repoGitHubTrust 82 · PrimaryPublished 10d agoLive · 10d ago
xcena-dev/maru
High-Performance KV Cache Storage Engine on CXL Shared Memory for LLM Inference
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 59%I mapped which local LLMs actually fit each RAM tier, 8 to 128GB (open dataset) →
- PossiblePossibly related (embedding) · 52%DeepSeek-V4-Flash (MXFP4): compute buffer scales ~3x just from KV cache quant type (f16 vs q8_0) — anyone else seeing this? Llama.cpp →
- PossiblePossibly related (embedding) · 51%OpenAI and Broadcom announce chip designed for LLM inference at scale →
- PossiblePossibly related (embedding) · 51%I compiled LLM inference pricing across 7 providers — the caching numbers are surprising(spreadsheet included) [R] →
- PossiblePossibly related (embedding) · 50%We'll benchmark an Open weights LLM on any GPU you choose — drop your model + hardware and we'll run it. [D] →
- PossiblePossibly related (embedding) · 60%FreqDepthKV: Frequency-Guided Depth Sharing for Robust KV Cache Compression in Long-Context LLM Inference →
- PossiblePossibly related (embedding) · 53%Reducing High-Bandwidth Memory Bottlenecks in JAX-Based LLM Training with Host Offloading - NVIDIA Developer →
- PossiblePossibly related (embedding) · 58%A JoLT for the KV Cache: Near-Lossless KV Cache Compression via Joint Tucker and JL-Residual Allocation for LLMs →
Covers
newsI mapped which local LLMs actually fit each RAM tier, 8 to 128GB (open dataset)newsDeepSeek-V4-Flash (MXFP4): compute buffer scales ~3x just from KV cache quant type (f16 vs q8_0) — anyone else seeing this? Llama.cppnewsOpenAI and Broadcom announce chip designed for LLM inference at scalenewsI compiled LLM inference pricing across 7 providers — the caching numbers are surprising(spreadsheet included) [R]newsWe'll benchmark an Open weights LLM on any GPU you choose — drop your model + hardware and we'll run it. [D]
Implements (incoming)
Covers (incoming)
Related across the graph
newsReducing High-Bandwidth Memory Bottlenecks in JAX-Based LLM Training with Host Offloading - NVIDIA DevelopernewsI compiled LLM inference pricing across 7 providers — the caching numbers are surprising(spreadsheet included) [R]newsOpenAI and Broadcom announce chip designed for LLM inference at scalepaperA JoLT for the KV Cache: Near-Lossless KV Cache Compression via Joint Tucker and JL-Residual Allocation for LLMsnewsRecent llama.cpp updates for SYCL/IntelnewsI mapped which local LLMs actually fit each RAM tier, 8 to 128GB (open dataset)newsDeepSeek-V4-Flash (MXFP4): compute buffer scales ~3x just from KV cache quant type (f16 vs q8_0) — anyone else seeing this? Llama.cppnewsWe'll benchmark an Open weights LLM on any GPU you choose — drop your model + hardware and we'll run it. [D]paperFreqDepthKV: Frequency-Guided Depth Sharing for Robust KV Cache Compression in Long-Context LLM Inference