Tariq receives his best paper award at EuroPar'2020

PhD Candidate Tariq Alturkestani gives his talk at the virtual EuroPar'2020 conference in Warsaw, Poland.

 

Maximizing I/O Bandwidth for Out-of-Core HPC Applications on Heterogeneous Large-Scale Systems

 

Best Paper Presentation by Tariq Alturkestani

PhD Student, Computer Science, KAUST

Thursday, Aug 27, 3:30pm - 4:00pm (AST)

https://zoom.us/j/99947879910

 

Tariq will present his best paper at EuroPar'2020 at the virtual EuroPar'2020 conference in Warsaw, Poland.

Warsaw, Poland

 

The reverse time migration (RTM) method is critical in seismic imaging for oil and gas industries. He demonstrates the effectiveness of his Multilayer Buffer System (MLBS) framework on Shaheen-2 (using 2048 compute nodes) and Summit supercomputers (using 2048 compute nodes equipped with a total of 12288 GPUs. MLBS achieves up to 2.5X and 1.4X performance speedups compared to the reference RTM implementation for a large 3D solution grid on Shaheen-2 and Summit, respectively.

Shaheen and Summit Supercomputers

 

The paper is available online at Springer LNCS. The talk is available on YouTube. Live Q&A session on Zoom following the talk. Questions can also be asked on Slack Channel.