servethehome

Author page description

August 5

  • UCIe 3.0 Spec Released with Big Speed Up for Chiplets

    ➀ The UCIe 3.0 specification doubles data transfer rates to 64GT/s for chiplet-based designs, significantly improving bandwidth density and enabling faster interconnects for 2D/2.5D packaging;

    ➁ Introduces new features like continuous transmission protocols, enhanced runtime recalibration, and L2 exit handshake to optimize power efficiency and link reliability;

    ➂ UCIe 3.0 lays the foundation for future chiplet ecosystems, with commercial products expected around 2026-2028 as adoption by major semiconductor companies accelerates.

    ChipletHPCUCIe
  • Broadcom Jericho4 51.2Tbps AI Router Chip Now Shipping with 3.2Tbps HyperPorts

    ➀ Broadcom Jericho4 51.2Tbps AI router chip is now shipping, targeting large AI clusters with 3.2Tbps HyperPorts and 3nm process technology;

    ➁ Features 100Km+ RoCE deployment for long-distance AI cluster communication and 200G PAM4 signaling;

    ➂ Part of Broadcom's expanded switch/routing portfolio, including Tomahawk 6 and Tomahawk Ultra, for high-tier AI network infrastructure.

    AIBroadcomNetworking
  • Storage.AI Project by SNIA Looks to Re-frame AI Storage Discussion

    ➀ SNIA launched the Storage.AI project to optimize AI storage efficiency by integrating storage with high-performance networks like InfiniBand and Ethernet;

    ➁ The initiative proposes shifting storage to scale-out networks to enhance GPU utilization and reduce pre-processing bottlenecks;

    ➂ Key industry players including AMD, Dell, and Intel have joined, but NVIDIA's absence raises questions about future collaboration.

    AIstorage

August 4

  • Dell PowerEdge R6715 Review A Spiffy 1U AMD EPYC Server

    ➀ The Dell PowerEdge R6715 is a 1U server with direct liquid cooling for AMD EPYC processors, emphasizing efficient thermal management for high-density computing;

    ➁ It features flexible configurations including multiple drive bays, dual OCP NIC 3.0 slots, and PCIe Gen5 expansion options;

    ➂ The system is designed for modern data centers with standardized rack liquid cooling integration and redundant Titanium-level power supplies.

    AMDCoolingDell

August 2

  • Deploying AMD Instead of Arm in our Infrastructure 2025 Here is Why

    ➀ STH explains its choice of AMD EPYC over Arm in 2025 infrastructure, citing Arm's challenges in enterprise adoption, including installed base compatibility, hardware limitations, and insufficient software support;

    ➁ Power efficiency gains from Arm are now minimal compared to AMD/Intel x86 CPUs, especially with AI GPU servers dominating power consumption (e.g., AMD Instinct MI325X at 1kW/accelerator);

    ➂ Enterprise Arm adoption faces hardware scarcity and vendor neglect: NVIDIA Grace CPUs are outdated for general workloads, while major vendors like Dell/HPE offer limited Arm options (e.g., HPE ProLiant RL300 Gen11's market failure).

    AMDArmNVIDIA

August 1

  • Seagate Expansion 28TB External Hard Drive and the HAMR HDD Within Mini-Review

    ➀ Seagate's 28TB Expansion external hard drive integrates HAMR (Heat-Assisted Magnetic Recording) technology, offering high-capacity storage in a compact design with USB 3.2 and external power support;

    ➁ Opening the enclosure reveals a Seagate Barracuda 28TB HAMR drive, typically not sold separately and possibly rebadged from enterprise-grade Exos or IronWolf Pro HDDs;

    ➂ Shucking the external drive provides significant cost savings ($240 less than standalone 28TB HDDs), though sacrificing warranty coverage.

    HDD

July 30

  • Enfabrica Elastic Memory Fabric System aka EMFASYS Launched

    ➀ Enfabrica launched the Elastic Memory Fabric System (EMFASYS), integrating its 3.2Tbps ACF-S SuperNIC with CXL memory to create a high-bandwidth RDMA solution for AI clusters;

    ➁ The system consolidates networking (RDMA) and memory expansion (CXL) into a single chip architecture, eliminating multi-chip complexity and enabling memory pooling over 144 CXL lanes;

    ➂ EMFASYS addresses AI memory scaling limitations by allowing flexible, large-scale memory access beyond traditional HBM/DRAM, with compatibility for devices like Astera Labs CXL modules.

    AICXLHPC

July 29

  • Micron 9650 PCIe Gen6 SSD Announced with Micron 6600 ION 122TB and 7650 SSDs

    ➀ Micron announced three new SSDs: the 9650 PCIe Gen6 (up to 30.72TB) for AI servers with 28GB/s reads and liquid cooling support;

    ➁ The 6600 ION offers massive 122TB capacity in E3.S form factor, targeting AI data storage;

    ➂ The mainstream 7650 PCIe Gen5 SSD focuses on low latency and capacities down to 1.92TB, replacing smaller Gen5 drives.

    MicronSSD
  • NVIDIA Starts to Tackle GPU Power Smoothing with the NVIDIA GB300 NVL72

    ➀ NVIDIA addresses GPU power fluctuation challenges in large AI clusters using the GB300 NVL72, targeting peak-to-valley transitions during AI training workloads;

    ➁ Implements GPU Burn and enhanced power supply capacitance to flatten power demand, reducing strain on power infrastructure like diesel generators;

    ➂ Highlights industry trends toward integrating battery systems in AI data centers to stabilize clusters with extreme GPU spike loads.

    AIGPUNVIDIA

July 28

  • HPE OCP NIC 3.0 Intel i350-T4 Quick Look

    ➀ The article provides a quick overview of the HPE OCP NIC 3.0 featuring the Intel i350-T4 controller, a quad-port 1GbE adapter commonly used in basic server configurations;

    ➁ The NIC follows the OCP NIC 3.0 design with an internal latch mechanism, which complicates maintenance due to the lack of a rear ejector or pull tab;

    ➂ Despite its widespread OS support and legacy relevance, the Intel i350-T4's PCIe Gen2 x4 interface and 1GbE speeds are considered outdated compared to modern high-speed solutions like 400GbE adapters.

    HPEIntelNetworking

July 27

  • QCT QuantaEdge EGX88D-1U Intel Xeon 6 SoC Xeon D Server Shown

    ➀ QCT unveiled the QuantaEdge EGX88D-1U 1U server powered by Intel Xeon 6 SoC (Xeon D), offering up to 24x25GbE ports for high-density networking;

    ➁ The server supports future Xeon 6 SoC CPUs up to 325W with 8-channel memory and includes specialized components like an OCXO oscillator and GNSS receiver for timing precision;

    ➂ Designed for hyperscalers and telcos, it highlights Intel's focus on integrating accelerators for Open RAN and edge applications.

    HPCcpuserver

July 26

  • Intel Looking to Spin and Sell its Networking and Edge Business

    ➀ X-Epic and Agile Analog announced a strategic collaboration to advance EUV lithography technology for semiconductor manufacturing;

    ➁ The partnership aims to optimize analog IP designs for EUV processes, enhancing performance and efficiency in advanced nodes;

    ➂ The collaboration will accelerate development of EDA tools tailored for 3nm and below process technologies, benefiting AI and HPC applications.

    Agile AnalogEUV

July 25

  • HPE ProLiant MicroServer Gen11 Review Great New Mini Server

    ➀ The HPE ProLiant MicroServer Gen11 features an upgraded design with four 3.5" SATA bays and tool-less drive installation;

    ➀ It includes two PCIe slots for expansion, enhanced cooling with a single large fan, and Broadcom 1GbE quad-port NICs;

    ➀ HPE retains iLO remote management and introduces a DC power retention clip for improved stability.

    BroadcomHPEServer Systems

July 24

  • OCP DC-MHS Demo at HPE Discover 2025

    ➀ HPE showcased an OCP DC-MHS demo at HPE Discover 2025, highlighting a modular server design to reduce costs through standardized I/O and layouts;

    ➁ The demo featured a single-socket Intel Xeon motherboard with OCP NIC 3.0 slots, DC-SCM modules for system control, and flexible storage/RAID configurations via MCIO connectivity;

    ➂ The modular approach emphasizes industry standardization, enabling adaptable server designs for diverse form factors (e.g., 1U/2U) and aligning with trends like NVIDIA MGX in AI hardware.

    HPEserver

July 23

  • Inside the NVIDIA Cedar Module with 1.6Tbps of Networking Capacity

    ➀ NVIDIA's Cedar Module integrates four ConnectX-7 NICs to deliver 1.6Tbps networking capacity, designed as a compact alternative to PCIe-based solutions;

    ➀ The custom horizontal form factor enables optimized cooling solutions and high-density deployment in DGX H100/H200 systems through optical cage connections;

    ➂ The module features specialized connectors similar to NVIDIA SXM GPUs, with future systems likely transitioning to ConnectX-8 integration.

    NVIDIANetworking

July 22

  • Kioxia LC9 SSD Hits 245.76TB of Capacity in a Single Drive

    ➀ Kioxia launches the LC9 SSD with a groundbreaking 245.76TB capacity in a 2.5-inch drive, utilizing QLC NAND and 8th-gen BiCS FLASH technology;

    ➁ Achieves high-density storage through 32-layer stacking of 2Tb NAND dies, offering sequential performance optimized for data center applications;

    ➂ Challenges HDD dominance by delivering ~8x the capacity of Seagate's 30TB 3.5-inch HDDs in a single drive, signaling rapid SSD capacity scaling trends.

    KioxiaSSDstorage

July 21

  • ASRock Rack Shows off Intel Xeon 6 SoC Platforms

    ➀ ASRock Rack showcased its Intel Xeon 6 SoC-based platforms at Computex 2025, featuring next-generation Xeon D processors;

    ➁ The GNRDD8HM-2O motherboard supports 100GbE, E1.S storage, and a 2U4N-EGB chassis for high-density 4-node configurations;

    ➂ While the design offers unique cooling and expandability, the higher cost of Xeon 6 SoC may limit adoption compared to traditional Xeon D solutions.

    ASRock RackIntelServer Systems

July 19

  • Rapidus Starts 2nm Gate All Around Prototype Production at IIM-1

    ➀ Rapidus announced it has commenced prototype production of 2nm Gate-All-Around (GAA) transistors at its new IMM-1 fab in Hokkaido, Japan, marking a significant advancement in semiconductor manufacturing;

    ➁ The facility achieved milestones including EUV tool installation in December 2024 and completed over 200 equipment installations, with volume production targeted for 2027;

    ➂ The earthquake-resistant design features specialized steel dampers and sliding column structures to ensure operational stability in seismically active regions.

    2nmRapidussemiconductor
  • STH Hardware Upgrades 2025 Edition

    ➀ STH plans to replace all remaining HDDs in production clusters with SSDs for higher reliability, accepting 3-4x higher costs;

    ➁ Deployment of Ampere Altra Max platforms faced unexpected challenges, offering insights for ARM-based server adoption;

    ➂ Transitioning from Google Drive to integrated storage solutions to handle growing 4K video production demands and reduce cost inefficiencies;

    ➃ Upgrading network testing capabilities using Supermicro Hyper SuperServer SYS-222HA-TN for advanced traffic generation up to 800GbE.

    SSDSupermicro

July 17

  • Phison Pascari X200P 7.68TB PCIe Gen5 NVMe SSD Review

    ➀ The Phison Pascari X200P 7.68TB PCIe Gen5 NVMe SSD offers a balance of high capacity and performance in a 2.5" U.2 form factor, targeting enterprise storage solutions;

    ➀ The drive features a unique heatsink fin design for efficient thermal management and delivers up to 30.72TB capacity in its product line, with the 7.68TB variant showcasing the best performance;

    ➂ Performance metrics vary significantly across different capacities, emphasizing the importance of checking specifications for targeted use cases following the X200P-X200E comparison.

    PCIe Gen5SSD

July 15

  • Broadcom Tomahawk Ultra Launch for Scale-up Ethernet

    ➀ Broadcom launched the Tomahawk Ultra, a 51.2Tbps Ethernet switch chip optimized for small packet sizes and ultra-low latency (250ns), targeting HPC and AI clusters;

    ➁ The chip challenges NVIDIA's NVLink and InfiniBand by offering 'Scale-Up Ethernet' (SUE) with features like in-network collectives, compressed headers, and lossless fabric technology;

    ➂ Pin-compatible with Tomahawk 5 and shipping alongside Tomahawk 6, it emphasizes scalability for AI workloads while maintaining Ethernet compliance.

    BroadcomHPCNVIDIA
  • Gigabyte G893-ZX1-AAX2 AMD Instinct MI325X Server Review

    ➀ Reviewed the Gigabyte G893-ZX1-AAX2 server with eight AMD Instinct MI325X GPUs and dual AMD EPYC CPUs, offering 2TB of HBM3E memory;

    ➁ Features air-cooled design for versatile deployment in AI and HPC workloads;

    ➂ Includes extensive connectivity options: PCIe Gen5 slots, NVMe storage bays, twelve 3kW power supplies, and management NICs.

    AIAMDGigabyte

July 14

  • Beelink Mate Mini Add Dual M.2 Storage to an Apple Mac Mini

    ➀ The Beelink Mate Mini is an external Thunderbolt 5 dock designed to add dual M.2 SSD storage and expand connectivity for Apple Mac Mini, offering a cost-effective alternative to Apple's internal storage upgrades.

    ➁ It features two M.2 slots, a 2.5GbE port, SD card reader, USB-A ports, and a headphone jack, with a design matching the Mac Mini's footprint for stacking compatibility.

    ➂ While the dock provides flexible mounting options (top or bottom), its placement may obstruct the Mac Mini's power button, requiring trade-offs between aesthetics and functionality.

    AppleBeelinkSSD

July 13

  • Cadence IP for LPDDR6 Launched

    ➀ Cadence launched its LPDDR6 IP with 14.4Gbps performance, enhancing future mobile devices and AI accelerators;

    ➁ The LPDDR6 memory subsystem supports AI chip designs combining HBM and LPDDR for bandwidth-sensitive workloads;

    ➂ The IP release aligns with industry trends, enabling next-gen chiplet integration for high-performance computing platforms.

    EDAHBM

July 12

  • GMKtec EVO-X2 Review an AMD Ryzen AI Max 395 Powerhouse

    ➀ The GMKtec EVO-X2 is a compact AI mini PC featuring AMD's Ryzen AI Max+ 395 APU with 16 CPU cores, a 40-core RDNA 3.5 GPU, and 128GB LPDDR5X memory;

    ➁ It offers extensive connectivity options, including dual USB4 ports, 2.5GbE, HDMI 2.1, and DisplayPort, with a vertical/horizontal hybrid design;

    ➂ The review highlights its cooling system, performance modes, and positioning as a high-performance yet portable solution for AI and compute-intensive tasks.

    AIAMDcpu

July 11

  • The ASUS Ascent GX10 Super NVIDIA GB10 System Has Changed

    ➀ The ASUS Ascent GX10 system underwent a chassis redesign, replacing its original clear plastic case to address thermal challenges observed in March 2025;

    ➁ The NVIDIA GB10 platform, targeting consumer markets, integrates 128GB memory and ConnectX networking, enabling cost-effective clustering solutions;

    ➂ OEM differentiation in this segment focuses on chassis aesthetics and support, as hardware components remain standardized across partners.

    AsusHPCNVIDIA

July 9

  • IBM Power11 Launched with Up To 2048 Threads and DDIMM Support

    ➀ IBM launched Power11 servers featuring up to 2048 threads in a 16-socket configuration and DDIMM memory technology for enhanced density;

    ➁ The platform emphasizes AI capabilities with IBM Watsonx Code Assistant and custom Spyre AI accelerators, extending beyond Z-series mainframes;

    ➂ Power11 introduces hybrid cloud support via PowerVS and reliability features like Spare Cores, targeting existing enterprise customers with upgraded hardware and software ecosystems.

    AIibm

July 8

  • This is the New Double Wide NIC with 8x 25GbE Ports Using the Intel E830 NIC

    ➀ Intel introduced a new double-wide OCP NIC 3.0 DSFF card with 8x 25GbE ports, delivering 200Gbps bandwidth via the E830 controller;

    ➁ The card uses a unique dual small form factor (DSFF) design with a single PCIe Gen4 x16/Gen5 x8 interface, enabling high-density networking in compact edge servers;

    ➂ Demonstrated at HPE Discover 2025, it targets edge computing scenarios where dual cards could provide 16x 25GbE ports in 1U servers for enhanced connectivity.

    25GbEHPEIntel

July 7

  • Kyocera Pegatron and SMART Show CXL Over Optics at Computex 2025

    ➀ Kyocera Pegatron and SMART showcased a prototype demonstrating CXL memory accessed remotely via optical connections at Computex 2025;

    ➁ The setup utilized SMART's CXL memory module, Kyocera's PCIe-to-optical converter, and MTP/MPO fiber cables, enabling dynamic memory allocation in distributed systems;

    ➂ The technology highlights future applications in AI clusters and aligns with upcoming PCIe Gen6 advancements, signaling progress toward optical co-packaging solutions.

    CXL

July 5

  • New Intel E610 NICs Shown for Low Power 10Gbase-T and 2.5GbE

    ➀ Intel showcased its new E610 series NICs at HPE Discover 2025, including a dual-port 10Gbase-T model and a quad-port 2.5GbE variant;

    ➁ The dual 10Gbase-T PCIe Gen4 x4 NIC consumes only 5.1W at full load, supporting 10GbE/5GbE/2.5GbE/1GbE speeds;

    ➂ The quad 2.5GbE OCP NIC 3.0 model offers up to 8.6W power efficiency, positioning it as an upgrade path for legacy Intel i350 users.

    2.5GbEIntel