Which Search content: type is faster in plain-text files?

[PeterPanino]

Which Search content: type is faster in plain-text files: iFilter or UTF-8 search?

[void]

UTF-8 search (utf8content:) is 3x faster than iFilter (content:) for me.

Code: Select all

search 'c:\dev\src\ ext:cpp;c;h utf8content:' filter '' sort 13
found 0 folders, size 0, db search time taken: 0.006393 seconds
found 60891 files, size 262384, db search time taken: 11.642357 seconds

Code: Select all

search 'c:\dev\src\ ext:cpp;c;h content:' filter '' sort 13
found 0 folders, size 0, db search time taken: 0.005335 seconds
found 60748 files, size 262384, db search time taken: 42.649868 seconds

Press Ctrl + ~ in Everything to show the debug console and check the performance you get for both functions.

[PeterPanino]

My first impression was that the content search is surprisingly FAST. The content search for a single term in 2000 source code files was almost instantaneous (< 1 second). I think that's what makes sense because I always know where to search and prepare several Search Bookmarks accordingly. On the other hand, it does not make sense to search for content with Everything among 1 million source code files.

What also might help a lot: A very fast SSD (look at the latest Intel models with 2.8 5 GB per second transfer speed) and a lot of RAM memory (32 GB or 64 GB DDR4 might also help). And of course a modern fast CPU.

[PeterPanino]

Have you also tried to make the content search multi-threaded? This should considerably improve the speed of the content search.

For example, you have to search 10,000 files for some content.Create 4 threads and let each thread search the content of 2,500 files. What do you think?