Add more GPU architectures support (#112)

* Add more GPU architectures support

* Update layout.py

* Optimize performance, Add SM90 support, Add 1D2D SM100 support

* Add fmtlib submodule at commit 553ec11

---------

Co-authored-by: fzyzcjy <5236035+fzyzcjy@users.noreply.github.com>

deep_gemm/testing/numeric.py

@@ -0,0 +1,19 @@
+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()
+    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