From 68e3aa7c5e8948e5957596de79492ba58a609277 Mon Sep 17 00:00:00 2001 From: Joseph Chu <122298491+cfjchu@users.noreply.github.com> Date: Thu, 10 Oct 2024 15:13:47 -0700 Subject: [PATCH] #13693: Fixes to tech_report/Programming Mesh of Devices (#13714) --- .../Programming Mesh of Devices with TT-NN.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/tech_reports/Programming Mesh of Devices/Programming Mesh of Devices with TT-NN.md b/tech_reports/Programming Mesh of Devices/Programming Mesh of Devices with TT-NN.md index 0ae6064b4fa..e507bbccf8f 100644 --- a/tech_reports/Programming Mesh of Devices/Programming Mesh of Devices with TT-NN.md +++ b/tech_reports/Programming Mesh of Devices/Programming Mesh of Devices with TT-NN.md @@ -289,7 +289,7 @@ Let's see an example of how to use the Ring All-Gather operation: ```py import ttnn -mesh_device = ttnn.open_mesh_device(ttnn.MeshShape(2, 4)) +mesh_device = ttnn.open_mesh_device(ttnn.MeshShape(2, 4), mesh_type=ttnn.MeshType.Ring) # Construct test tensor of data; 8 chunks of 32x32 torch_tensor = torch.rand((1,1,32,128), dtype=torch.bfloat16) @@ -318,7 +318,7 @@ This time, we'll issue the CCL Line All-Gather operation along the cluster y-axi ```py import ttnn -mesh_device = ttnn.open_mesh_device(ttnn.MeshShape(2, 4)) +mesh_device = ttnn.open_mesh_device(ttnn.MeshShape(2, 4), mesh_type=ttnn.MeshType.Ring) # Construct test tensor of data; 8 chunks of 32x32 torch_tensor = torch.rand((1,1,32,128), dtype=torch.bfloat16) @@ -516,7 +516,7 @@ torch_hidden_states = (torch.rand(batch_size, 1, sequence_length, config.hidden_ torch_output = model.forward(torch_hidden_states) # Device Initialization -mesh_device = ttnn.open_mesh_device(ttnn.MeshShape(2,4)) +mesh_device = ttnn.open_mesh_device(ttnn.MeshShape(2,4), mesh_type=ttnn.MeshType.Ring) # Initialize input activations on all devices in the mesh # Alternatively, we can shard the input activations on the height dimension and