Re: [patch] gsoc, improving hash index v2

<div dir="ltr">Hi, hackers. Here is some test I run on a bigger set.<br /><br />Use a word list of  39916800 unique
words<br/>The table size is 1529MB.<br />Index      BuildTime    IndexSize<br />----        ----            ----<br />
btree     874470.339ms  1027MB<br /> hash-patch   513026.381 ms   1024MB<br /><br />I use pgbench to test the time of a
customquery script.<br />There are 2000 statements in the script.<br />It looks like this:<br />select * from dict
whereword='123456789a0'<br /> ...<br />The time of the two index is<br />btree: 1/0.174700=5.00250125<br />hash-patch:
1/0.199900=5.724098<br/><br />---------------btree------------------<br /> $ pgbench -n -f /tmp/query.sql dict<br
/>transactiontype: Custom query<br /> scaling factor: 1<br />query mode: simple<br />number of clients: 1<br />number
oftransactions per client: 10<br />number of transactions actually processed: 10/10<br />tps = 0.174694 (including
connectionsestablishing)<br />tps = 0.174700 (excluding connections establishing)<br clear="all" /><br
/>---------------hash patch-------------<br />$ pgbench -n -f /tmp/query.sql dict<br />transaction type: Custom
query<br/>scaling factor: 1<br />query mode: simple<br />number of clients: 1<br />number of transactions per client:
10<br/> number of transactions actually processed: 10/10<br />tps = 0.199892 (including connections establishing)<br
/>tps= 0.199900 (excluding connections establishing)<br /><br />-- <br />Best Regards,<br />Xiao Meng<br /><br />DKERC,
HarbinInstitute of Technology, China<br /> Gtalk: <a href="mailto:mx.cogito@gmail.com"
target="_blank">mx.cogito@gmail.com</a><br/>MSN: <a href="mailto:cnEnder@live.com"
target="_blank">cnEnder@live.com</a><br/><a href="http://xiaomeng.yo2.cn" target="_blank">http://xiaomeng.yo2.cn</a><br
/></div>