KV Cache Visualizations
Interactive explorations of GPU memory, KV cache optimization, and the scaling challenges facing modern LLM inference
KV Cache Memory Growth
Watch GPU memory fill up as context length grows across 5 different LLM architectures
Training Memory Simulation
Explore memory requirements during model training with different batch sizes and optimizers
Memory Projections 2020–2030
See how KV cache memory demands scale with projected context length growth over the decade
KV Cache Orchestrators
Compare LMCache, SGLang HiCache, UCM, and FlexKV approaches to cache management
Zipfian Distribution & Engram Cache
Visualize token access patterns and how Engram cache exploits frequency distribution
Agentic Cache Offloading
See how AI agents coordinate KV cache offloading across tiered storage