Skip to content
Snippets Groups Projects

Add single quotes and warning message.

Merged Adam Caprez requested to merge gpu-syntax into master
1 file
+ 13
9
Compare changes
  • Side-by-side
  • Inline
@@ -5,7 +5,7 @@ description = "How to submit GPU (CUDA/OpenACC) jobs on HCC resources."
@@ -5,7 +5,7 @@ description = "How to submit GPU (CUDA/OpenACC) jobs on HCC resources."
### Available GPUs
### Available GPUs
Crane has four types of GPUs available in the **gpu** partition. The
Crane has four types of GPUs available in the `gpu` partition. The
type of GPU is configured as a SLURM feature, so you can specify a type
type of GPU is configured as a SLURM feature, so you can specify a type
of GPU in your job resource requirements if necessary.
of GPU in your job resource requirements if necessary.
@@ -46,6 +46,13 @@ a feature. To run on K40 GPUs for example:
@@ -46,6 +46,13 @@ a feature. To run on K40 GPUs for example:
--partition=gpu --gres=gpu --constraint=gpu_k40
--partition=gpu --gres=gpu --constraint=gpu_k40
{{< /highlight >}}
{{< /highlight >}}
 
{{% notice info %}}
 
You may request multiple GPUs by changing the` --gres` value to
 
-`-gres=gpu:2`. Note that this value is **per node**. For example,
 
`--nodes=2 --gres=gpu:2 `will request 2 nodes with 2 GPUs each, for a
 
total of 4 GPUs.
 
{{% /notice %}}
 
The GPU memory feature may be used to specify a GPU RAM amount either independent of architecture, or in combination with it.
The GPU memory feature may be used to specify a GPU RAM amount either independent of architecture, or in combination with it.
For example, using
For example, using
@@ -54,20 +61,17 @@ For example, using
@@ -54,20 +61,17 @@ For example, using
{{< /highlight >}}
{{< /highlight >}}
will request a GPU with 16GB of RAM, independent of the type of card (K20, K40, P100, etc.). You may also
will request a GPU with 16GB of RAM, independent of the type of card (K20, K40, P100, etc.). You may also
request both a GPU type _and_ memory amount using the `&` operator. For example,
request both a GPU type _and_ memory amount using the `&` operator (single quotes are used because `&` is a special character).
 
For example,
{{< highlight batch >}}
{{< highlight batch >}}
--partition=gpu --gres=gpu --constraint=gpu_32gb&gpu_v100
--partition=gpu --gres=gpu --constraint='gpu_32gb&gpu_v100'
{{< /highlight >}}
{{< /highlight >}}
will request a V100 GPU with 32GB RAM.
will request a V100 GPU with 32GB RAM.
{{% notice warning %}}
{{% notice info %}}
You must verify the GPU type and memory combination is valid based on the [available GPU types.]({{< relref "submitting_cuda_or_openacc_jobs/#available-gpus" >}}). Requesting a nonexistent combination will cause your job to be rejected with a `Requested node configuration is not available` error.
You may request multiple GPUs by changing the` --gres` value to
-`-gres=gpu:2`. Note that this value is **per node**. For example,
`--nodes=2 --gres=gpu:2 `will request 2 nodes with 2 GPUs each, for a
total of 4 GPUs.
{{% /notice %}}
{{% /notice %}}
### Compiling
### Compiling
Loading