lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Harry Yu" <502437...@qq.com>
Subject 回复: Speed up searching in multiple-thread?
Date Mon, 15 Sep 2014 08:15:46 GMT
Hi Michael,


Thanks for your reply. Yes, core i5 is 2 real (4 with hyperthreading) cores. And 
is this the core reason? Are my test methods wrong? 


In my test environment, I used JMeter to simulate 30 threads, and access my search ‍
service. I recorded per-search time and analyzed them.


Best Regards,


Harry Yu




------------------ 原始邮件 ------------------
发件人: "Michael McCandless";<lucene@mikemccandless.com>;
发送时间: 2014年9月15日(星期一) 下午3:48
收件人: "Lucene Users"<java-user@lucene.apache.org>; 

主题: Re: Speed up searching in multiple-thread?



If you run 30 search threads on a core i5 it's expected there will be
big slowdowns in the per-query latency since core i5 only has 2 real
(4 with hyperthreading) cores?

Mike McCandless

http://blog.mikemccandless.com


On Mon, Sep 15, 2014 at 3:10 AM, Harry Yu <502437499@qq.com> wrote:
> Dear members at Lucene project,
>
> I'm developing poi search application using lucene 4.8 . Recently, I met a trouble that
the performance of IndexSearcher.search is bad in multiple-thread environment. According the
test results, I found that if thread number is 1,  the response time of searching is only
17ms. But that would be 400ms when I increased threads to 30.
>
>
> In our application, the size of index is 882M, cpu is 2.4GHZ i5, and test machine is
mac pro. I read How to Make Searching Faster in Lucene-java wiki. And there are only singleton
instance of IndexWriter, IndexReader and IndexSearcher.
>
>
> Please give me some tips or solutions to speeding up searching in multiple-thread. Thanks.
>
>
> Best regards,
>
>
> Harry Yu‍

---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org
Mime
  • Unnamed multipart/alternative (inline, 8-Bit, 0 bytes)
View raw message