Hyak

Hyak is UW's high performance computing resource. Research Computing Club maintains a large account on Hyak's Klone cluster. This includes standard compute, GPU, and large memory nodes. Hyak is well suited to 'traditional' job-based computing workflows such as molecular dynamics, DFT, and machine learning model training. Most users should start with Hyak before considering cloud resources. Information on using Hyak can be found in the [documentation].

Hyak Trainings

Find step-by-step instructions and tutorials for common tools in Hyak's [Documentation]

Hyak training sessions are recorded and available through the [Research Computing Training Playlist on UW-IT's YouTube channel]

[Subscribe to subscribe to the Hyak mailing list] to receive monthly training announcements.

RCC Resources

RCC has the following resources on Klone. Thanks to a block funding allocation from the UW Student Technology Fee committee, we are constantly expanding our klone slice footprint.

1. Partition	2. Number of Slices	3. CPU Cores (per slice)	4. Memory (per slice)	5. GPU Cards (per slice)	6. Number of Nodes	7. Memory (per Node)	8. GPUs
compute	16	16	192 GB	N/A	16	192GB	N/A
cpu-g2	6	32	256 GB	N/A	1	1536 GB	N/A
cpu-g2-mem2x	6	32	512 GB	N/A	1	3072 GB	N/A
gpu-2080ti	1	40	384 GB	8	1	384 GB	8
gpu-l40	4	32	384 GB	8	1	1536 GB	8
gpu-l40s	4	32	384 GB	8	1	1536 GB	8
compute-hugemem	3	40	750 GB	N/A	3	750 GB	N/A
interactive	1	1	192 GB	N/A	1	192 GB	N/A