cwpearson/hugo-cwpearson
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

stencil paper revision

Browse Source
This commit is contained in:
Carl Pearson
2020-03-13 17:31:55 -06:00
parent 309b8b7fb6
commit d924f2963b
1 changed files with 4 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
content/publication/2020522_pearson_iwapt/index.md
View File

@@ -26,9 +26,10 @@ abstract = """
High-performance distributed computing systems increasingly feature nodes that have multiple CPU sockets and multiple GPUs.
The communication bandwidth between these components is non-uniform.
Furthermore, these systems can expose different communication capabilities between these components.
For communication-heavy applications, optimally using these capabilities is challenging and essential for performance.
This work presents approaches for automatic data placement and communication implementation for 3D stencil codes on multi-GPU nodes with non-homogeneous communication performance and capabilities.
Benchmarking results in the Summit system show that choices in placement can result in a 20% improvement in single-node exchange, and communication specialization canyield a further 6x improvement in exchange time in a single node, and a 16% improvement at 1536 GPUs"""
For communication-heavy applications, optimally using these capabilities is challenging and essential for performance.
Bespoke codes with optimized communication may be non-portable across run-time/software/hardware configurations, and existing stencil frameworks neglect optimized communication.
This work presents node-aware approaches for automatic data placement and communication implementation for 3D stencil codes on multi-GPU nodes with non-homogeneous communication performance and capabilities.
Benchmarking results in the Summit system show that choices in placement can result in a 20% improvement in single-node exchange, and communication specialization can yield a further 6x improvement in exchange time in a single node, and a 16% improvement at 1536 GPUs."""
abstract_short = ""