LLM Inference in C++ (Paperback)
Language: English
Published by Independently Published, 2026
- Softcover
- New
Seller: Grand Eagle Retail, Bensenville, IL, U.S.A.Grand Eagle Retail
AbeBooks seller since October 12, 2005
Condition: New
£ 28.18
Quantity: 1 available
Add to basketItem description from seller
Paperback. Stop Wasting GPU Compute. Build the High-Throughput, Low-Latency AI Infrastructure of 2026. The "VRAM Wall" is the biggest bottleneck in modern AI. Standard Python wrappers and out-of-the-box runtimes are fine for prototyping, but at scale, memory fragmentation and Global Interpreter Lock (GIL) overhead will destroy your throughput. LLM Inference in C++ is the definitive engineering manual for bypassing Python entirely and building custom, bare-metal inference engines that maximize hardware utilization. Focusing on the cutting-edge 2026 landscape, this book bridges the gap between high-level AI concepts and low-level GPU execution. You will learn how to implement enterprise-grade features like PagedAttention, FlashAttention-3, and Continuous Batching directly in C++ and CUDA, unlocking massive performance gains for large-scale language models.Inside, you will discover: Hardware-Aware Memory Management: Eliminate memory waste by implementing PagedAttention logic and custom allocators to bypass std:: malloc overhead.Accelerated Tensor Algebra: Master C++23's std:: mdspan and write fused SIMD kernels with AVX-512 to minimize GPU context switching.Custom CUDA Kernels: Write high-speed FlashAttention-3, LayerNorm, and RMSNorm kernels while managing CUDA streams for maximum GPU occupancy.The Cost Killer (Quantization): Slash VRAM requirements with bit-level manipulation for 4-bit (AWQ) and 8-bit (FP8) inference using NVIDIA Tensor Cores.Distributed & Speculative Execution: Scale across clusters using zero-copy NCCL/RDMA interconnects and implement Draft Models to accelerate massive architectures.The Production Serving Layer: Build lock-free C++ request queues for continuous batching and track P99 "Time to First Token" (TTFT) at the systems level.THE IMPLEMENTATION VAULT (Appendix) Built for the infrastructure engineer in the trenches, the Appendix provides immediate, battle-tested utility: The 15-Point Production-Ready Checklist: Your mandatory safety and performance audit before deploying any custom engine.Latency vs. Throughput Reference Table: The ultimate cheat sheet for balancing batch sizes against user wait times.Troubleshooting Guide: Direct solutions for the top 10 most common and devastating CUDA and C++ memory errors.Don't let inefficient software architecture throttle your hardware. Master C++ LLM inference and build the fastest, most cost-effective AI engines in the industry. This item is printed on demand. Shipping may be from multiple locations in the US or from the UK, depending on stock availability.…
Seller Inventory # 9798259069299
- Title
- LLM Inference in C++ (Paperback)
- Author
- Billie S. Lightner
- Publisher
- Independently Published
- Publication year
- 2026
- Condition
- new
- Binding
- Paperback
- Language
- English
- ISBN 13
- 9798259069299
- Series
- Book 7 of 11: High-Performance C++ Engineering
The "VRAM Wall" is the biggest bottleneck in modern AI. Standard Python wrappers and out-of-the-box runtimes are fine for prototyping, but at scale, memory fragmentation and Global Interpreter Lock (GIL) overhead will destroy your throughput. LLM Inference in C++ is the definitive engineering manual for bypassing Python entirely and building custom, bare-metal inference engines that maximize hardware utilization.
Focusing on the cutting-edge 2026 landscape, this book bridges the gap between high-level AI concepts and low-level GPU execution. You will learn how to implement enterprise-grade features like PagedAttention, FlashAttention-3, and Continuous Batching directly in C++ and CUDA, unlocking massive performance gains for large-scale language models.
Inside, you will discover:
- Hardware-Aware Memory Management: Eliminate memory waste by implementing PagedAttention logic and custom allocators to bypass std::malloc overhead.
- Accelerated Tensor Algebra: Master C++23's std::mdspan and write fused SIMD kernels with AVX-512 to minimize GPU context switching.
- Custom CUDA Kernels: Write high-speed FlashAttention-3, LayerNorm, and RMSNorm kernels while managing CUDA streams for maximum GPU occupancy.
- The Cost Killer (Quantization): Slash VRAM requirements with bit-level manipulation for 4-bit (AWQ) and 8-bit (FP8) inference using NVIDIA Tensor Cores.
- Distributed & Speculative Execution: Scale across clusters using zero-copy NCCL/RDMA interconnects and implement Draft Models to accelerate massive architectures.
- The Production Serving Layer: Build lock-free C++ request queues for continuous batching and track P99 "Time to First Token" (TTFT) at the systems level.
Built for the infrastructure engineer in the trenches, the Appendix provides immediate, battle-tested utility:
- The 15-Point Production-Ready Checklist: Your mandatory safety and performance audit before deploying any custom engine.
- Latency vs. Throughput Reference Table: The ultimate cheat sheet for balancing batch sizes against user wait times.
- Troubleshooting Guide: Direct solutions for the top 10 most common and devastating CUDA and C++ memory errors.
"Synopsis" may belong to another edition of this title.
Grand Eagle Retail
Bensenville, IL, U.S.A.
AbeBooks seller since October 12, 2005
Shipping rates within U.S.A.
| Item | 6 to 14 business days | 6 to 16 business days |
|---|---|---|
| First item | £ 0.00 | £ 0.00 |
Payment methods
Seller's business information
APOLLO ONLINE CORP.
605 Geddes Street
Wilmington, DE U.S.A. 19805
Terms of sale
We guarantee the condition of every book as it¿s described on the Abebooks web sites. If you¿ve changed
your mind about a book that you¿ve ordered, please use the Ask bookseller a question link to contact us
and we¿ll respond within 2 business days.
Books ship from California and Michigan.
Right of withdrawal
If you are a consumer you can withdraw from the contract in accordance with the following. Consumer means any natural person who is acting for purposes which are outside his trade, business, craft or profession.
Information regarding the right of withdrawal
Statutory right to withdraw
You have the right to withdraw from this contract within 14 days without giving any reason.
The withdrawal period will expire after 14 days from the day on which you acquire, or a third party other than the carrier and indicated by you acquires, physical possession of the last good or the last lot or piece.
To exercise the right of withdrawal, electronically fill in and submit a clear statement on our website, under "My Purchases" in "My Account". We will communicate to you an acknowledgement of receipt of such a withdrawal on a durable medium (e.g. by e-mail) without delay.
To meet the withdrawal deadline, it is sufficient for you to send your communication concerning your exercise of the right of withdrawal before the withdrawal period has expired.
Effects of withdrawal
If you withdraw from this contract, we will reimburse to you all payments received from you, including the costs of delivery (except for the supplementary costs arising if you chose a type of delivery other than the least expensive type of standard delivery offered by us).
We may make a deduction from the reimbursement for loss in value of any goods supplied, if the loss is the result of unnecessary handling by you.
We will make the reimbursement without undue delay, and not later than 14 days after the day on which we are informed about your decision to withdraw from this contract.
We will make the reimbursement using the same means of payment as you used for the initial transaction, unless you have expressly agreed otherwise; in any event, you will not incur any fees as a result of such reimbursement.
We may withhold reimbursement until we have received the goods back, or you have supplied evidence of having sent back the goods, whichever is the earliest.
You shall send back the goods or hand them over to Grand Eagle Retail, Bensenville, Illinois, U.S.A., without undue delay and in any event not later than 14 days from the day on which you communicate your withdrawal from this contract to us. The deadline is met if you send back the goods before the period of 14 days has expired. You will have to bear the direct cost of returning the goods. You are only liable for any diminished value of the goods resulting from the handling other than what is necessary to establish the nature, characteristics and functioning of the goods.
Exceptions to the right of withdrawal
The right of withdrawal does not apply to:
- The delivery of newspapers, journals or magazines with the exception of subscription contracts; and
- The supply of digital content which is not supplied on a tangible medium (e.g. on a CD or DVD) if you accepted when you placed your order that we could start to deliver it, and that you could not withdraw once delivery had started.
Shipping terms
Orders usually ship within 2 business days. All books within the US ship free of charge. Delivery is 4-14 business days anywhere in the United States.
Books ship from California and Michigan.
If your book order is heavy or oversized, we may contact you to let you know extra shipping is required.