[backend.aws.cpu-big] region = "eu-west-3" image_id = "ami-01c204b15b504da60" instance_type = "m6i.32xlarge" user = "ubuntu" [backend.aws.cpu-big_fallback] region = "us-east-1" image_id = "ami-0c6e54f3be8d03234" instance_type = "m6i.32xlarge" user = "ubuntu" [backend.aws.cpu-small] region = "eu-west-3" image_id = "ami-01c204b15b504da60" instance_type = "m6i.4xlarge" user = "ubuntu" [backend.aws.bench] region = "eu-west-1" image_id = "ami-067f3893108afe8cd" instance_type = "hpc7a.96xlarge" user = "ubuntu" [backend.aws.bench-hpc8] region = "eu-north-1" image_id = "ami-0d7f32ee367bc2a69" instance_type = "hpc8a.96xlarge" user = "ubuntu" # Profile used to build CUDA code without the need to get p-like instance. [backend.aws.gpu-build] region = "us-east-1" image_id = "ami-0250f541bc1358116" instance_type = "m6i.4xlarge" user = "ubuntu" [backend.aws.single-h100] region = "us-east-1" image_id = "ami-0250f541bc1358116" instance_type = "p5.4xlarge" user = "ubuntu" [backend.aws.single-h100_1stfallback] region = "us-east-2" image_id = "ami-0c924a8e0a210a709" instance_type = "p5.4xlarge" user = "ubuntu" [backend.aws.single-h100_2ndfallback] region = "us-west-1" image_id = "ami-0fe3273232bc2fb8a" instance_type = "p5.4xlarge" user = "ubuntu" [backend.aws.single-h100_3rdfallback] region = "us-west-2" image_id = "ami-057682b0cfedeaf20" instance_type = "p5.4xlarge" user = "ubuntu" [backend.aws.4-l40] region = "us-east-1" image_id = "ami-0250f541bc1358116" instance_type = "g6e.24xlarge" user = "ubuntu" fallbacks = [ "hyperstack.4-l40", "terraform.scaleway-4-l40" ] [backend.aws.small-gpu] region = "us-east-1" image_id = "ami-0250f541bc1358116" instance_type = "g6.2xlarge" user = "ubuntu" [backend.hyperstack.gpu-test] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-L40x1" user = "ubuntu" fallbacks = [ "terraform.scaleway-small-gpu", "hyperstack.rtx-a6000" ] [backend.hyperstack.gpu-test-sanitizer] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-L40x1" user = "ubuntu" fallbacks = [ "terraform.scaleway-small-gpu-sanitizer" ] [backend.hyperstack.single-h100] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-H100x1" user = "ubuntu" # Primary target of gpu_protocol_stress_h100_tests: one H100 PCIe, falling back to a single # Scaleway H100 and then to a 2 GPU Scaleway instance. The fallbacks deliberately name the # profiles that carry no fallbacks of their own, so the chain stays exactly these three. [backend.hyperstack.single-h100-protocol-stress] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-H100x1" user = "ubuntu" fallbacks = [ "terraform.scaleway-single-h100", "terraform.scaleway-2-h100-sxm5", ] [backend.hyperstack.2-h100] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-H100x2" user = "ubuntu" [backend.hyperstack.4-h100] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-H100x4" user = "ubuntu" [backend.hyperstack.multi-h100] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-H100x8" user = "ubuntu" [backend.hyperstack.multi-h100-nvlink] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-H100x8-NVLink" user = "ubuntu" [backend.hyperstack.multi-h100-sxm5] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-H100-SXM5x8" user = "ubuntu" [backend.hyperstack.multi-h100-sxm5_fallback] environment_name = "us-1" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-H100-SXM5x8" user = "ubuntu" [backend.hyperstack.multi-a100-nvlink] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-A100x8-NVLink" user = "ubuntu" [backend.hyperstack.l40] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-L40x1" user = "ubuntu" [backend.hyperstack.l40_fallback] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-RTX-A6000x1" user = "ubuntu" # Fallback target only. [backend.hyperstack.rtx-a6000] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-RTX-A6000x1" user = "ubuntu" [backend.hyperstack.4-l40] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-L40x4" user = "ubuntu" [backend.hyperstack.4-l40_fallback] environment_name = "norway" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-RTX-A4000x4" user = "ubuntu" [backend.hyperstack.8-l40] environment_name = "canada" image_name = "Ubuntu Server 22.04 LTS R570 CUDA 12.8" flavor_name = "n3-L40x8" user = "ubuntu" [backend.terraform.scaleway-multi-h100-sxm5] # Due to a bug that can't be figured out on SXM-8 on scaleway, we need to use a dedicated script # for now to get the instance to register a github actions runner script_path = "ci/terraform_scripts/scaleway/terraform_sxm_x8.tf" instance_type = "H100-SXM-8-80G" user = "ubuntu" [backend.terraform.scaleway-4-h100-sxm5] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "H100-SXM-4-80G" user = "ubuntu" [backend.terraform.scaleway-2-h100-sxm5] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "H100-SXM-2-80G" user = "ubuntu" [backend.terraform.scaleway-2-h100-sxm5-tests] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "H100-SXM-2-80G" user = "ubuntu" fallbacks = [ "hyperstack.2-h100", "terraform.scaleway-4-h100-sxm5", "hyperstack.4-h100", ] [backend.terraform.scaleway-single-h100] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "H100-1-80G" user = "ubuntu" [backend.terraform.scaleway-single-h100_fallback] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "H100-SXM-2-80G" user = "ubuntu" [backend.terraform.scaleway-single-h100-tests] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "H100-1-80G" user = "ubuntu" fallbacks = [ "hyperstack.single-h100", ] [backend.terraform.scaleway-4-l40] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L40S-4-48G" user = "ubuntu" fallbacks = [ "hyperstack.4-l40", "aws.4-l40" ] [backend.terraform.scaleway-4-l40-tests] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L40S-4-48G" user = "ubuntu" fallbacks = [ "terraform.scaleway-2-l40", "terraform.scaleway-4-l4", "terraform.scaleway-2-l4", "hyperstack.4-l40", ] [backend.terraform.nebius-2-l40] script_path = "ci/terraform_scripts/nebius/terraform_l40.tf" instance_type = "2gpu-64vcpu-384gb" hardware_name = "nebius-L40" user = "ubuntu" [backend.terraform.generic-4-l4orl40-tests] script_path = "ci/terraform_scripts/nebius/terraform_l40.tf" instance_type = "4gpu-128vcpu-768gb" hardware_name = "nebius-L40" user = "ubuntu" fallbacks = [ "terraform.nebius-2-l40", "terraform.scaleway-2-l40", "terraform.scaleway-4-l4", "terraform.scaleway-2-l4", "hyperstack.4-l40", ] [backend.terraform.scaleway-2-l40] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L40S-2-48G" user = "ubuntu" [backend.terraform.scaleway-1-l4] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L4-1-24G" user = "ubuntu" [backend.terraform.scaleway-4-l4] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L4-4-24G" user = "ubuntu" [backend.terraform.scaleway-2-l4] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L4-2-24G" user = "ubuntu" [backend.terraform.scaleway-small-gpu] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L40S-1-48G" user = "ubuntu" fallbacks = [ "hyperstack.gpu-test", "terraform.scaleway-2-l4" ] [backend.terraform.scaleway-small-gpu-noise-check] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L40S-1-48G" user = "ubuntu" fallbacks = [ "hyperstack.gpu-test", "hyperstack.rtx-a6000" ] [backend.terraform.scaleway-small-gpu-classic-tests] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L40S-1-48G" user = "ubuntu" fallbacks = [ "terraform.scaleway-1-l4", "terraform.scaleway-2-l40", "hyperstack.gpu-test", ] [backend.terraform.scaleway-small-gpu-sanitizer] script_path = "ci/terraform_scripts/scaleway/terraform.tf" instance_type = "L40S-1-48G" user = "ubuntu" fallbacks = [ "hyperstack.gpu-test-sanitizer" ] [backend.terraform.nebius-single-h100] script_path = "ci/terraform_scripts/nebius/terraform_h100.tf" instance_type = "1gpu-16vcpu-200gb" # Nebius presets do not name the GPU, this is the name reported in benchmark results hardware_name = "nebius-H100-SXM-1-80G" user = "ubuntu" [backend.terraform.nebius-multi-h100-sxm5] script_path = "ci/terraform_scripts/nebius/terraform_h100.tf" instance_type = "8gpu-128vcpu-1600gb" hardware_name = "nebius-H100-SXM-8-80G" user = "ubuntu" [backend.terraform.generic-1or2-h100] script_path = "ci/terraform_scripts/nebius/terraform_h100.tf" instance_type = "1gpu-16vcpu-200gb" hardware_name = "nebius-H100-SXM-1-80G" user = "ubuntu" fallbacks = [ "terraform.scaleway-2-h100-sxm5", "terraform.scaleway-single-h100", "hyperstack.single-h100", ]