We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.AR

Change to browse by:

cs

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Hardware Architecture

Title: Buddy Compression: Enabling Larger Memory for Deep Learning and HPC Workloads on GPUs

Abstract: GPUs offer orders-of-magnitude higher memory bandwidth than traditional CPU-only systems. However, GPU device memory tends to be relatively small and the memory capacity can not be increased by the user. This paper describes Buddy Compression, a scheme to increase both the effective GPU memory capacity and bandwidth while avoiding the downsides of conventional memory-expanding strategies. Buddy Compression compresses GPU memory, splitting each compressed memory entry between high-speed device memory and a slower-but-larger disaggregated memory pool (or system memory). Highly-compressible memory entries can thus be accessed completely from device memory, while incompressible entries source their data using both on and off-device accesses. Increasing the effective GPU memory capacity enables us to run larger-memory-footprint HPC workloads and larger batch-sizes or models for DL workloads than current memory capacities would allow. We show that our solution achieves an average compression ratio of 2.2x on HPC workloads and 1.5x on DL workloads, with a slowdown of just 1~2%.
Subjects: Hardware Architecture (cs.AR)
Cite as: arXiv:1903.02596 [cs.AR]
  (or arXiv:1903.02596v2 [cs.AR] for this version)

Submission history

From: Esha Choukse [view email]
[v1] Wed, 6 Mar 2019 19:55:19 GMT (7124kb,D)
[v2] Mon, 15 Apr 2019 23:28:48 GMT (2473kb,D)

Link back to: arXiv, form interface, contact.