AI Briefing#performance-tuning
The latest AI and developer news about #performance-tuning, with the original source and a short summary.
The latest AI and developer news about #performance-tuning, with the original source and a short summary.
Removing duplicate Alloy collection and optimizing labels reduced Loki CPU usage by 98% on a sandbox basis.
The engineer reduced multi-hour sequential debugging cycles to parallel runs by isolating experiments via Git branches and automating deployment with AI agents.
Ollama's default settings for context length, parallel requests, and resident model limits significantly increase memory usage, causing GPU offloading failures.
It explains the relationship between compute, memory bandwidth, and overhead that determines deep learning performance, along with the underlying optimization principles.
To solve API latency issues arising from SDUI adoption, Oliveyoung introduced a dual caching strategy using Caffeine and Redis, improving response speed to under 1ms.
This analysis compares the operational principles of MySQL's Copy, In-Place, and Instant ALTER DDL algorithms and their Metadata Lock acquisition patterns.
This covers runtime flags, parallel processing, and Executor optimization methods to maximize the performance of ZIO applications.
A practical guide to minimizing STW by using GOMEMLIMIT instead of GOGC, and reducing GC load through Heap Profile analysis and code optimization tips.
A Kurly developer pointed out the inefficiency of JPA saveAll and improved 1:N data processing performance using JDBC Bulk Insert leveraging Auto Increment PK inference.
The video playback flickering issue on iOS 15-16 was resolved by changing the object-fit property and branching code by version.
Kurly applied a Kafka Batch Listener to prevent rebalancing and significantly increase search indexing processing speed.
Before deploying VSMS, Market Kurly resolved a deadlock and optimized the DB to achieve the target TPS of 1500.
Eliminated AntMatcher pattern matching and applied equals comparison to improve lock contention and memory usage.
To stabilize the real-time server for Alex's comment service with 600k DAU, 10k connections were verified using a GoLang client, and incidents were resolved by tuning VIP TPS limits and file descriptors.