diff --git a/README.md b/README.md index a848380..a3bd416 100644 --- a/README.md +++ b/README.md @@ -21,8 +21,9 @@ FlexSearch also provides you a non-blocking asynchronous processing model as wel Installation Guide  •  API Reference  •  Example Options  •  Custom Builds Comparison: -- Library Benchmarks +- Library Query Benchmarks - Library Relevance Scoring +- Library Memory Consumption Supported Platforms: - Browser @@ -83,6 +84,7 @@ Workers get its own dedicated memory. Especially for larger indexes, web worker __Note:__ It is slightly faster to use no web worker when the index or query isn't too big (index < 500,000 words, query < 25 words). + ## Installation @@ -894,17 +897,35 @@ The required memory for the index depends on several options: + +#### Compare Memory Consumption + +The book "Gulliver's Travels" (Swift Jonathan 1726) was used for this test. + +
+ + ## Built-in Profiles You can pass a built-in profile during creation/initialization. They have these following settings: +Standard profile: __"default"__ +```js +{ + encode: "icase", + mode: "forward" +} +``` + Memory-optimized profile: __"memory"__ ```js { encode: "extra", mode: "strict", - threshold: 7 + threshold: 7, + stemmer: true, + filter: true } ``` diff --git a/doc/memory-comparison.svg b/doc/memory-comparison.svg new file mode 100644 index 0000000..e17ede1 --- /dev/null +++ b/doc/memory-comparison.svg @@ -0,0 +1,3 @@ + + +661035 bytes3518156 bytes4341000 bytes117021480 bytes2398442 bytes2227612 bytes4746825 bytes0500000100000015000002000000250000030000003500000400000045000005000000RAMFlexSearchBulkSearchElasticlunrLunrWadeJSiiBM25Memory Consumption of Search Engines (lower ist better)https://github.com/nextapps-de/flexsearch/