-
Notifications
You must be signed in to change notification settings - Fork 53
gpu
William Rowell edited this page Sep 9, 2025
·
3 revisions
Starting in workflow version 2.0.0, we have added support for running workflows on GPU-enabled nodes. The first task to take advantage of this is the deepvariant_call_variants task in the DeepVariant workflow, which can use 1 GPU. To run the DeepVariant workflow on a GPU-enabled node, you will need to provide some additional configuration in your inputs JSON file.
| Type | Name | Description | Notes |
|---|---|---|---|
| Boolean | gpu | Use GPUs. | default = false
|
| String | gpuType | Type of GPU/Accelerator to use. | This value will depend on your backend configuration. |
| Backend | GPU Type | Notes |
|---|---|---|
| AWS-HealthOmics | ["nvidia-tesla-a10g", "nvidia-tesla-t4", "nvidia-tesla-t4-a10g"] |
GPU availability varies by zone. |
| Azure | GPU support not yet implemented, but monitoring microsoft/ga4gh-tes#717. | |
| GCP | ["nvidia-tesla-t4", "nvidia-tesla-v100"] |
GPU availability varies by zone. |
| HPC | This will depend on HPC and miniwdl or Cromwell configuration. Reach out to support@pacb.com |