Resources

CPU

By default, all Snakemake jobs will run on a machine with 4 CPUs available. To modify the number of CPUs allocated to the job, use the resources directive of the Snakefile rule as follows:

rule <rule_name>:
    ...
    resources:
        cpus=8
    ...

Snakemake requires that the user specify the number of cores available to the workflow via the --cores command line argument. To define the number of cores available to the job, set the cores keyword in your SnakemakeMetadata. The cores field will default to 4 if there is no value provided.

# latch_metadata/__init__.py
from latch.types.metadata import SnakemakeMetadata, LatchAuthor
from latch.types.directory import LatchDir

from .parameters import generated_parameters

SnakemakeMetadata(
    output_dir=LatchDir("latch:///your_output_directory"),
    display_name="Your Workflow Name",
    author=LatchAuthor(
        name="Your Name",
    ),
    parameters=generated_parameters,
    cores=8, # updated
)

Memory

By default, all Snakemake jobs will run on a machine with 8 GB of RAM. To modify the amount of memory allocated to the job, use the resources directive of the Snakefile rule. For example, to allocate 32 GB of RAM to a task:

rule <rule_name>:
    ...
    resources:
        mem_mb=34360
    ...

GPU

To run a Snakemake job on a GPU instance, modify the resources directive of the Snakefile rule. For example:

rule <rule_name>:
    ...
    resources:
        nvidia_gpu=1
    ...

GPU tasks will execute as either a small_gpu_task or large_gpu_task as defined here. To request a large GPU instance, add CPU and memory requirements as follows:

rule <rule_name>:
    ...
    resources:
        nvidia_gpu=1
        cpus=8
        mem_mb=33286
    ...

Limitations:

Using the container directive inside GPU instances is currently not supported. Use conda or add runtime dependencies to your Dockerfile to use GPUs.
Multi-GPU instances are currently not supported. The JIT workflow will fail if more than 1 GPU is requested.

Workflows & SDK

Developer Tools

Workflow User Interface

Testing

Latch Console Dashboard

Ready-to-Use Workflows

Resources

Resources

CPU

Memory

GPU

​Resources

​CPU

​Memory

​GPU

Resources

CPU

Memory

GPU