Oracle Floating point vs Integer 1

[tcorum]

Does an Oracle8 server do more floating point or integer based caculations? I need to know this for performance reasons and I know little about databases. Any links explaining this would be a great help also.

Thanks,

Tcorum

 

[jimbopalmer]

the Oracle data type number is a float with about 22 digits, so USER numbers are pretty much all floats. internally it is mostly working with addresses so it is mostly integers.


Over the years I have upgraded my ERP database hardware many times, 40 Mhz to 55 to 125Mhz, then 50 Mhz to 85 Mhz, then 167 Mhz to 250 Mhz. I have always perceived the speedups as a funcion of the increase in size of the L2 cache, then integer performance. Floating point claims did not effect my ERP use of Oracle.

40 Mhz had 36k L2
55 had 256k
125 had 512k
50 had 1meg
85 had 1 meg
167 had 512k
250 had 4 meg

If I was asked to upgrade today I would have no hesitaation asking for 8 meg L2 cache as my 2:1 speed up, no matter what CPU speed

BTW, RAM interleave is also a effective speedup in a multiCPU box

I tried to remain child-like, all I acheived was childish.

 

[tcorum]

Thanks for your post Jim that was just the response I was looking for!

Thanks again,

Tcorum