Re: Optimizing..

Ray Garrison wrote: > > Performance tuning recommendations in the docs are *very* basic, and sometimes > at odds with experience. One specific example - the number of CPU virtual > processors. Docs specifically state you should not configure more CPU VPs into > the engine than physical CPUs in the box. Through trial and error, we have > found 50% more CPU VPs than physical CPUs provides faster throughput, measured > in the sense of "How many minutes does query XXX take to run?". On our > 4 CPU V2200 box we get best performance configuring 6 CPU VPs (we use KAIO, don't > know if that impacts results.) (yet ;) ) > > I saw a posting to this effect from someone at hp.com a while back, and tried > it. He was correct. I have also heard of this method. But if you think about it logically, one CPUVP per CPU should give you total CPU power. What is happening with more CPUVPs giving better performance, is that they are being switched too often. They are then being replaced with another process, and with so many CPUVPs around, with any luck it is one of them. I suggest this goes to process priority. You might try increasing the priority of your CPUVP processes. I think on HP this is best achieved manually, as opposed to setting NOAGE. Cheers, -- Mark. +----------------------------------------------------------+-----------+ |Mark D. Stock - Informix SA http://www.informix.com |//////// /| |mailto:mdstock@informix.com FAQ http://www.iiug.org |///// / //| | +-----------------------------------+//// / ///| | Tel: +27 11 807 0313 |If it's slow, the users complain. |/// / ////| | Fax: +27 11 807 2594 |If it's fast, the users keep quiet.|// / /////| |Cell: +27 83 250 2325 |Therefore, "No news: travels fast"!|/ ////////| +----------------------+-----------------------------------+-----------+