The MySQL optimizer can select an inappropriate index due to inaccurate statistics or sampling errors, but you can correct this by refreshing statistics with ANALYZE TABLE, increasing sample pages, or explicitly forcing the right index using the FORCE INDEX hint.
This article systematically dissects the engineering decisions behind frontier large‑language‑model training—covering architecture choices, attention variants, optimizer evolution, data‑curation strategies, scaling‑law insights, and post‑training SFT/RL pipelines—based on open‑source reports from seven leading AI laboratories.
This article walks through the complete lifecycle of a MySQL SELECT statement, detailing how the client, connector, optional query cache, parser, optimizer, and executor interact to transform SQL text into an execution plan and finally return results.
This article provides a thorough analysis of nanochat’s source code, detailing transformer component differences, precise parameter‑size formulas, FlashNorm and ReLU² innovations, scaling‑law insights, memory‑usage estimations, and the distributed optimizer and training pipelines used to build the model.
This article breaks down the core deep‑learning optimizers—gradient descent, Momentum, AdaGrad, RMSProp and Adam—showing why vanilla gradient descent converges slowly, how each method uses exponential moving averages to accelerate training, and why Adam is generally the preferred choice.
This article explains Oracle's step‑by‑step processing of a SQL statement—including the client submission, parsing checks, optimizer decision‑making, row‑source generation, and final execution—while illustrating hard and soft parses, optimizer components, and practical query‑plan queries.
This article walks through a suite of PyTorch techniques—including automatic mixed precision, BF16, gradient checkpointing, gradient accumulation, tensor sharding, efficient data loading, in‑place ops, lightweight optimizers, memory profiling, TorchScript, and kernel fusion—that together can cut peak GPU memory usage by up to twenty‑fold while preserving model accuracy.
This article explains the underlying principles of the Spark SQL optimizer and presents three classic optimization paradigms—push‑down optimization, operator elimination/merging, and expression elimination/replacement—illustrating each with concrete rule implementations and code examples.
This tutorial walks Java developers through the complete workflow of building, training, and evaluating a neural network in PyTorch, covering network definition, data iteration, forward and backward passes, loss calculation, and parameter updates with detailed code examples and Java‑centric analogies.
This article explains MySQL's layered architecture, how the optimizer generates multiple execution plans, evaluates their I/O and engine costs, and selects the plan with the lowest total cost, illustrated with example queries, cost tables, and JSON output.
This article recounts a real‑world incident where MySQL 5.7 chose a sub‑optimal index, causing a three‑second slow query, and explains how the team identified the root cause, used EXPLAIN and optimizer trace, and applied fixes such as forced index hints and data cleanup.
This article explains the complete lifecycle of a MySQL SELECT statement, covering the server and storage‑engine layers, the roles of the connector, query cache, parser, optimizer and executor, and provides concrete command‑line and SQL examples.
This article explains how MySQL handles queries that involve two separate indexed columns, demonstrates the index_merge execution plan with real‑world examples, details the three index‑merge strategies (intersect, union, sort_union), and shows how to control or disable the feature for better performance.
This article provides a comprehensive overview of MySQL 8.0 enhancements, covering performance improvements such as contention‑aware scheduling and hash joins, security upgrades including authentication plugins and password policies, optimizer refinements, and various other features like persistent variables, GIS support, and binlog expiration handling.
This article compiles practical deep learning training and hyperparameter tuning advice from Zhihu contributors, covering model debugging, learning‑rate strategies, optimizer choices, data preprocessing, regularization techniques, initialization methods, common pitfalls, recommended research papers, and ensemble approaches.
This article explains MySQL’s internal execution flow, covering the connector’s role, permission verification, caching, parsing, optimization, execution stages, processlist states, SQL execution order, and best practices for arranging WHERE‑clause conditions to improve query performance.
This article explains how MySQL 8.0's prefer_ordering_index optimizer switch can be toggled to improve LIMIT queries that involve ORDER BY and GROUP BY, demonstrates the effect with test data, and shows why manual parameter control can yield better execution plans.
A MySQL 5.6+ optimizer bug causes queries that filter by uid and order by id with a small LIMIT to use a costly full table scan instead of the appropriate idx_uid_stat index, and the article explains the root cause, shows optimizer_trace output, and offers two practical work‑arounds.
This article investigates a MySQL crash triggered by a specific SELECT statement, analyzes the stack trace and optimizer behavior that leads to an invalid memory access, and presents three practical solutions including disabling DuplicateWeedout, upgrading MySQL, and normalizing table character sets.
PolarDB‑X evolved from Alibaba’s internal sharding middleware TDDL to the DRDS cloud service and finally to a fully distributed MySQL‑compatible database, introducing a full SQL optimizer, MPP engine, global secondary indexes, strong‑consistent distributed transactions via TSO + 2PC, and full binlog compatibility.
This article explains MySQL's client‑server architecture, detailing how the server processes connections, parses and optimizes queries, generates execution plans, and interacts with various storage engines, while also covering configuration variables, connection monitoring, and practical command examples.
This article explains how to configure and use PyTorch optimizers, their attributes and methods, and demonstrates various learning‑rate scheduling techniques—including manual updates and built‑in schedulers such as LambdaLR, StepLR, MultiStepLR, ExponentialLR, CosineAnnealingLR, and ReduceLROnPlateau—through clear code examples.
The article examines a MySQL optimizer bug where a query with a WHERE clause on uid, ORDER BY id ASC, and LIMIT 1 incorrectly triggers a full table scan despite a usable composite index, explains the optimizer’s cost calculations, and offers workarounds such as FORCE INDEX or a harmless arithmetic trick to force index usage.
The tutorial walks through building a complete LLVM‑based Kaleidoscope compiler—from a tokenizing lexer and recursive‑descent parser, through AST construction, LLVM IR generation with SSA and phi nodes, optimization passes, JIT execution, and mutable variable handling via stack allocation—providing full C++ source examples.
The article analyzes a MySQL optimizer bug where a query with ORDER BY id ASC LIMIT 1 triggers a full‑table scan despite an applicable idx_uid_stat index, explains the cost‑based decision process, and presents two practical work‑arounds—using FORCE INDEX or a harmless arithmetic expression—to ensure the index is used.
This article explains MySQL 8.0’s optimizer architecture and walks through the detailed transformation steps for subqueries, scalar‑to‑derived conversions, flattening, join simplification, partition pruning, and condition push‑down, illustrating each phase with code examples and execution‑plan screenshots.
The article explains MySQL 8.0.22's new optimizer_switch variable derived_condition_pushdown, shows how enabling it can cause unexpected results with user‑defined variables in outer WHERE clauses, provides test cases, explains the underlying behavior, and offers three practical solutions.
This article explains MySQL's query cache settings, clustered and non‑clustered index design, optimizer cost calculations, how to interpret EXPLAIN output, and the roles of redo/undo logs and transaction isolation levels, providing practical examples and performance tips.
This article explains the data structures behind MySQL indexes, demonstrates how B+Tree indexes are built, and shows why the optimizer may choose full‑table scans over index scans in various scenarios such as pattern matching, OR conditions, type conversion, and NULL handling.
This article examines a slow MySQL 5.7 query, uses the optimizer trace tool to reveal why the optimizer chose an inefficient plan, explains the role of fan‑out estimation on information_schema tables, and shows how hints can be applied to correct the plan.
An overview of MySQL 8.0’s release timeline, including major GA versions, and a detailed summary of its new capabilities such as transactional data dictionary, atomic DDL, enhanced security, role support, InnoDB improvements, JSON enhancements, optimizer extensions, backup lock, connection management, and other performance and management features.
This article provides a comprehensive overview of SparkSQL's logical plan architecture, detailing the stages of logical plan creation, analysis, rule‑based optimization, and the underlying catalog and rule systems that transform SQL queries into efficient execution plans.
This article introduces Xiaomi's open‑source SOAR tool, explains how to download and install the binary or source versions, walks through quick‑start commands, details key YAML configuration options and output formats, and describes four typical usage scenarios for SQL optimization and rewriting.
MySQL 8.0.17 introduces server cloning via MySQL Shell, multi‑value indexes for JSON arrays, enhanced JSON schema validation, optimizer and Volcano iterator upgrades, new collations, replication and routing enhancements, plus numerous bug fixes and deprecations, all detailed in the official release notes.
This article explains how MySQL 8.0 improves the optimizer by introducing configurable cost model constants and a histogram feature, showing how to query and update system tables, use ANALYZE TABLE to create and manage histograms, and explore the underlying code structures and future tuning possibilities.
Xiaomi is open‑sourcing its Go‑based SQL Optimizer and Rewriter (SOAR) at OSCAR Days, a tool that automatically checks SQL quality, suggests index improvements, rewrites queries, provides EXPLAIN analysis, and integrates with editor plugins and a web self‑audit platform to boost database reliability and developer productivity.
This article describes how Meituan‑Dianping leverages deep learning, especially the Wide & Deep model, to improve its recommendation system by addressing business diversity, user context, feature engineering challenges, optimizer and loss function choices, and presents offline and online experimental results showing significant CTR gains.
The latest MySQL release introduces a transactional data dictionary, role‑based account permission management, multiple InnoDB enhancements—including stable auto‑increment counters, memcached multi‑key support, dynamic deadlock detection, shared temporary tables, and NOWAIT/SKIP LOCKED options—plus advanced JSON operators, aggregation functions, pretty‑print utility, and optimizer features such as hidden and descending indexes.
This guide explains the role of optimizer statistics in Oracle, when to gather them, how to use automatic and manual collection jobs, tune ESTIMATE_PERCENT and METHOD_OPT, manage histograms, handle bind peeking, apply dynamic sampling, and safely test changes with pending statistics.
The article recaps a MySQL 8.0 optimizer meetup where experts discussed version naming, enterprise vs community edition differences, character‑set choices, storage‑engine evolution, hint usage, sharding strategies, high‑availability architectures, market position, and upcoming feature requests from the community.
