LogoLogo
Continuum WebsiteContinuum ApplicationsContinuum KnowledgeAxolotl Platform
Continuum Knowledge
Continuum Knowledge
  • Continuum
  • Data
    • Datasets
      • Pre Training Data
      • Types of Fine Tuning
      • Self Instruct Paper
      • Self-Alignment with Instruction Backtranslation
      • Systematic Evaluation of Instruction-Tuned Large Language Models on Open Datasets
      • Instruction Tuning
      • Instruction Fine Tuning - Alpagasus
      • Less is More For Alignment
      • Enhanced Supervised Fine Tuning
      • Visualising Data using t-SNE
      • UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction
      • Training and Evaluation Datasets
      • What is perplexity?
  • MODELS
    • Foundation Models
      • The leaderboard
      • Foundation Models
      • LLama 2 - Analysis
      • Analysis of Llama 3
      • Llama 3.1 series
      • Google Gemini 1.5
      • Platypus: Quick, Cheap, and Powerful Refinement of LLMs
      • Mixtral of Experts
      • Mixture-of-Agents (MoA)
      • Phi 1.5
        • Refining the Art of AI Training: A Deep Dive into Phi 1.5's Innovative Approach
      • Phi 2.0
      • Phi-3 Technical Report
  • Training
    • The Fine Tuning Process
      • Why fine tune?
        • Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?
        • Explanations in Fine Tuning
      • Tokenization
        • Tokenization Is More Than Compression
        • Tokenization - SentencePiece
        • Tokenization explore
        • Tokenizer Choice For LLM Training: Negligible or Crucial?
        • Getting the most out of your tokenizer for pre-training and domain adaptation
        • TokenMonster
      • Parameter Efficient Fine Tuning
        • P-Tuning
          • The Power of Scale for Parameter-Efficient Prompt Tuning
        • Prefix-Tuning: Optimizing Continuous Prompts for Generation
        • Harnessing the Power of PEFT: A Smarter Approach to Fine-tuning Pre-trained Models
        • What is Low-Rank Adaptation (LoRA) - explained by the inventor
        • Low Rank Adaptation (Lora)
        • Practical Tips for Fine-tuning LMs Using LoRA (Low-Rank Adaptation)
        • QLORA: Efficient Finetuning of Quantized LLMs
        • Bits and Bytes
        • The Magic behind Qlora
        • Practical Guide to LoRA: Tips and Tricks for Effective Model Adaptation
        • The quantization constant
        • QLORA: Efficient Finetuning of Quantized Language Models
        • QLORA and Fine-Tuning of Quantized Language Models (LMs)
        • ReLoRA: High-Rank Training Through Low-Rank Updates
        • SLoRA: Federated Parameter Efficient Fine-Tuning of Language Models
        • GaLora: Memory-Efficient LLM Training by Gradient Low-Rank Projection
      • Hyperparameters
        • Batch Size
        • Padding Tokens
        • Mixed precision training
        • FP8 Formats for Deep Learning
        • Floating Point Numbers
        • Batch Size and Model loss
        • Batch Normalisation
        • Rethinking Learning Rate Tuning in the Era of Language Models
        • Sample Packing
        • Gradient accumulation
        • A process for choosing the learning rate
        • Learning Rate Scheduler
        • Checkpoints
        • A Survey on Efficient Training of Transformers
        • Sequence Length Warmup
        • Understanding Training vs. Evaluation Data Splits
        • Cross-entropy loss
        • Weight Decay
        • Optimiser
        • Caching
      • Training Processes
        • Extending the context window
        • PyTorch Fully Sharded Data Parallel (FSDP)
        • Train Short, Test Long: Attention with Linear Biases Enables Input Length Extrapolation
        • YaRN: Efficient Context Window Extension of Large Language Models
        • Sliding Window Attention
        • LongRoPE
        • Reinforcement Learning
        • An introduction to reinforcement learning
        • Reinforcement Learning from Human Feedback (RLHF)
        • Direct Preference Optimization: Your Language Model is Secretly a Reward Model
  • INFERENCE
    • Why is inference important?
      • Grouped Query Attention
      • Key Value Cache
      • Flash Attention
      • Flash Attention 2
      • StreamingLLM
      • Paged Attention and vLLM
      • TensorRT-LLM
      • Torchscript
      • NVIDIA L40S GPU
      • Triton Inference Server - Introduction
      • Triton Inference Server
      • FiDO: Fusion-in-Decoder optimised for stronger performance and faster inference
      • Is PUE a useful measure of data centre performance?
      • SLORA
  • KNOWLEDGE
    • Vector Databases
      • A Comprehensive Survey on Vector Databases
      • Vector database management systems: Fundamental concepts, use-cases, and current challenges
      • Using the Output Embedding to Improve Language Models
      • Decoding Sentence-BERT
      • ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT
      • SimCSE: Simple Contrastive Learning of Sentence Embeddings
      • Questions Are All You Need to Train a Dense Passage Retriever
      • Improving Text Embeddings with Large Language Models
      • Massive Text Embedding Benchmark
      • RocketQAv2: A Joint Training Method for Dense Passage Retrieval and Passage Re-ranking
      • LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders
      • Embedding and Fine-Tuning in Neural Language Models
      • Embedding Model Construction
      • Demystifying Embedding Spaces using Large Language Models
      • Fine-Tuning Llama for Multi-Stage Text Retrieval
      • Large Language Model Based Text Augmentation Enhanced Personality Detection Model
      • One Embedder, Any Task: Instruction-Finetuned Text Embeddings
      • Vector Databases are not the only solution
      • Knowledge Graphs
        • Harnessing Knowledge Graphs to Elevate AI: A Technical Exploration
        • Unifying Large Language Models and Knowledge Graphs: A Roadmap
      • Approximate Nearest Neighbor (ANN)
      • High Dimensional Data
      • Principal Component Analysis (PCA)
      • Vector Similarity Search - HNSW
      • FAISS (Facebook AI Similarity Search)
      • Unsupervised Dense Retrievers
    • Retrieval Augmented Generation
      • Retrieval-Augmented Generation for Large Language Models: A Survey
      • Fine-Tuning or Retrieval?
      • Revolutionising Information Retrieval: The Power of RAG in Language Models
      • A Survey on Retrieval-Augmented Text Generation
      • REALM: Retrieval-Augmented Language Model Pre-Training
      • Retrieve Anything To Augment Large Language Models
      • Generate Rather Than Retrieve: Large Language Models Are Strong Context Generators
      • Active Retrieval Augmented Generation
      • DSPy: LM Assertions: Enhancing Language Model Pipelines with Computational Constraints
      • DSPy: Compiling Declarative Language Model Calls
      • DSPy: In-Context Learning for Extreme Multi-Label Classification
      • Optimizing Instructions and Demonstrations for Multi-Stage Language Model Programs
      • HYDE: Revolutionising Search with Hypothetical Document Embeddings
      • Enhancing Recommender Systems with Large Language Model Reasoning Graphs
      • Retrieval Augmented Generation (RAG) versus fine tuning
      • RAFT: Adapting Language Model to Domain Specific RAG
      • Summarisation Methods and RAG
      • Lessons Learned on LLM RAG Solutions
      • Stanford: Retrieval Augmented Language Models
      • Overview of RAG Approaches with Vector Databases
      • Mastering Chunking in Retrieval-Augmented Generation (RAG) Systems
    • Semantic Routing
    • Resource Description Framework (RDF)
  • AGENTS
    • What is agency?
      • Rephrase and Respond: Let Large Language Models Ask Better Questions for Themselves
      • Types of Agents
      • The risk of AI agency
      • Understanding Personality in Large Language Models: A New Frontier in AI Psychology
      • AI Agents - Reasoning, Planning, and Tool Calling
      • Personality and Brand
      • Agent Interaction via APIs
      • Bridging Minds and Machines: The Legacy of Newell, Shaw, and Simon
      • A Survey on Language Model based Autonomous Agents
      • Large Language Models as Agents
      • AI Reasoning: A Deep Dive into Chain-of-Thought Prompting
      • Enhancing AI Reasoning with Self-Taught Reasoner (STaR)
      • Exploring the Frontier of AI: The "Tree of Thoughts" Framework
      • Toolformer: Revolutionising Language Models with API Integration - An Analysis
      • TaskMatrix.AI: Bridging Foundational AI Models with Specialised Systems for Enhanced Task Completion
      • Unleashing the Power of LLMs in API Integration: The Rise of Gorilla
      • Andrew Ng's presentation on AI agents
      • Making AI accessible with Andrej Karpathy and Stephanie Zhan
  • Regulation and Ethics
    • Regulation and Ethics
      • Privacy
      • Detecting AI Generated content
      • Navigating the IP Maze in AI: The Convergence of Blockchain, Web 3.0, and LLMs
      • Adverse Reactions to generative AI
      • Navigating the Ethical Minefield: The Challenge of Security in Large Language Models
      • Navigating the Uncharted Waters: The Risks of Autonomous AI in Military Decision-Making
  • DISRUPTION
    • Data Architecture
      • What is a data pipeline?
      • What is Reverse ETL?
      • Unstructured Data and Generatve AI
      • Resource Description Framework (RDF)
      • Integrating generative AI with the Semantic Web
    • Search
      • BM25 - Search Engine Ranking Function
      • BERT as a reranking engine
      • BERT and Google
      • Generative Engine Optimisation (GEO)
      • Billion-scale similarity search with GPUs
      • FOLLOWIR: Evaluating and Teaching Information Retrieval Models to Follow Instructions
      • Neural Collaborative Filtering
      • Federated Neural Collaborative Filtering
      • Latent Space versus Embedding Space
      • Improving Text Embeddings with Large Language Models
    • Recommendation Engines
      • On Interpretation and Measurement of Soft Attributes for Recommendation
      • A Survey on Large Language Models for Recommendation
      • Model driven recommendation systems
      • Recommender AI Agent: Integrating Large Language Models for Interactive Recommendations
      • Foundation Models for Recommender Systems
      • Exploring the Impact of Large Language Models on Recommender Systems: An Extensive Review
      • AI driven recommendations - harming autonomy?
    • Logging
      • A Taxonomy of Anomalies in Log Data
      • Deeplog
      • LogBERT: Log Anomaly Detection via BERT
      • Experience Report: Deep Learning-based System Log Analysis for Anomaly Detection
      • Log-based Anomaly Detection with Deep Learning: How Far Are We?
      • Deep Learning for Anomaly Detection in Log Data: A Survey
      • LogGPT
      • Adaptive Semantic Gate Networks (ASGNet) for log-based anomaly diagnosis
  • Infrastructure
    • The modern data centre
      • Enhancing Data Centre Efficiency: Strategies to Improve PUE
      • TCO of NVIDIA GPUs and falling barriers to entry
      • Maximising GPU Utilisation with Kubernetes and NVIDIA GPU Operator
      • Data Centres
      • Liquid Cooling
    • Servers and Chips
      • The NVIDIA H100 GPU
      • NVIDIA H100 NVL
      • Lambda Hyperplane 8-H100
      • NVIDIA DGX Servers
      • NVIDIA DGX-2
      • NVIDIA DGX H-100 System
      • NVLink Switch
      • Tensor Cores
      • NVIDIA Grace Hopper Superchip
      • NVIDIA Grace CPU Superchip
      • NVIDIA GB200 NVL72
      • Hopper versus Blackwell
      • HGX: High-Performance GPU Platforms
      • ARM Chips
      • ARM versus x86
      • RISC versus CISC
      • Introduction to RISC-V
    • Networking and Connectivity
      • Infiniband versus Ethernet
      • NVIDIA Quantum InfiniBand
      • PCIe (Peripheral Component Interconnect Express)
      • NVIDIA ConnectX InfiniBand adapters
      • NVMe (Non-Volatile Memory Express)
      • NVMe over Fabrics (NVMe-oF)
      • NVIDIA Spectrum-X
      • NVIDIA GPUDirect
      • Evaluating Modern GPU Interconnect
      • Scalable Hierarchical Aggregation and Reduction Protocol (SHARP)
      • Next-generation networking in AI environments
      • NVIDIA Collective Communications Library (NCCL)
    • Data and Memory
      • NVIDIA BlueField Data Processing Units (DPUs)
      • Remote Direct Memory Access (RDMA)
      • High Bandwidth Memory (HBM3)
      • Flash Memory
      • Model Requirements
      • Calculating GPU memory for serving LLMs
      • Transformer training costs
      • GPU Performance Optimisation
    • Libraries and Complements
      • NVIDIA Base Command
      • NVIDIA AI Enterprise
      • CUDA - NVIDIA GTC 2024 presentation
      • RAPIDs
      • RAFT
    • Vast Data Platform
      • Vast Datastore
      • Vast Database
      • Vast Data Engine
      • DASE (Disaggregated and Shared Everything)
      • Dremio and VAST Data
    • Storage
      • WEKA: A High-Performance Storage Solution for AI Workloads
      • Introduction to NVIDIA GPUDirect Storage (GDS)
        • GDS cuFile API
      • NVIDIA Magnum IO GPUDirect Storage (GDS)
      • Vectors in Memory
Powered by GitBook
LogoLogo

Continuum - Accelerated Artificial Intelligence

  • Continuum Website
  • Axolotl Platform

Copyright Continuum Labs - 2023

On this page
  • Liquid Cooling
  • Cost of Liquid Cooling
  • Power Conservation
  • Increasing Density
  • QCT's L04V air-assist liquid cooling solution

Was this helpful?

  1. Infrastructure
  2. The modern data centre

Liquid Cooling

PreviousData CentresNextServers and Chips

Last updated 11 months ago

Was this helpful?

Liquid Cooling

Here are the key points related to the cost of liquid cooling, power conservation, and the implications of increasing density in data centres:

Cost of Liquid Cooling

Implementing liquid cooling in existing data centres is a significant challenge and can be costly, as it requires modifications to the infrastructure, such as connecting to existing pipework, ensuring leak detection and containment, and integrating with the building management system (BMS).

The cost of liquid cooling depends on the specific technology used, such as immersion cooling or cold plate cooling, and the scale of implementation.

Retrofitting liquid cooling into legacy data centres may involve additional operational costs and risks, such as the need for specialised expertise and the potential for disruptions to existing customers.

Designing a data centre from the ground up to accommodate liquid cooling can be more cost-effective than retrofitting an existing facility, as it allows for proper planning and infrastructure integration.

The cost of liquid cooling also depends on the level of redundancy and availability required. Implementing features like leak detection, containment, and separate cooling loops for each customer can add significant costs to the system.

Standardisation and collaboration among industry players could help reduce the costs of liquid cooling by promoting common practices, interfaces, and equipment compatibility.

Power Conservation

  1. Liquid cooling can potentially reduce the power consumption associated with cooling in data centres, as it allows for more efficient heat removal compared to traditional air cooling methods.

  2. By using liquid cooling, data centres can operate at higher temperatures, which reduces the need for energy-intensive chillers. In some cases, chillers may be eliminated altogether, replaced by dry coolers or other more efficient cooling methods.

  3. The power savings from liquid cooling can be significant, with estimates ranging from a 10-15% reduction in overall data centre power consumption, as cooling can account for 20-25% of total energy usage.

  4. However, the power savings from liquid cooling may be relatively small compared to the increasing power demands of high-density IT equipment, such as AI accelerators and high-performance computing systems.

  5. To maximise power conservation, data centres can explore techniques like heat reuse, where the waste heat from liquid-cooled systems is captured and used for other purposes, such as heating nearby buildings or industrial processes.

Increasing Density

  1. The demand for higher-density computing, driven by AI and other advanced workloads, is rapidly increasing, with rack power densities reaching 55-120 kW or more.

  2. Increasing density poses significant challenges for data centre cooling, as traditional air cooling methods struggle to keep up with the heat generated by high-density racks.

  3. Liquid cooling is seen as a key solution to enable higher densities, as it can effectively remove heat from high-power components and allow for more compact rack designs.

  4. However, increasing density also has implications for power distribution and overall data centre infrastructure. Higher-density racks require more power, which may strain existing electrical systems and require upgrades to transformers, switchgear, and other components.

  5. Increasing density can also impact data centre layout and space utilisation, as high-density racks may require specialised containment systems, dedicated cooling loops, and other infrastructure modifications.

  6. To accommodate increasing density, data centres may need to adopt modular or containerised designs, which allow for more flexible and scalable deployment of high-density systems.

  7. Ultimately, the ability to support increasing density will depend on a data centre's ability to provide adequate power, cooling, and infrastructure to meet the demands of high-performance computing systems.

In summary, the adoption of liquid cooling in data centres is driven by the need to support increasing rack densities and improve power efficiency.

However, implementing liquid cooling can be costly and challenging, particularly in legacy data centres.

Power conservation benefits from liquid cooling, while significant, may be outpaced by the growing power demands of high-density systems.

As density continues to increase, data centres will need to carefully plan and invest in infrastructure upgrades, modular designs, and advanced cooling technologies to keep pace with the evolving requirements of AI and other demanding workloads.

QCT's L04V air-assist liquid cooling solution

QCT's L04V air-assist liquid cooling solution is designed to address the growing need for efficient cooling in data centres as the power per socket increases, especially beyond 300W.

The system is built on a standard 42 unit rack and is compatible with various rack sizes.

Here's a detailed explanation of how it works and its key features:

Liquid-to-Air Cooling

  • The system uses a closed-loop liquid cooling system with cold plates covering the CPUs.

  • Cold coolant enters the server through blue cables, absorbs heat from the CPUs, and exits as hot coolant through red cables.

  • The hot coolant is then passed through a rear door heat exchanger, where fans dissipate the heat, cooling the liquid before it returns to the servers.

Liquid-to-Liquid Cooling

  • For data centres with existing liquid cooling infrastructure, QCT offers a liquid-to-liquid version of the L04V.

  • In this configuration, the rear door radiator and fans are replaced with a plate exchanger and flow control valve.

  • This allows the system to integrate with the data centre's existing liquid cooling loop.

Cooling Distribution Unit (CDU)

  • The heart of the L04V system is the 4U CDU located at the bottom of the rack.

  • 4U CDU Placement: Positioned at the bottom of the rack, taking up 4 units of rack space.

  • It features hot-swappable pumps to ensure continuous operation even if one pump fails.

  • The CDU also includes a filter to maintain liquid quality and an LCD monitor for status monitoring.

Server-grade BMC in CDU

  • QCT has integrated a server-grade Baseboard Management Controller (BMC) chip inside the CDU, running open BMC code.

  • This allows for advanced management capabilities, such as monitoring server health, cooling requirements, and leak detection.

  • The CDU can intelligently control cooling pumps and rear door exchanger fans to optimize power efficiency based on server load and cooling needs.

Performance and Cost Benefits

  • Liquid cooling enables better thermal management, allowing CPUs to run at higher and more consistent clock speeds (turbo boost) compared to air cooling.

  • The L04V system can achieve up to 70% power savings compared to traditional air cooling solutions by efficiently cooling servers and operating at higher ambient temperatures (up to 40°C).

  • The higher ambient temperature reduces the need for power-hungry HVAC systems, significantly lowering the data centre's Power Usage Effectiveness (PUE).

Total Cost of Ownership (TCO) and Return on Investment (ROI)

  • Although liquid cooling has a higher initial capital expenditure (CAPEX) compared to air cooling, its operating expenses (OPEX) are significantly lower.

  • The TCO of liquid cooling grows at a slower rate over time, making it more cost-effective in the long run.

  • Depending on the system configuration and electricity costs, data centres can start seeing a positive ROI within a few years of deploying liquid cooling.

Management and Monitoring

  • The L04V system offers a web UI for live monitoring of the liquid cooling solution, individual server health, and location.

  • It also provides Redfish API support for integration with existing data centre management tools.

  • An LCD monitor on the rack provides real-time information for on-site servicing.

Best practices for implementing liquid cooling include

  1. Thorough planning and design to ensure compatibility with existing infrastructure and future scalability.

  2. Regular maintenance of the liquid cooling system, including monitoring coolant quality and replacing filters as needed.

  3. Utilizing the management and monitoring capabilities to optimize cooling performance and efficiency.

  4. Training personnel on liquid cooling technology and service procedures.

  5. Gradually transitioning from air cooling to liquid cooling, allowing for a smooth migration and minimizing disruption to data centre operations.

In summary, QCT's L04V air-assist liquid cooling solution addresses the growing need for efficient cooling in high-density data centres. Its innovative design, advanced management features, and cost benefits make it an attractive option for data centers looking to optimise performance and reduce operating expenses in the face of increasing power demands.

Page cover image