Back

Deep Kondah

Ghost
deep-kondah.com· irregular· Published 16 days ago

Long-form technical posts by a single author explaining computer systems and AI concepts with equations and diagrams, covering memory consistency on multicore CPUs, GPU architecture with CUDA and GEMM kernels, and the history of attention leading to multi-agent LLM systems. Readers get in-depth, first-principles walkthroughs of how these systems work.

memory consistency modelmemory reorderingmulticore processorscuda programminggpu architecturegemmtransformer attentionmulti-agent systemsllm agentsautoregressive language modelspipeliningcpu microarchitecture

Website: https://www.deep-kondah.com/