Measured data

Many IBM-compatible processor models are also measured for the LSPR, whenever machine time can be obtained.

In addition to actual processor measurements, there are two types of LSPR data which IBM believes carry essentially the same degree of accuracy as measurements. These are based on the actual processor measurements made and various analysis techniques.

Projections

Primary models of each processor series are always measured for the LSPR. However, it is not practical for IBM to allocate the necessary resource to measure every individual processor model announced. For those that are not measured, ITR numbers are projected. A projected ITR is based on the known processor internal design, and on the known characteristics of the subject workload on similarly designed processor models. Projections have been shown to have accuracy comparable to that of measurements, where subsequent testing has occurred.

Estimates

Many older processor models (both IBM and IBM compatible) that have been measured in the past, are no longer accessible for testing. It is highly desirable to maintain these processors in LSPR tables, for those SCPs that are supported. Therefore, as the LSPR moves ahead to more current software levels, ITRs for these processors must be estimated. Estimates are based on known deltas between the older software and the current software on similarly designed processor models.

As the capacities of today's processors grow ever larger, it becomes more difficult to commit the time and/or the external resources necessary for full, unconstrained LSPR measurements. Experience has shown that there are estimation techniques, based on making reduced resource measurements and analysis that can yield reliable results.

Estimated ITRs are considered to have an accuracy very close to that of actual measurements. Estimates are not provided in the LSPR without a high degree of confidence in the process, and in the resulting ITR values.