Add volta runner - p3.8xlarge. Remove unused windows memory runner (#6921)

Add Volta runner to test sm70 architecture with CUDA 12.8. Required for
testing of : pytorch/pytorch#157517
Remove windows memory runner that was used to test Windows CUDA 12.9
regression and not required anymore

Author: atalman

.github/canary-scale-config.yml

@@ -67,6 +67,11 @@ runner_types:
     instance_type: c5.12xlarge
     is_ephemeral: true
     os: linux
+  c.linux.p3.8xlarge.nvidia.gpu:
+    disk_size: 150
+    instance_type: p3.8xlarge
+    is_ephemeral: true
+    os: linux
   c.linux.16xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g3.16xlarge
@@ -214,22 +219,6 @@ runner_types:
     variants:
       wincanary:
         ami: Windows 2019 GHA CI - 20250526202723
-  c.windows.16xlarge.memory:
-    disk_size: 256
-    instance_type: r5n.16xlarge
-    is_ephemeral: true
-    os: windows
-    variants:
-      wincanary:
-        ami: Windows 2019 GHA CI - 20250526202723
-  c.windows.16xlarge.memory.nonephemeral:
-    disk_size: 256
-    instance_type: r5n.16xlarge
-    is_ephemeral: true
-    os: windows
-    variants:
-      wincanary:
-        ami: Windows 2019 GHA CI - 20250526202723
   c.windows.8xlarge.nvidia.gpu:
     disk_size: 256
     instance_type: p3.2xlarge

.github/lf-canary-scale-config.yml

@@ -67,6 +67,11 @@ runner_types:
     instance_type: c5.12xlarge
     is_ephemeral: true
     os: linux
+  lf.c.linux.p3.8xlarge.nvidia.gpu:
+    disk_size: 150
+    instance_type: p3.8xlarge
+    is_ephemeral: true
+    os: linux
   lf.c.linux.16xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g3.16xlarge
@@ -214,22 +219,6 @@ runner_types:
     variants:
       wincanary:
         ami: Windows 2019 GHA CI - 20250526202723
-  lf.c.windows.16xlarge.memory:
-    disk_size: 256
-    instance_type: r5n.16xlarge
-    is_ephemeral: true
-    os: windows
-    variants:
-      wincanary:
-        ami: Windows 2019 GHA CI - 20250526202723
-  lf.c.windows.16xlarge.memory.nonephemeral:
-    disk_size: 256
-    instance_type: r5n.16xlarge
-    is_ephemeral: true
-    os: windows
-    variants:
-      wincanary:
-        ami: Windows 2019 GHA CI - 20250526202723
   lf.c.windows.8xlarge.nvidia.gpu:
     disk_size: 256
     instance_type: p3.2xlarge

.github/lf-scale-config.yml

@@ -67,6 +67,11 @@ runner_types:
     instance_type: c5.12xlarge
     is_ephemeral: true
     os: linux
+  lf.linux.p3.8xlarge.nvidia.gpu:
+    disk_size: 150
+    instance_type: p3.8xlarge
+    is_ephemeral: true
+    os: linux
   lf.linux.16xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g3.16xlarge
@@ -214,22 +219,6 @@ runner_types:
     variants:
       wincanary:
         ami: Windows 2019 GHA CI - 20250526202723
-  lf.windows.16xlarge.memory:
-    disk_size: 256
-    instance_type: r5n.16xlarge
-    is_ephemeral: true
-    os: windows
-    variants:
-      wincanary:
-        ami: Windows 2019 GHA CI - 20250526202723
-  lf.windows.16xlarge.memory.nonephemeral:
-    disk_size: 256
-    instance_type: r5n.16xlarge
-    is_ephemeral: true
-    os: windows
-    variants:
-      wincanary:
-        ami: Windows 2019 GHA CI - 20250526202723
   lf.windows.8xlarge.nvidia.gpu:
     disk_size: 256
     instance_type: p3.2xlarge

.github/scale-config.yml

@@ -63,6 +63,11 @@ runner_types:
     instance_type: c5.12xlarge
     is_ephemeral: true
     os: linux
+  linux.p3.8xlarge.nvidia.gpu:
+    disk_size: 150
+    instance_type: p3.8xlarge
+    is_ephemeral: true
+    os: linux
   linux.16xlarge.nvidia.gpu:
     disk_size: 150
     instance_type: g3.16xlarge
@@ -210,22 +215,6 @@ runner_types:
     variants:
       wincanary:
         ami: Windows 2019 GHA CI - 20250526202723
-  windows.16xlarge.memory:
-    disk_size: 256
-    instance_type: r5n.16xlarge
-    is_ephemeral: true
-    os: windows
-    variants:
-      wincanary:
-        ami: Windows 2019 GHA CI - 20250526202723
-  windows.16xlarge.memory.nonephemeral:
-    disk_size: 256
-    instance_type: r5n.16xlarge
-    is_ephemeral: true
-    os: windows
-    variants:
-      wincanary:
-        ami: Windows 2019 GHA CI - 20250526202723
   windows.8xlarge.nvidia.gpu:
     disk_size: 256
     instance_type: p3.2xlarge