#gpu
18pages tagged with "gpu"
infrastructure
- ai infrastructure boom timeline— detailed chronicle of ai datacenter infrastructure from pre-chatgpt era through 2025, tracking 149 ai/ml projects, chatgpt launch impact, gpu deplo...
- ai infrastructure competition— competitive analysis of specialized ai data center providers including coreweave, crusoe, applied digital, and the stargate initiative
- ai/ml data center projects analysis— comprehensive analysis of 140 ai/ml data center projects representing 538b investment and 65 gw capacity, driving unprecedented infrastructure demand
- datacenter technology overview— comprehensive overview of datacenter technology evolution from traditional compute to ai-optimized infrastructure with gpu clusters and extreme cooling requirements
- technology shifts and innovation trends— analysis of datacenter technology evolution including gpu acceleration, liquid cooling adoption, power density increases, and infrastructure innova...
programming
- building nvidia drivers with clang/lto— how to get the nvidia open kernel modules (dkms) to build against a linux kernel compiled with clang and thinlto: two independent failures, gnu ld vs. llvm bitcode and objtool vs. the prebuilt blobs, with symptom, cause, fix, and verification for each
- cuda 13.0 release overview— comprehensive guide to nvidia cuda 13.0 features, changes, and migration considerations
- cuda docker containers— gpu acceleration in docker containers with nvidia container toolkit
- cuda gpu computing setup— overview and decision guide for cuda/gpu setup on linux systems
- cuda native installation— installing cuda toolkit directly on linux for development and compilation
- gh200 setup for pytorch training (october 2026)— a tested, reproducible pytorch training environment on an nvidia gh200 grace hopper box: uv, torch 2.13 cu129 on aarch64, torch.compile, document-masked attention without flash-attn, moe grouped gemm kernels, and the gotchas
- moore threads— chinese gpu design company founded in 2020
- nvidia gpu compute capability reference— complete list of nvidia gpus and their cuda compute capabilities
- pytorch cuda setup for nvidia gh200 (arm64)— comprehensive guide to configuring pytorch with cuda support on nvidia gh200 grace hopper systems running arm64 architecture
- small encoder training throughput on consumer gpus— measured training throughput for 5m and 21m parameter transformer encoders across rtx 5070 ti, 4070 ti super, 4060 ti, and zen 4 cpus
technology
- GPU Infrastructure: The 1M+ GPU Deployment Powering US AI— Comprehensive analysis of GPU infrastructure deployed across US datacenters, documenting the 1M+ GPUs powering AI workloads with detailed specifica...
- Rack Density Evolution: From 5kW to 350kW Per Rack— Comprehensive analysis of datacenter rack density evolution from traditional 5-15kW enterprise racks to ultra-high-density 350kW AI infrastructure, with survey data on how far the installed-base average (7-27kW depending on who is counting) lags the leading edge.