johnj/llds
View on GitHub ↗ ★ 122 C
Low-Level Data Structure - efficient data structures, and fast data access in the 3.x/4.x/5.x/6.x kernel
9 mentions on Hacker News by 1 person: 2 stories, 7 comments · first 2012-05-30 · latest 2018-01-03
20082026
-
…the past it has been reserved for use cases like search engines with massive in-memory trees. [1] https://github.com/johnj/llds
comment by zippie · 2018-01-03 · in Linux page table isolation is not needed on AMD processors -
Supporting benchmarks that might be interesting alongside the OP: https://github.com/johnj/llds#wall-timings-in-seconds Same concept applies, reducing translations for page lookups reduces latency.
comment by zippie · 2014-02-23 · in How Bad Can 1 GB Pages Be? -
There is, of sorts (rbtree), http://github.com/johnj/llds
comment by zippie · 2013-03-28 · in Common Pitfalls in Writing Lock-Free Algorithms -
…I've run into this problem with our btree search indexes on bare metal (~96GB), some more reading: https://github.com/johnj/llds
comment by zippie · 2013-01-22 · in EC2 High Memory Cluster Eight Extra Large Instance (244 GiB RAM) -
…prefetching on modern processors with large data structures (usually limited to L1d cache lines): https://github.com/johnj/llds#l1-data-prefetch-misses-200000... A plausible analogy for CPU prefetching would be "readahead" mechanisms…
comment by zippie · 2012-10-10 · in Why is processing a sorted array faster than an unsorted array? -
…for instance treats cache lines as registers and heavily optimizes for cache friendly operations: https://github.com/johnj/llds
comment by zippie · 2012-09-02 · in Cache Lines Are The New Registers -
…memory access. We used Low-Level Data Structure to get the 2x improvements in memory latency: https://github.com/johnj/llds
comment by zippie · 2012-05-31 · in Latency numbers every programmer should know