Embed DeepGEMM source (not submodule) for SM100 raw CUDA GEMM primitives

third_party/DeepGEMM/deep_gemm/testing/numeric.py

@@ -0,0 +1,21 @@
+import torch
+from typing import Iterable
+
+
+def calc_diff(x: torch.Tensor, y: torch.Tensor):
+    x, y = x.double(), y.double()
+    denominator = (x * x + y * y).sum()
+    if denominator == 0:    # Which means that all elements in x and y are 0
+        return 0.0
+    sim = 2 * (x * y).sum() / denominator
+    return 1 - sim
+
+
+def count_bytes(*tensors):
+    total = 0
+    for t in tensors:
+        if isinstance(t, (tuple, list)):
+            total += count_bytes(*t)
+        elif t is not None:
+            total += t.numel() * t.element_size()
+    return total