Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Fri, 13 Oct 2023 10:28:38 GMT

Current Revision posted to Wiki by Brian John on 10/13/2023 10:28:38 AM

Language Weaver Edge is deployed in Kubernetes as pods, running on standard CPU nodes. Translation Engine pods & Training Engine pods could optionally be deployed in Kubernetes on nodes with Nvidia Tesla GPUs. Helm charts provided by RWS will try to deploy Training Engines & Translation Engines on nodes with Nvidia Tesla GPUs if detected & specified in the “values.yaml”. No pods are scheduled on Kubernetes GPU nodes by default, unless specified in the “values.yaml”.

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia Tesla GPUs.

Supported Nvidia Tesla Drivers

https://confluence.sdl.com/display/AIP/About+NVidia+drivers

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Tue, 10 Oct 2023 16:16:57 GMT

Revision 10 posted to Wiki by Brian John on 10/10/2023 4:16:57 PM

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia Tesla GPUs.

Supported Nvidia Tesla Drivers

https://confluence.sdl.com/display/AIP/About+NVidia+drivers

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Tue, 10 Oct 2023 16:16:21 GMT

Revision 9 posted to Wiki by Brian John on 10/10/2023 4:16:21 PM

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia Tesla GPUs.

Supported Nvidia Tesla Drivers

https://confluence.sdl.com/display/AIP/About+NVidia+drivers

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Tue, 10 Oct 2023 16:16:09 GMT

Revision 8 posted to Wiki by Brian John on 10/10/2023 4:16:09 PM

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia Tesla GPUs.

Supported Nvidia Tesla Drivers

https://confluence.sdl.com/display/AIP/About+NVidia+drivers

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Tue, 10 Oct 2023 12:05:20 GMT

Revision 7 posted to Wiki by Brian John on 10/10/2023 12:05:20 PM

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia Tesla GPUs.

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Fri, 06 Oct 2023 10:47:27 GMT

Revision 6 posted to Wiki by Brian John on 10/6/2023 10:47:27 AM

Language Weaver Edge is deployed in Kubernetes as pods, running on standard CPU nodes. Translation Engine pods & Training Engine pods could optionally be deployed in Kubernetes on nodes with Nvidia GPUs. Helm charts provided by RWS will try to deploy Training Engines & Translation Engines on nodes with Nvidia GPUs if detected & specified in the “values.yaml”. No pods are scheduled on Kubernetes GPU nodes by default, unless specified in the “values.yaml”.

For Translation Engines, the cost of deploying on GPU nodes, tends to be comparable with deploying on non-GPU nodes. The additional complexity may not be warranted.

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Training Adaptable Language Pairs on GPUs is cost effective when a Language Pair takes unacceptably long time to complete a training on CPUs.

Advantages

Training Engines could be considerably faster (up to 10x speedup).

Disadvantages

Kubernetes cluster must have GPU nodes.
Auto scaling Translation Engines is not yet supported on GPUs.
The hardware cost of performing translations is usually comparable with CPU deployment.

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia GPUs.

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Fri, 06 Oct 2023 10:47:07 GMT

Revision 5 posted to Wiki by Brian John on 10/6/2023 10:47:07 AM

Language Weaver Edge is deployed in Kubernetes as pods, running on standard CPU nodes. Translation Engine pods & Training Engine pods could optionally be deployed in Kubernetes on nodes with Nvidia GPUs. Helm charts provided by RWS will try to deploy Training Engines & Translation Engines on nodes with Nvidia GPUs if detected & specified in the “values.yaml”. No pods are scheduled on Kubernetes GPU nodes by default, unless specified in the “values.yaml”.

For Translation Engines, the cost of deploying on GPU nodes, tends to be comparable with deploying on non-GPU nodes. The additional complexity may not be warranted.

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Training Adaptable Language Pairs on GPUs is cost effective when a Language Pair takes unacceptably long time to complete a training on CPUs.

Advantages

Training Engines could be considerably faster (up to 10x speedup).

Disadvantages

Kubernetes cluster must have GPU nodes.
Auto scaling Translation Engines is not yet supported on GPUs.
The hardware cost of performing translations is usually comparable with CPU deployment.

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia GPUs.

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Fri, 06 Oct 2023 10:46:02 GMT

Revision 4 posted to Wiki by Brian John on 10/6/2023 10:46:02 AM

Language Weaver Edge is deployed in Kubernetes as pods, running on standard CPU nodes. Translation Engine pods & Training Engine pods could optionally be deployed in Kubernetes on nodes with Nvidia GPUs. Helm charts provided by RWS will try to deploy Training Engines & Translation Engines on nodes with Nvidia GPUs if detected & specified in the “values.yaml”. No pods are scheduled on Kubernetes GPU nodes by default, unless specified in the “values.yaml”.

For Translation Engines, the cost of deploying on GPU nodes, tends to be comparable with deploying on non-GPU nodes. The additional complexity may not be warranted.

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Training Adaptable Language Pairs on GPUs is cost effective when a Language Pair takes unacceptably long time to complete a training on CPUs.

Advantages

Training Engines could be considerably faster (up to 10x speedup).

Disadvantages

Kubernetes cluster must have GPU nodes.
Auto scaling Translation Engines is not yet supported on GPUs.
The hardware cost of performing translations is usually comparable with CPU deployment.

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia GPUs.

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

Brian John — Fri, 06 Oct 2023 10:45:42 GMT

Revision 3 posted to Wiki by Brian John on 10/6/2023 10:45:42 AM

Language Weaver Edge is deployed in Kubernetes as pods, running on standard CPU nodes. Translation Engine pods & Training Engine pods could optionally be deployed in Kubernetes on nodes with Nvidia GPUs. Helm charts provided by RWS will try to deploy Training Engines & Translation Engines on nodes with Nvidia GPUs if detected & specified in the “values.yaml”. No pods are scheduled on Kubernetes GPU nodes by default, unless specified in the “values.yaml”.

For Translation Engines, the cost of deploying on GPU nodes, tends to be comparable with deploying on non-GPU nodes. The additional complexity may not be warranted.

For Training Engines, GPU deployment offers significant speed gains versus CPU deployment (up to 10x faster).

Training Adaptable Language Pairs on GPUs is cost effective when a Language Pair takes unacceptably long time to complete a training on CPUs.

Advantages

Training Engines could be considerably faster (up to 10x speedup).

Disadvantages

Kubernetes cluster must have GPU nodes.
Auto scaling Translation Engines is not yet supported on GPUs.
The hardware cost of performing translations is usually comparable with CPU deployment.

Supported GPU Platforms

The current release of Language Weaver Edge supports only Nvidia GPUs.

Configuration

Adding below under Training Engine or Translation Engine Stateful Set configuration in “values.yaml” will force the helm deployment to use GPU nodes to deploy Training Engine or Translation Engine pods.

gpu: true

Adaptation of Language Pairs in Kubernetes GPU Nodes

RWS Community — Thu, 05 Oct 2023 14:00:11 GMT

Revision 2 posted to Wiki by RWS Community on 10/5/2023 2:00:11 PM

Adaptation of Language Pairs in Kubernetes GPU Nodes