efficiency for modern data center environments," said Danny Hsu, General Manager of Enterprise Platform Solutions.

JEDEC Solid State Technology Association, the global leader in standards development for the microelectronics industry, today announced upcoming raw card designs currently in development in JEDEC's JC-45 Committee for DRAM Modules in collaboration with the JC-40 and JC-42 Committees. These raw card memory device standards are intended for use in client computing applications such as laptops and desktops and will be supported by related appendix specifications. The forthcoming raw cards will also complement two DDR5 Clock Driver standards published earlier this year: JESD323: DDR5 Clocked Unbuffered Dual Inline Memory Module (CUDIMM) Common Specification and JESD324: DDR5 Clocked Small Outline Dual Inline Memory Module (CSODIMM) Common Specification.

Integrating a Clock Driver (CKD) into a DDR5 DIMM provides numerous advantages, particularly in memory stability and performance, and enhances signal integrity and reliability at high speeds. By regenerating the clock signal locally on the DIMM, a CKD ensures stable operation even at elevated clock speeds. With a DDR5 CKD, DIMM data rates can be increased from 6400 Mbps to 7200 Mbps in the initial version of the standard, and targeting up to 9200 Mbps in future versions.

Micron Technology, Inc., today announced the availability of a brand-new category of clock driver memory with the launch of its Crucial DDR5 clocked unbuffered dual inline memory modules (CUDIMM) and clocked small outline dual memory modules (CSODIMM), which are now shipping in volume. The JEDEC-standard solutions run at speeds up to 6,400 MT/s (megatransfers per second), more than twice as fast as DDR4 and 15% faster than traditional non-clock-driver-based DDR5. Designed to provide more speed stability, faster downloads and better refresh rates, these solutions represent a completely new frontier of memory form factors for next-generation PCs. Micron's CUDIMM and CSODIMM solutions are the industry's first commercially available JEDEC-standard DDR5 CUDIMM and CSODIMM solutions to hit the market since JEDEC standardized the specification earlier this year.

In addition, Intel has validated Micron DDR5 CUDIMM and CSODIMM solutions up to capacities of 64 gigabytes (GB) for use with its Intel Core Ultra processors (Series 2), which were launched last week on Oct. 10.

Rambus Inc., a premier chip and silicon IP provider making data faster and safer, today unveiled industry-first, complete memory interface chipsets for Gen 5 DDR5 RDIMMs and next-generation DDR5 Multiplexed Rank Dual Inline Memory Modules (MRDIMMs). These innovative new products for RDIMMs and MRDIMMs will seamlessly extend DDR5 performance with unparalleled bandwidth and memory capacity for compute-intensive data center and AI workloads.

"The voracious memory demands of AI and HPC require the relentless pursuit of higher performance through continued innovation and technology leadership," said Sean Fan, chief operating officer at Rambus. "With our 30-plus years of renowned high-speed signal integrity and memory system expertise, the Rambus Gen5 RCD, and next-generation MRCD, MDB, and PMIC will be critical enabling chips in future-generation servers leveraging DDR5 RDIMM 8000 and MRDIMM 12800."

AMD has unveiled its latest EPYC processors, codenamed "Turin," featuring Zen 5 and Zen 5C dense cores. Phoronix's thorough testing reveals remarkable advancements in performance, efficiency, and value. The new lineup includes the EPYC 9575F (64-core), EPYC 9755 (128-core), and EPYC 9965 (192-core) models, all showing impressive capabilities across various server and HPC workloads. In benchmarks, a dual-socket configuration of the 128-core EPYC 9755 Turin outperformed Intel's dual Xeon "Granite Rapids" 6980P setup with MRDIMM-8800 by 40% in the geometric mean of all tests. Surprisingly, even a single EPYC 9755 or EPYC 9965 matched the dual Xeon 6980P in expanded tests with regular DDR5-6400. Within AMD's lineup, the EPYC 9755 showed a 1.55x performance increase over its predecessor, the 96-core EPYC 9654 "Genoa". The EPYC 9965 surpassed the dual EPYC 9754 "Bergamo" by 45%.

These gains come with improved efficiency. While power consumption increased moderately, performance improvements resulted in better overall efficiency. For example, the EPYC 9965 used 32% more power than the EPYC 9654 but delivered 1.55x the performance. Power consumption remains competitive: the EPYC 9965 averaged 275 Watts (peak 461 Watts), the EPYC 9755 averaged 324 Watts (peak 500 Watts), while Intel's Xeon 6980P averaged 322 Watts (peak 547 Watts). AMD's pricing strategy adds to the appeal. The 192-core model is priced at $14,813, compared to Intel's 128-core CPU at $17,800. This competitive pricing, combined with superior performance per dollar and watt, has resonated with hyperscalers. Estimates suggest 50-60% of hyperscale deployments now use AMD processors.

As a global leader in memory solutions, Team Group Inc. today unveiled its latest T-Force CKD (Client Clock Driver) DDR5 Gaming Memory. Following the worldwide attention received for the 7200 MHz T-CREATE EXPERT Ai CKD DDR5 showcased at COMPUTEX 2024, the T-Force gaming brand takes performance to new heights with the upcoming T-Force XTREEM CKD DDR5 8200 to 9000 (Gear 2) 2x24GB, featuring overclocking capabilities reaching up to 9600 MHz (Gear 4). Consumers can now push the limits of CUDIMM overclocking on the Intel Z890 platform with next-generation CKD DDR5 overclocking memory, achieving unprecedented performance breakthroughs.

Team Group continues to lead the industry in pushing boundaries by crafting the T-Force XTREEM CKD DDR5, which breaks beyond JEDEC frequency specifications. Utilizing CKD components and Intel XMP profiles on the Intel Z890 motherboard, users can effortlessly overclock to DDR5 speeds beyond 9000 MHz with a single click. Unlike JEDEC-compliant memory modules, the T-Force XTREEM CKD DDR5 leverages CKD components to enhance and buffer high-frequency signals from the CPU, ensuring more stable signal transmission to the memory modules. This enables DDR5 to push overclocking performance to even higher frequencies, surpassing traditional U-DIMM overclocking limitations.

ASRock Rack Inc., a leading innovative server company, announced upgrades to its extensive lineup to support AMD EPYC 9005 Series processors. Among these updates is the introduction of the new 6U8M-TURIN2 GPU server. This advanced platform features AMD Instinct MI325X accelerators, specifically optimized for intensive enterprise AI applications, and will be showcased at AMD Advancing AI 2024.

ASRock Rack Introduce GPU Servers Powered by AMD EPYC 9005 series processors
AMD today revealed the 5th Generation AMD EPYC processors, offering a wide range of core counts (up to 192 cores), frequencies (up to 5 GHz), and expansive cache capacities. Select high-frequency processors, such as the AMD EPYC 9575F, are optimized for use as host CPUs in GPU-enabled systems. Additionally, the just launched AMD Instinct MI325X accelerators feature substantial HBM3E memory and 6 TB/s of memory bandwidth, enabling quick access and efficient handling of large datasets and complex computations.

Asgard joined the ranks of V-Color to unveil its first high-frequency DDR5 memory, the Thor DDR5-9600. V-Color recently launched its DDR5-9200 memory, and Asgard one-upped it with DDR5-9600. Both these are CUDIMMs—UDIMMs that feature a client clock driver (CKD) component. These modules feature the highest bins of DDR5-9600 chips from SK Hynix. The advertised speed of DDR5-9600 is achieved using timings of 44-56-56-136, and a scorching DRAM voltage of 1.50 V. The module includes an XMP 3.0 profile to enable these settings on an Intel platform. With a single-rank configuration and 24 GB module density, the dual-channel kit gives you 48 GB. Asgard revealed that it is working on a DDR5-10000 kit. The company didn't announce US availability.

V-COLOR is set to revolutionize the future of memory technology with the launch of the world's first RGB DDR5 O CUDIMM (High Speed Overclocked CUDIMM). This groundbreaking innovation for the next-generation desktop platform, offering unmatched speeds and efficiency for desktop applications. This memory module has a new heatsink patent (Patent No. 113208127) to increase heat dissipation efficiency to a new level, and the introduction of the exclusive light guide rod patented coating technology (Patent No. M653290) allows the light guide rod to attract attention regardless of whether it has RGB effects or not.

The DDR5 Clocked Unbuffered Dual Inline Memory Module (CUDIMM) technology dynamically adjusts clock frequency and voltage based on workload and system conditions. This cutting-edge capability results in faster data transmission, lower power consumption, and enhanced stability, making it a game-changer in the industry.

SMART Modular Technologies, Inc. ("SMART"), a division of SGH and a global leader in memory solutions, solid-state drives, and advanced memory, has launched a new line of DDR5 Registered DIMMs (RDIMMs) with conformal coating which are specifically designed for use in liquid immersion servers. This innovative product line combines the superior performance of DDR5 technology with enhanced protection, ensuring reliability and longevity in the most demanding data center environments.

Arthur Sainio, DRAM product director for SMART explains the significance of this introduction, "Our new DDR5 RDIMMs with conformal coating represent the perfect fusion of cutting-edge performance and rugged reliability for the next generation of immersion-cooled data centers. By combining the speed and efficiency of DDR5 technology with advanced protective coatings, we're enabling our customers to push the boundaries of computing power while ensuring long-term durability in demanding liquid immersion environments. This product embodies our commitment to innovation and our drive to meet the evolving needs of high-performance computing applications."

Giga Computing, a subsidiary of GIGABYTE and an industry leader in generative AI servers and advanced cooling technologies, today released two GIGABYTE R-series servers (R183-ZK0 and R283-ZK0) with enhanced performance and reliability for cloud services and data-intensive applications. These highly scalable memory capacity servers support AMD EPYC 9004 processors and are ready for select 5th generation AMD EPYC processors with up to 192 CPU cores.

These new GIGABYTE servers are the first and only ones in the market that are one node with two CPUs that support 48 memory DIMMs. GIGABYTE's rich history in motherboard design and engineering with great signal integrity make this possible, a server with 12 TB memory using 256 GB DDR5 3DS RDIMMs. To accommodate a 12-memory channel platform with a 2DPC configuration and without compromising, a new memory layout was developed.

Marvell Technology, Inc., a leader in data infrastructure semiconductor solutions, today launched the Marvell Structera product line of Compute Express Link (CXL) devices that enable cloud data center operators to overcome memory performance and scaling challenges in general-purpose servers.

To address memory-intensive applications, data center operators add extra servers to get higher memory bandwidth and higher memory capacity. The compute capabilities from the added processors are typically not utilized for these applications, making the servers inefficient from cost and power perspectives. The CXL industry standard addresses this challenge by enabling new architectures that can efficiently add memory to general-purpose servers.

Micron Technology, Inc., today announced it is now sampling its multiplexed rank dual inline memory module (MRDIMMs). The MRDIMMs will enable Micron customers to run increasingly demanding workloads and obtain maximum value out of their compute infrastructure. For applications requiring more than 128 GB of memory per DIMM slot, Micron MRDIMMs outperform current TSV RDIMMs by enabling the highest bandwidth, largest capacity with the lowest latency and improved performance per watt to accelerate memory-intensive virtualized multi-tenant, HPC and AI data center workloads.1 The new memory offering is the first generation in the Micron MRDIMM family and will be compatible with Intel Xeon 6 processors.

"Micron's latest innovative main memory solution, MRDIMM, delivers the much-needed bandwidth and capacity at lower latency to scale AI inference and HPC applications on next-generation server platforms," said Praveen Vaidyanathan, vice president and general manager of Micron's Compute Products Group. "MRDIMMs significantly lower the amount of energy used per task while offering the same reliability, availability and serviceability capabilities and interface as RDIMMs, thus providing customers a flexible solution that scales performance. Micron's close industry collaborations ensure seamless integration into existing server infrastructures and smooth transitions to future compute platforms."

ASUS Republic of Gamers today announced the launch of the ROG NUC gaming PC. The first-ever ROG NUC PC is packed with cutting-edge technology and performance for AAA gaming and more.

Gaming, redefined
Featuring Intel Core Ultra 9 or Intel Core Ultra 7 processors and NVIDIA GeForce RTX 4070 or 4060 discrete graphics, the ROG NUC PC is a compact powerhouse that delivers unmatched gaming experiences. The high-performance hybrid architecture of Intel Core Ultra processors provides gamers with the power needed to stream, edit, record and play without skipping a beat. The 2.5-liter chassis is designed for quick system access to make upgrades and cleaning easy. Plus, the compatible ROG Raikiri Pro controller can be used to deliver a console-like experience for gamers seeking uncompromised performance and flexibility in a compact form factor.

SK hynix presented its leading AI memory solutions at COMPUTEX Taipei 2024 from June 4-7. As one of Asia's premier IT shows, COMPUTEX Taipei 2024 welcomed around 1,500 global participants including tech companies, venture capitalists, and accelerators under the theme "Connecting AI". Making its debut at the event, SK hynix underlined its position as a first mover and leading AI memory provider through its lineup of next-generation products.

"Connecting AI" With the Industry's Finest AI Memory Solutions
Themed "Memory, The Power of AI," SK hynix's booth featured its advanced AI server solutions, groundbreaking technologies for on-device AI PCs, and outstanding consumer SSD products. HBM3E, the fifth generation of HBM1, was among the AI server solutions on display. Offering industry-leading data processing speeds of 1.18 terabytes (TB) per second, vast capacity, and advanced heat dissipation capability, HBM3E is optimized to meet the requirements of AI servers and other applications. Another technology which has become crucial for AI servers is CXL as it can increase system bandwidth and processing capacity. SK hynix highlighted the strength of its CXL portfolio by presenting its CXL Memory Module-DDR5 (CMM-DDR5), which significantly expands system bandwidth and capacity compared to systems only equipped with DDR5. Other AI server solutions on display included the server DRAM products DDR5 RDIMM and MCR DIMM. In particular, SK hynix showcased its tall 128-gigabyte (GB) MCR DIMM for the first time at an exhibition.

On June 4th, COMPUTEX 2024 was successfully held at the Taipei Nangang Exhibition Center. Mnemonic Electronic Co., Ltd., the Taiwanese subsidiary of Longsys, showcased industry-leading high-capacity SSDs under the theme "Embracing the Era of High-Capacity SSDs." The products on display included the Mnemonic MS90 8TB SATA SSD, FORESEE ORCA 4836 series enterprise NVMe SSDs, FORESEE XP2300 PCIe Gen 4 SSDs, and rich product lines comprising embedded storage, memory modules, memory cards, and more. The company offers reliable industrial-grade, automotive-grade, and enterprise-grade storage products, providing high-capacity solutions for global users.

High-Capacity SSDs
For SSDs, Mnemonic Electronic presented products in various form factors and interfaces, including PCIe M.2, PCIe BGA, SATA M.2, and SATA 2.5-inch. The Mnemonic MS90 8 TB SATA SSD supports the SATA interface with a speed of up to 6 Gb/s (Gen 3) and is backward compatible with Gen 1 and Gen 2. It also supports various SATA low-power states (Partial/Sleep/Device Sleep) and can be used for nearline HDD replacement, surveillance, and high-speed rail systems.

Kingston at the 2024 Computex showed us their latest gaming and overclocking PC memory products, under its FURY brand. The star attraction here, is the FURY Impact DDR5 CAMM2 module. Kingston is partnering with MSI to bring this standard to the gaming PC space, as part of its new Project Zero ecosystem. A DDR5 CAMM2 is basically a dual-channel (4 sub-channel) DDR5 memory-on-a-stick. The module has wiring for both the channels and all four sub-channels, and a total of two ranks (1R/DPC). The module sits on a land-grid, just like the processor, and is held in place by a series of screws. When installed, a CAMM2 looks a lot more discreet than a bunch of DIMMs that stick out perpendicularly to the plane of the motherboard.

Kingston DDR5 CAMM2 will come in typical dual-channel capacities, such as 32 GB; with typical speeds. The CAMM2 standard was originally designed for mainstream notebooks and mini-PC desktops, but MSI adapted it to an ATX desktop motherboard with its Project Zero backside connectivity. Next up, we have the FURY Renegade RGB DDR5 series, with its high DDR5 speeds, and plenty of RGB bling. Densities range between 16 GB (1x 16 GB) to 96 GB (2x 48 GB), with popular densities along the way, such as 2x 16 GB, and 2x 24 GB. Speeds range between DDR5-6000, all the way till DDR5-8000 (with XMP 3.0).

Biwin is a licensee of SSDs, PC memory, and flash storage products for some for the biggest PC brands out there, including Acer and HP. This year, the company decided to launch a whole product stack under its own brand, so it could sell to the retail channel directly. We also spotted several licensed products under the coveted Acer Predator brand. Let's start our tour with them: the company showed us an Acer Predator Hera memory kit with 48 GB (2x 24 GB), which does an impressive DDR5-8000, at 40-48-48-128, and 1.35 V. The kit includes a DDR5-8000 @ 1.35 V XMP. The module features a mirror-finish metal heat spreader, and an RGB illuminated top. There are also 32 GB (2x 16 GB) kits in the series that go up to DDR5-8200.

Biwin's own first-party brand isn't too far behind the Predator Hera, the company showed us the Biwin (Editor's note: Wookong is local for Asia and will not be part of international branding) DW100 RGB, a high-end memory series, with kit capacities ranging from 32 GB (2x 16 GB) to 64 GB (2x 32 GB), speeds ranging from DDR5-6000 to DDR5-8200, and vDIMM going up to 1.45 V on the top-spec kit. There's also the DX100, which trades a little bit of performance for a more elaborate RGB LED setup. It comes in capacities up to 64 GB (2x 32 GB), and speeds of up to DDR5-8000. The HX100 is the mid-range kit, it lacks any lighting, capacities range up to 64 GB, and speeds of up to DDR5-7200. The timings aren't as tight as the ones on the DX100. Biwin also has a DDR5 LPCAMM2, with capacities of up to 64 GB, and speeds of up to 9600 MT/s. Most of Biwin's DRAM products launch in July 2024.

GeIL booth was packed with all kinds of memories, anyone could find something for their taste, but our eyes gazed in hope at a few (yet) CAMM2 / LPCAMM2 memory samples. GeIL says that those are capable of supporting data transfer rates up to 8533 MT/s with up to 128 GB capacity. Just next to those a myriad of DDR5 with up to 8400 MT/s and RGB features were trying to catch attention. TUF Gaming Alliance, Orion V RGB (with integrated power management on the DIMM), the EVO V series with active dual fan cooling system, or Polaris family to which GeIL feels the need to mention that it fits most CPU coolers without mechanical interference.More pictures follow.

MSI, a leading global server provider, today announced the availability of high-performance server platforms supporting AMD EPYC 4004 Processors for small and medium businesses and regional-hosted IT service providers to deliver essential security capabilities and energy efficiency. "Businesses across all scales are discovering the advantages of advanced computing, connectivity, and analytics capabilities as applications and services become more widespread," said Danny Hsu, General Manager of Enterprise Platform Solutions. "MSI server platforms, supporting AMD EPYC 4004 Processors, empower our customers to implement high-performance computing with cost-effective, ease of deployment, and manageability features. This capability addresses challenges such as system costs, limited IT expertise, and other infrastructure constraints that were previously prohibitive."

"The new AMD EPYC 4004 Series CPUs, along with our strong ecosystem of technology partners, bring enterprise solutions to a traditionally underserved market and ensure that small and medium businesses have access to highly-performant technologies that help them stay competitive," said John Morris, corporate vice president, Enterprise and HPC Business Group, AMD. "The AMD EPYC 4004 CPUs fill an important market gap, providing cost-optimized solutions with enterprise-grade dependability, scalability and security in cost-optimized system configurations that make sense for smaller businesses and dedicated hosters."

Micron Technology, Inc., today announced the availability of Crucial LPCAMM2, the disruptive next-generation laptop memory form factor that features LPDDR5X mobile memory to level up laptop performance for professionals and creators. Consuming up to 58% less active power and with a 64% space savings compared to DDR5 SODIMMs, LPCAMM2 delivers higher bandwidth and dual-channel support with a single module. LPCAMM2 is an ideal high-performance memory solution for handling AI PC and complex workloads and is compatible with the powerful and versatile Lenovo ThinkPad P1 Gen 7 mobile workstations.

"LPCAMM2 is a game-changer for mobile workstation users who want to enjoy the benefits of the latest mobile high performance memory technology without sacrificing superior performance, upgradeability, power efficiency or space," said Jonathan Weech, senior director of product marketing for Micron's Commercial Products Group. "With LPCAMM2, we are delivering a future-proof memory solution, enabling faster speeds and longer battery life to support demanding creative and AI workloads."

Thanks to recent developments in the AI field, and following in the wake of the world's recovery from COVID-19, the transition of factories to partial or full automation proceeds with unstoppable momentum. And the best place to learn about the latest technologies that aim to make this transition as painless as possible is at Automate 2024. This is North America's largest robotics and automation event, and it will be held in Chicago, Illinois from May 6 to 9.

Automate attracts professionals from around the world, and Apacer is no exception. Apacer team will be on hand to discuss the latest technological developments created by our experienced R&D team. Many of these developments were specifically created to reduce the pain points commonly experienced by fully automated facilities. Take CoreSnapshot, for example. This backup and recovery technology can restore a crashed system to full operation in just a few seconds, reducing downtime and associated maintenance costs. Apacer recently updated CoreSnapshot, creating CoreRescue ASR and CoreRescue USR. The name of CoreRescue ASR refers to Auto Self Recovery. This technology will harness AI to learn the system booting process and analyze how long a boot should take. If this average boot time is significantly longer than usual, the system will trigger the self-recovery process and revert to an earlier, uncorrupted version of the drive's content. CoreRescue USR offers similar functionality, except the self-recovery process is triggered by connecting a small USB stick drive.

Micron Technology, Inc. (Nasdaq: MU), today announced it is leading the industry by validating and shipping its high-capacity monolithic 32Gb DRAM die-based 128 GB DDR5 RDIMM memory in speeds up to 5,600 MT/s on all leading server platforms. Powered by Micron's industry-leading 1β (1-beta) technology, the 128 GB DDR5 RDIMM memory delivers more than 45% improved bit density, up to 22% improved energy efficiency and up to 16% lower latency over competitive 3DS through-silicon via (TSV) products.

Micron's collaboration with industry leaders and customers has yielded broad adoption of these new high-performance, large-capacity modules across high-volume server CPUs. These high-speed memory modules were engineered to meet the performance needs of a wide range of mission-critical applications in data centers, including artificial intelligence (AI) and machine learning (ML), high-performance computing (HPC), in-memory databases (IMDBs) and efficient processing for multithreaded, multicore count general compute workloads. Micron's 128 GB DDR5 RDIMM memory will be supported by a robust ecosystem including AMD, Hewlett Packard Enterprise (HPE), Intel, Supermicro, along with many others.

Rambus Inc. (NASDAQ: RMBS), a premier chip and silicon IP provider making data faster and safer, today announced the availability of its new family of state-of-the-art DDR5 server Power Management ICs (PMICs), including an industry-leading extreme current device for high-performance applications. With this new family of server PMICs, Rambus offers module manufacturers a complete DDR5 RDIMM memory interface chipset supporting a broad range of data center use cases.

"Advanced data center workloads like generative AI require the highest bandwidth and capacity server RDIMMs tailored to meet ever-increasing memory needs of a growing data pipeline," said Sean Fan, chief operating officer at Rambus. "With the addition of this new family of server PMICs, we expand our foundational technology and offer customers a comprehensive memory interface chipset that supports multiple DDR5 server platform generations."

SK hynix showcased its next-generation technologies and strengthened key partnerships at the TSMC 2024 Technology Symposium held in Santa Clara, California on April 24. At the event, the company displayed its industry-leading HBM AI memory solutions and highlighted its collaboration with TSMC involving the host's CoWoS advanced packaging technology.

TSMC, a global semiconductor foundry, invites its major partners to this annual conference in the first half of each year so they can share their new products and technologies. Attending the event under the slogan "Memory, the Power of AI," SK hynix received significant attention for presenting the industry's most powerful AI memory solution, HBM3E. The product has recently demonstrated industry-leading performance, achieving input/output (I/O) transfer speed of up to 10 gigabits per second (Gbps) in an AI system during a performance validation evaluation.