tarantool/tarantool

Optimize large allocations

Open

#2,705 创建于 2017年8月22日

在 GitHub 查看
 (0 评论) (0 反应) (0 负责人)Lua (3,633 star) (406 fork)batch import
featuregood first issue

描述

Now that we support tuples bigger than 1M it's become necessary to optimize large allocations. jemalloc can reuse memory for large allocations, for example, while glibc malloc would always use mmap/munmap for large tuples.

Investigate whether we should import jemalloc or only the part of it relevant for large allocation management.

贡献者指南

技术栈
c
领域
performancebackend
议题类型
performance
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
3
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
3-5 days
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
fresh
清晰度议题是否清楚说明期望改动、验收标准和下一步。
mostly clear
前置要求
C programmingunderstanding of memory allocators
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
20
研究方向
Investigate the performance characteristics of jemalloc and glibc malloc for large allocations, especially tuples over 1M. Look at the current memory allocation code in tarantool, likely in tarantool/src/ or similar. Profile the allocation patterns and determine if jemalloc's reuse of memory provides a benefit. Consider importing only the relevant parts of jemalloc for large allocations. The discussion may involve maintainers to decide on the approach.
Optimize large allocations · tarantool/tarantool#2705 | Good First Issue