December 11, 2002 at 3:52 am
Could be a number of things. I am not really sure the CPU time is accurate thou as I see various results that I cannot get a picture of anywhere else. Even then it may depend on server load. Now the 5s to 120s depending on which one was first could be if the time got shorter was pulling from the cache an not the physical Hard Drive and it saw the last time it was run was still marked in memory. If got slower could be a bottleneck or server load or a number of other factors that caused it, hard to know 100% without acutally monitoring al the possiblities.
December 11, 2002 at 7:15 am
I know Elapsed time may depend on ... many things, but should I think CPU time is not
an accurate and steady data?
Many performance tests I make are based
on it!
December 11, 2002 at 7:44 am
I believe it is accurate but it is also subjective to various factors. Unless you are running on a system with no load and you use the DBCC processes to flush the execution plans and cached data from memory the CPU is not a good value to go on. Run it once, then immediately run again and you will see a change most times for the better. However strain the server with a major load that will push the data from the cache in memory then it will go up to handle the data movement. Use it with a grain of salt IMHO.
Viewing 3 posts - 16 through 17 (of 17 total)
You must be logged in to reply to this topic. Login to reply