SDPA tutorial requires newer CUDA, so host it on an A10G

.jenkins/get_files_to_run.py

@@ -46,6 +46,9 @@ def add_to_shard(i, filename):
     needs_gpu_nvidia_medium = list(
         filter(lambda x: get_needs_machine(x) == "gpu.nvidia.large", all_files,)
     )
+    needs_a10g = list(
+        filter(lambda x: get_needs_machine(x) == "linux.g5.4xlarge.nvidia.gpu", all_files,)
+    )
     for filename in needs_gpu_nvidia_small_multi:
         # currently, the only job that uses gpu.nvidia.small.multi is the 0th worker,
         # so we'll add all the jobs that need this machine to the 0th worker
@@ -56,6 +59,11 @@ def add_to_shard(i, filename):
         # so we'll add all the jobs that need this machine to the 1st worker
         add_to_shard(1, filename)
         all_other_files.remove(filename)
+    for filename in needs_a10g:
+        # currently, workers 2-5th use linux.g5.4xlarge.nvidia.gpu, so, arbitrarily,
+        # we'll add all the jobs that need this machine to the 5th worker
+        add_to_shard(5, filename)
+        all_other_files.remove(filename)
 
     sorted_files = sorted(all_other_files, key=get_duration, reverse=True,)
 

.jenkins/metadata.json

@@ -30,5 +30,8 @@
   },
   "intermediate_source/torch_compile_tutorial.py": {
     "needs": "gpu.nvidia.large"
+  },
+  "intermediate_source/scaled_dot_product_attention_tutorial.py": {
+    "needs": "linux.g5.4xlarge.nvidia.gpu"
   }
 }