GPU

Related topics

• GPU is part of Cloud Run From the official documentation: “The GPU feature offers on-demand availability with no reservations needed, similar to the way on-demand [CPU](/run/docs/configuring/jobs/cpu) and on-demand [memory](/run/docs/configuring/jobs/memory-limits) work in Cloud Run.”
• GPU is part of Compute Engine From the official documentation: “Compute Engine offers the option of adding one or more GPUs to your virtual machine instances.”
• GPU works with reservation From the official documentation: “For all GPUs and most Local SSD disks, the total numbers and types must match between your commitment and attached reservations.”
• GPU works with Spot VMs From the official documentation: “You can also use GPUs with [Spot VMs](/kubernetes-engine/docs/concepts/spot-vms) if your workloads can tolerate frequent node disruptions.”
• GPU works with vLLM From the official documentation: “Demonstrates how to run machine learning inference by using vLLM and GPUs.”
• Cloud Monitoring works with GKE From the official documentation: “Enable Cloud Monitoring on GKE Clusters”
• Compute Engine works with service account From the official documentation: “A common method is to attach a service account to a resource like a Compute Engine instance.”
• Cloud Run depends on service account From the official documentation: “The check ensures that the service account that runs the container has permission to access these secrets.”
• Compute Engine is an alternative to GKE From the official documentation: “You can also deploy a third-party model to either Vertex AI Model Garden or self-host on GKE or Compute Engine.”
• Cloud Run works with Google Cloud Storage From the official documentation: “The Cloud Run function needs a service account with permissions to write to GCS bucket.”
• Cloud TPU works with GKE From the official documentation: “Running this SFT workload on Cloud TPU and GKE involves the following components:”
• Cloud Run is an alternative to GKE From the official documentation: “For your containerized websites, you can also deploy them to [Cloud Run](/run) in addition to [using GKE](#kubernetes-engine).”
• Cloud Run works with Google Kubernetes Engine From the official documentation: “Connecting Cloud Run services to Cloud Service Mesh lets you securely route traffic to workloads on GKE.”
• Cloud TPU works with Compute Engine From the official documentation: “You can use TPUs through Compute Engine, Google Kubernetes Engine, and Gemini Enterprise Agent Platform.”
• Compute Engine is an alternative to Google Kubernetes Engine From the official documentation: “You can install MongoDB on a Compute Engine instance or you can deploy MongoDB on Google Kubernetes Engine.”
• GKE depends on Compute Engine From the official documentation: “For example, GKE deploys clusters on Compute Engine VMs.”
• Cloud Monitoring works with Cloud Run From the official documentation: “Cloud Monitoring records metrics regarding the execution of Cloud Run.”
• Cloud Monitoring works with Google Kubernetes Engine From the official documentation: “When Google Kubernetes Engine (GKE) [system metrics](/kubernetes-engine/docs/how-to/configure-metrics#system-metrics) are enabled, Cloud Monitoring collects the following metrics.”
• Cloud Run works with GKE From the official documentation: “Leverage Google Kubernetes Engine (GKE) and Cloud Run to deploy applications to the Google Cloud, optimizing costs, performance, and management overhead.”
• Compute Engine works with Google Cloud Storage From the official documentation: “The usage export bucket is a Google Cloud Storage bucket where Compute Engine usage reports are stored.”
• GKE works with vLLM From the official documentation: “GKE supports these quantization options with vLLM:”
• Spot VMs is part of Compute Engine From the official documentation: “Spot VMs are virtual machine (VM) instances that are excess Compute Engine capacity.”
• Cloud Monitoring works with Cloud TPU From the official documentation: “This guide explains how to use [Cloud Monitoring](https://cloud.google.com/monitoring) to monitor your TPU VMs. Cloud Monitoring automatically collects [metrics](https://cloud.google.com/monitoring/api/metrics_gcp) and [logs](https://cloud.google.com/logging) from your TPU and…”
• Cloud Monitoring works with Compute Engine From the official documentation: “Many Google services like Compute Engine send telemetry to Cloud Monitoring.”
• Cloud Monitoring works with vLLM From the official documentation: “To observe your model's performance, you can use the vLLM dashboard integration in Cloud Monitoring.”
• Cloud Run works with Compute Engine From the official documentation: “This lets Cloud Run clients use this backend service route to send requests directly to Compute Engine instances.”
• Cloud Run is an alternative to Google Kubernetes Engine From the official documentation: “**Agent and tool runtimes**: hosting and executing agents or MCP servers on runtimes such as Agent Runtime on Gemini Enterprise Agent Platform, Cloud Run, or Google Kubernetes Engine (GKE) incurs standard compute and runtime charges for those services.”
• Cloud Run works with service account From the official documentation: “Map a custom Service Account to a Cloud Run service”
• Cloud TPU works with Google Cloud Storage From the official documentation: “add-gcs - ensure Cloud TPU is authorized for Cloud Storage”
• Cloud TPU works with Google Kubernetes Engine From the official documentation: “You can use TPUs through Compute Engine, Google Kubernetes Engine, and Gemini Enterprise Agent Platform.”
• Cloud TPU works with Spot VMs From the official documentation: “In this block, the `spot: true` setting enables Spot VMs for the Cloud TPU node pool.”
• Cloud TPU works with vLLM From the official documentation: “Cloud TPU supports large language model (LLM) inference through two open-source serving libraries: [vLLM](https://docs.vllm.ai/en/latest/) and [SGLang](https://docs.sglang.io/).”
• Compute Engine works with Google Kubernetes Engine From the official documentation: “Compute Engine integrates with Google Cloud technologies such as Cloud Storage, Google Kubernetes Engine, and BigQuery, to extend beyond the basic computational capability to create more complex and sophisticated applications.”
• Compute Engine works with reservation From the official documentation: “After you attach your reservations, Compute Engine automatically populates the fields in the **Commitment details** section to match the numbers and types of resources from your attached reservations.”
• Compute Engine depends on service account From the official documentation: “The Service Account is used by Compute Engine to access other Google APIs.”
• Google Cloud Storage works with vLLM From the official documentation: “The prebuilt vLLM container can download models from Hugging Face or from Cloud Storage.”
• Google Kubernetes Engine works with reservation From the official documentation: “In the **Google Cloud Services** section, specify whether to share this reservation with other services like Google Kubernetes Engine or Vertex AI.”
• Google Kubernetes Engine works with vLLM From the official documentation: “Deploy a vLLM container to your GKE cluster.”
• reservation is part of Compute Engine From the official documentation: “If Google Cloud approves your request, then Compute Engine creates reservations with your requested resources at your specified date and time.”
• service account depends on Cloud Run From the official documentation: “The service account that you created in the previous step needs permission to use Cloud Run so that the BigQuery remote function can use the Cloud Run functions function.”
• service account depends on Google Cloud Storage From the official documentation: “Google SecOps uses a unique service account to read data from your GCS bucket. You must grant this service account access to your bucket.”