{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

Make sure that you use the m thread option when you

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: r these measurements with a brief explanation of your choice. If you use HPCToolkit, list also the sampling rates used for these measurements. 2. Scaling efficiency: Without using any tools, run the bt.A solver with 1, 2, 4 and 8 threads and record the total operation throughput reported by the benchmark (Mop/s total) for each run. Plot this throughput and contrast it against an ideal scaling throughput. Discuss what you observe. 3. Differential analysis: For this step you have to use HPCToolkit. Collect only time measurements (PAPI_TOT_CYC) for two runs with 2 and 4 threads, respectively. Compute a performance database with the data for thread 0 from each run, as shown in the lecture slides example. Make sure that you use...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online