Re: Re: [metabase-dev] RE: [PEAR-DEV] Re: [binarycloud-dev] FW: letstalk "metapear" - politics aside:-)

Thanks Thies, I found out that my benchmark used: OCIExecute($stmt) when I should be using OCIExecute($stmt, OCI_DEFAULT) so it was commiting every record. Uggh. Tested your benchmark on a sun server (editted for clarity) timing 5000 inserts using parse time elapsed 3.56 timing 5000 inserts using parse & bind time elapsed 3.48 timing 5000 inserts using the same statment & bind time elapsed 1.98 Eg. Using bind vars was nearly twice as fast. Regards, John "Thies C. Arntzen" <thies@thiesos.org> wrote in message news:20020319184200.GB13320@schnuffel.thieso.net... > On Wed, Mar 20, 2002 at 01:06:28AM +0800, John Lim wrote: > > > > > Query caching: Some databases, like Oracle, have a cache of parsed and > > > optimized queries. By using prepare(), the database will need to parse > > > and optimize your query just once instead of every time you execute it. > > > This can be a really big speedup. > > > > > Hi Stig, > > > > Strangely enough, I recently benchmarked binding of variables and got some > > strange results. They were 6 times slower than sending pure sql. > > try my attached "insert_speed" benchmark. fior me it shows a > 3,5 fold speedup using bind and reusing the statement. > > re, > tc > >