June 25, 2026 · AI
Speeding up LLM inference with MTP and diffusion
MTP and diffusion inference on Gemma 4 and Qwen 3.6, fp8 on one H100
#LLM
#vLLM
#MTP
#Speculative Decoding
#Benchmark
한국어May 21, 2026 · AI
GraphDB Benchmark (2/2) — Workload Matrix and Final Recommendations
Eight graph engines measured across OLTP, memory, analytics, and differentiation queries
#GraphDB
#PostgreSQL
#RCTE
#Neo4j
May 21, 2026 · DB
PG-Strom SSB Benchmark — Arrow FDW Comes Before GPU
Why Arrow+GPU achieves 10x at SF=100 and Heap+GPU loses to CPU on wide tables
#PG-Strom
#PostgreSQL
#GPU
#OLAP
May 20, 2026 · AI
PDF to Markdown — Five Tools Compared
Five PDF-to-Markdown converters (markitdown, pdftotext, pymupdf, mineru, opendataloader-pdf) scored against a seven-criterion 100-point rubric
#PDF
#Markdown
#RAG
#Ingestion
May 20, 2026 · AI
Apple Silicon LLM Inference — Five Backends Compared
Benchmarking Qwen3.5-9B on Apple Silicon across MLX, llama.cpp, Ollama, omlx, and vLLM Metal — single-request throughput, prefill scaling, decode vs input length, and concurrency response
#LLM
#Apple Silicon
#MLX
#llama.cpp
May 20, 2026 · AI
LightRAG Without Apache AGE — Graph Storage in Recursive CTE
Implementing LightRAG's BaseGraphStorage on plain PostgreSQL with RCTE — why a 1-hop-dominant retrieval pattern fits flat SQL
#LightRAG
#PostgreSQL
#RAG
#GraphRAG
May 20, 2026 · AI
Long-Context Evaluation — NIAH and Lost in the Middle
NIAH limits, the Lost in the Middle effect, alternative benchmarks, and measured recall across four reasoning-effort modes
#LLM
#Long-context
#NIAH
#Benchmark
May 20, 2026 · AI
GraphDB Benchmark, Eight Engines (Part 1) — Decomposing the RCTE 290x Gap
On a 1.14M-edge knowledge-graph workload, PostgreSQL RCTE beats Apache AGE by 290x — tracing the cypher() wrapper's 13ms cost and PG plan generation accumulation
#GraphDB
#PostgreSQL
#RCTE
#Apache AGE
