Experimental design and statistical modeling tool for workload characterization

A method for a cache architecture simulation, comprising:

• performing a uniform cache architecture experiment on a sample space to obtain preliminary data, wherein the preliminary data comprises a dense sample region;
• obtaining the first sample set using the non-stationary Gaussian field model, wherein the non-stationary Gaussian field model uses the preliminary data to obtain the first sample set;
• performing a cache architecture simulation using the first sample set to produce a first set of simulation data;
• fitting a first multivariate model to the first set of simulation data;
• fitting a first constrained multivariate model to the first multivariate model using an isonotizing algorithm;
• obtaining a second sample set using non-stationary Gaussian field model and the preliminary data if the first multivariate model is not within an acceptable error range, wherein the second sample set includes more data points than the first sample set;
• performing the cache architecture simulation using the second sample set to produce a second set of simulation data;
• fitting a second multivariate model to the second set of simulation data;
• and
• fitting a second constrained multivariate model to the second multivariate model using an isonotizing algorithm.

A method for cache architecture simulation comprising:

• determining a sample space;
• determining a dense sample region and a sparse sample region in the sample space;
• generating a sample set using a non-stationary Gaussian field model, wherein the non-stationary model separates the dense sample region and the sparse sample region;
• running a cache architecture simulation using the sample set to obtain simulation data;
• fitting a multivariate model to the simulation data;
• and
• fitting a constrained multivariate model to the multivariate model using an isonotizing algorithm.

A computer-readable medium having recorded thereon instructions executable by a processor to perform a cache architecture simulation, the instructions for:

• obtaining a first sample set for the cache architecture using a non-stationary Gaussian field model;
• performing a cache architecture simulation using the first sample set to produce a first set of simulation data;
• and
• fitting a first multivariate model to the first set of simulation data.

A computer system for cache architecture simulation, comprising:

• a processor;
• a memory;
• a display device;
• and
• software instructions stored in the memory for enabling the computer system under control of the processor, to perform:
• obtaining a first sample set for the cache architecture using a non-stationary Gaussian field model;
• performing a cache architecture simulation using the first sample set to produce a first set of simulation data;
• and
• fitting a first multivariate model to the first set of simulation data.

An apparatus for a cache architecture simulation, comprising:

• means for performing a uniform cache architecture experiment on a sample space to obtain preliminary data, wherein the preliminary data comprises a dense sample region;
• means for obtaining the first sample set using the non-stationary Gaussian field model, wherein the non-stationary Gaussian field model uses the preliminary data to obtain the first sample set;
• means for performing a cache architecture simulation using the first sample set to produce a first set of simulation data;
• means for fitting a first multivariate model to the first set of simulation data;
• and means for fitting a first constrained multivariate model to the first multivariate model using an isonotizing algorithm.

An apparatus for cache architecture simulation comprising;

• means for determining a sample space;
• means for determining a dense sample region and a sparse sample region in the sample space;
• means for generating a sample set using a non-stationary Gaussian field model, wherein the non-stationary model separates the dense sample region and the sparse sample region;
• means for running a cache architecture simulation using the sample set to obtain simulation data;
• means for fitting a multivariate model to the simulation data;
• and
• means for fitting a constrained multivariate model to the multivariate model using an isonotizing algorithm.