Archives of the TeradataForum
Message Posted: Mon, 05 Dec 2011 @ 07:11:09 GMT
| Subj: | | AmpCPUTime |
| |
| From: | | Attila_Finta |
In DBQL, Cpu Time measures come in two varieties: Normalized and non-Normalized, e.g. AmpCpuTime and AmpCpuTimeNorm.?
Teradata manuals say this about Norm vs. non-Norm CPU times: ?In regards to CPU information for co-existence system, the CpuNorm field stores
normalized CPU time. ?In a co-existence systems, AMPs running on faster nodes process data with fewer CPU time than AMPs running on nodes with
slower CPUs even though they are processing the same amount of data. CpuNorm is calculated using Teradata-supplied normalization factors to make
performance statistics comparable across different hardware platforms.?
Our production system has homogeneous nodes, yet DBQL shows different measurements for normalized and non-normalized CPU times.? CpuTimeNorm is
always 39.8x CpuTime.? So I suppose either number is equally usable, but why are these numbers different in a homogeneous system??
Attila Finta
| 
