Sydney date / AI technology
News archive
Sunday 15 November 2026 · 26 stories
AI technology
26
CUDA adoption: from early GPU experiments to the AlexNet breakthrough
AlexNet: the CUDA engineering behind the 2012 breakthrough
Inside cuda-convnet: AlexNet dual-GPU layers and the ImageNet era
Tensor Cores, Triton, residual networks and vision transformers
Self-attention, CUDA kernels and FlashAttention
FlashAttention scaling, multi-GPU training and weight quantization
PagedAttention, speculative decoding, MoE and Blackwell
CUDA Graphs, reasoning models and disaggregated serving
Rubin, custom AI chips and compressed KV caches
LLM inference, fused kernels and GPU networking
GPU benchmarks, 70B memory sizing and AI compilers
FSDP2, GPU profiling and mobile compute
Practical GPU engineering: Metal shaders, profiling and edge deployment
Choosing AI optimization strategies for your hardware
Modern CUDA C++ and Triton: control, productivity and workload choices
GEMM in CUDA C++ and Triton, torch.compile and ROCm portability
Generated Triton code, GEMM examples and ROCm warp-size pitfalls
SwiGLU fusion, FlexAttention and a GPU optimization cheat-sheet
Google expands multi-agent research workflows
Hospital-local AI highlights evidence-linked clinical support
Anthropic extends free Claude access for US teachers
OpenShell contribution puts agent governance in focus
SpaceX successfully catches Starship rocket returning to Earth
Musk warns Nvidia could lose critical supply-chain deals without Huang
