Industry Tradeshow
Open Compute Project (OCP) Global Summit 2024
Join us at OCP 2024!
October 15-17, 2024
San Jose Convention Center
San Jose, California
Booth C26
Empowering the AI Evolution
TE Connectivity (TE) is empowering the artificial intelligence (AI) and machine learning (ML) evolution through a robust portfolio of interconnected solutions engineered for high speed and power efficiency in data center architectures.
The advent of the AI/ML era is driving a major revolution in scale, complexity, and time-to-market of compute architectures at both the data center and at the edge. Interconnect solutions are more critical than ever in providing low-latency GPU-to-GPU clustering and connectivity; memory pooling and disaggregation; and efficient thermal management.
TE's product displays and partnerships at the OCP Global Summit will demonstrate how our our latest innovations around cabled backplane, high-speed cables, and liquid and thermal cooling, are enabling our customers to build the next generation of AI systems.
TE Expert Talks at OCP Global Summit 2024
- Next Gen PCIe Active and Passive Cable Solution for Enhanced Signal
Integrity and Reach: This paper explores the pivotal role of PCIe technology in AI and ML applications, driven by its high bandwidth, low latency, and reliable communication channels. CopprLink cables are crucial for meeting increasing bandwidth demands and maintaining PCIe's competitive edge. The study focuses on optimizing near-chip interconnect designs, tackling challenges such as signal integrity issues from PAM4 modulation, ground resonances, and mechanical reliability. It identifies current-generation signal integrity hurdles and proposes design improvements for future PCIe iterations.
- Connectivity Solutions Empowering the AI Evolution: The advent of the AI/ML era is driving a major revolution in scale, complexity, and time-to-market of compute architectures at both the data center and at the edge. Interconnect solutions are more critical than ever in providing low-latency GPU-to-GPU clustering and connectivity; memory pooling and disaggregation; and efficient thermal management. This session explores trends around GPU clustering and highlights key challenges our customers face in building and scaling up these systems. Our latest innovations around cabled backplane, PCIe connectivity, and liquid cooled busbars are enabling our customers to build the next generation of AI systems.
- Beyond ORv3 HPR Rack Vertical Busbar: Next Gen Rack Busbar Enabling >200kW AI Racks: The AI revolution is driving the need for higher rack power levels. Busbar performance is a key constraint the AI industry is facing in increasing per rack power capacity. Incorporating liquid cooling in busbars enables a 2-4x increase in busbar power carrying capacity compared to existing air cooled solutions.The liquid cooled busbar design utilizes existing liquid cooling rack infrastructure, thus supporting the most demanding AI workloads and latest chip designs without increasing busbar footprint. TE & Meta will demonstrate a vertical busbar design proposal to achieve higher power capacity for next gen 200kW+ OCP AI platforms.
Product Solutions on Display at OCP 2024
224G Architectures Enabling AI & Networking: In this live demo, TE is showcasing a cabled backplane architecture plus OTB and near-chip connectivity, utilizing 224G AdrenaLINE Catapult near-chip connector and AdrenaLINE Slingshot cabled backplane connector (cable-to-cable and cable-to-board). This demo is driven by Marvell's 224 Gbps DSP SerDes silicon.
Liquid Cooled Power Busbar Assembly: In partnership with Cooler Master, TE is enabling the next-gen power busbar solutions helping to empower the AI evolution. As rack level power requirements continue to increase, power busbar capacities need to follow suit. The use of liquid cooled infrastructures to cool the power busbars allows for higher current capacities while maintaining lower temperatures.
Power Distribution for Data Center Rack: Featuring TE’s latest solutions compatible with Open Rack v3 (ORv3) power distribution. TE's solutions for high power distribution support effectiveness, safety, high/low volume production, and reliability. This product showcase will highlight the latest power distribution products, featuring input/output interconnect solutions, integrated busbar assemblies, custom power cable assemblies, quick connect and disconnect high power RAPID LOCK cable assemblies, CROWN CLIP Jr. busbar clip, ELCON Mini connector solutions, high current card edge products, and next generation MULTI-BEAM Plus power connectors.
Advanced High-Speed I/O Liquid Cooling & Thermal Bridge Hybrid Solution: In this demonstration, TE is showcasing our next-generation thermal solution for future high-speed I/O products. By combining TE's patented thermal bridge components with a liquid cooling cold plate structure, TE presents the market with this efficient technology for solving some of the most challenging thermal problems.
Networking, Outside the Box: TE's AOC and DAC breakout cables provide an effective way to provide Switch-NIC connectivity. TE linear drive optics (LRO) AOCs offer interoperability and improved latency in comparison to full DSP optics, which recah optimized for backend AI and HPC applications. TE linear pluggable optics (LPO) provide attractive 112G solutions, helping to overcome the limitations of traditional DSP/CDR implementation by taking advantage of evolving AI and HPC architectures.
Industry Tradeshow
Open Compute Project (OCP) Global Summit 2024
Join us at OCP 2024!
October 15-17, 2024
San Jose Convention Center
San Jose, California
Booth C26
Empowering the AI Evolution
TE Connectivity (TE) is empowering the artificial intelligence (AI) and machine learning (ML) evolution through a robust portfolio of interconnected solutions engineered for high speed and power efficiency in data center architectures.
The advent of the AI/ML era is driving a major revolution in scale, complexity, and time-to-market of compute architectures at both the data center and at the edge. Interconnect solutions are more critical than ever in providing low-latency GPU-to-GPU clustering and connectivity; memory pooling and disaggregation; and efficient thermal management.
TE's product displays and partnerships at the OCP Global Summit will demonstrate how our our latest innovations around cabled backplane, high-speed cables, and liquid and thermal cooling, are enabling our customers to build the next generation of AI systems.
TE Expert Talks at OCP Global Summit 2024
- Next Gen PCIe Active and Passive Cable Solution for Enhanced Signal
Integrity and Reach: This paper explores the pivotal role of PCIe technology in AI and ML applications, driven by its high bandwidth, low latency, and reliable communication channels. CopprLink cables are crucial for meeting increasing bandwidth demands and maintaining PCIe's competitive edge. The study focuses on optimizing near-chip interconnect designs, tackling challenges such as signal integrity issues from PAM4 modulation, ground resonances, and mechanical reliability. It identifies current-generation signal integrity hurdles and proposes design improvements for future PCIe iterations.
- Connectivity Solutions Empowering the AI Evolution: The advent of the AI/ML era is driving a major revolution in scale, complexity, and time-to-market of compute architectures at both the data center and at the edge. Interconnect solutions are more critical than ever in providing low-latency GPU-to-GPU clustering and connectivity; memory pooling and disaggregation; and efficient thermal management. This session explores trends around GPU clustering and highlights key challenges our customers face in building and scaling up these systems. Our latest innovations around cabled backplane, PCIe connectivity, and liquid cooled busbars are enabling our customers to build the next generation of AI systems.
- Beyond ORv3 HPR Rack Vertical Busbar: Next Gen Rack Busbar Enabling >200kW AI Racks: The AI revolution is driving the need for higher rack power levels. Busbar performance is a key constraint the AI industry is facing in increasing per rack power capacity. Incorporating liquid cooling in busbars enables a 2-4x increase in busbar power carrying capacity compared to existing air cooled solutions.The liquid cooled busbar design utilizes existing liquid cooling rack infrastructure, thus supporting the most demanding AI workloads and latest chip designs without increasing busbar footprint. TE & Meta will demonstrate a vertical busbar design proposal to achieve higher power capacity for next gen 200kW+ OCP AI platforms.
Product Solutions on Display at OCP 2024
224G Architectures Enabling AI & Networking: In this live demo, TE is showcasing a cabled backplane architecture plus OTB and near-chip connectivity, utilizing 224G AdrenaLINE Catapult near-chip connector and AdrenaLINE Slingshot cabled backplane connector (cable-to-cable and cable-to-board). This demo is driven by Marvell's 224 Gbps DSP SerDes silicon.
Liquid Cooled Power Busbar Assembly: In partnership with Cooler Master, TE is enabling the next-gen power busbar solutions helping to empower the AI evolution. As rack level power requirements continue to increase, power busbar capacities need to follow suit. The use of liquid cooled infrastructures to cool the power busbars allows for higher current capacities while maintaining lower temperatures.
Power Distribution for Data Center Rack: Featuring TE’s latest solutions compatible with Open Rack v3 (ORv3) power distribution. TE's solutions for high power distribution support effectiveness, safety, high/low volume production, and reliability. This product showcase will highlight the latest power distribution products, featuring input/output interconnect solutions, integrated busbar assemblies, custom power cable assemblies, quick connect and disconnect high power RAPID LOCK cable assemblies, CROWN CLIP Jr. busbar clip, ELCON Mini connector solutions, high current card edge products, and next generation MULTI-BEAM Plus power connectors.
Advanced High-Speed I/O Liquid Cooling & Thermal Bridge Hybrid Solution: In this demonstration, TE is showcasing our next-generation thermal solution for future high-speed I/O products. By combining TE's patented thermal bridge components with a liquid cooling cold plate structure, TE presents the market with this efficient technology for solving some of the most challenging thermal problems.
Networking, Outside the Box: TE's AOC and DAC breakout cables provide an effective way to provide Switch-NIC connectivity. TE linear drive optics (LRO) AOCs offer interoperability and improved latency in comparison to full DSP optics, which recah optimized for backend AI and HPC applications. TE linear pluggable optics (LPO) provide attractive 112G solutions, helping to overcome the limitations of traditional DSP/CDR implementation by taking advantage of evolving AI and HPC architectures.
