[mpact][file-formats] add matrix market and extended frostt utils (#66)

* [mpact][file-formats] add matrix market and extended frostt utils

* add mm back

* add benchmark util dep to test

benchmark/python/utils/tensor_generator.py

@@ -75,3 +75,56 @@ def generate_tensor(
 
     result = np.reshape(flat_output, shape).astype(dtype)
     return torch.from_numpy(result)
+
+
+def print_matrix_market_format(tensor: torch.Tensor):
+    """Prints the matrix market format for a sparse matrix.
+
+    Args:
+      tensor: sparse matrix (real type)
+    """
+    if len(tensor.shape) != 2:
+        raise ValueError("Unexpected rank : %d (matrices only)" % len(tensor.shape))
+    if tensor.dtype != torch.float32 and tensor.dtype != torch.float64:
+        raise ValueError("Unexpected type : %s" % tensor.dtype)
+
+    h = tensor.shape[0]
+    w = tensor.shape[1]
+    nnz = sum([1 if tensor[i, j] != 0 else 0 for i in range(h) for j in range(w)])
+    density = (100.0 * nnz) / tensor.numel()
+    print("%%MatrixMarket matrix coordinate real general")
+    print("% https://math.nist.gov/MatrixMarket/formats.html")
+    print("%")
+    print("%% density = %4.2f%%" % density)
+    print("%")
+    print(h, w, nnz)
+    for i in range(h):
+        for j in range(w):
+            if tensor[i, j] != 0:
+                print(i + 1, j + 1, tensor[i, j].item())
+
+
+def print_extended_frostt_format(tensor: torch.Tensor):
+    """Prints the Extended FROSTT format for a sparse tensor.
+
+    Args:
+      tensor: sparse tensor
+    """
+    a = tensor.numpy()
+    nnz = sum([1 if x != 0 else 0 for x in np.nditer(a)])
+    density = (100.0 * nnz) / tensor.numel()
+    print("# Tensor in Extended FROSTT file format")
+    print("# http://frostt.io/tensors/file-formats.html")
+    print("# extended with two metadata lines:")
+    print("#   rank nnz")
+    print("#   dims (one per rank)")
+    print("#")
+    print("# density = %4.2f%%" % density)
+    print("#")
+    print(len(tensor.shape), nnz)
+    print(*tensor.shape, sep=" ")
+    it = np.nditer(a, flags=["multi_index"])
+    for x in it:
+        if x != 0:
+            print(*[i + 1 for i in it.multi_index], sep=" ", end=" ")
+            print(x)

test/CMakeLists.txt

@@ -11,6 +11,7 @@ configure_lit_site_cfg(
 
 set(MPACT_TEST_DEPENDS
         FileCheck count not
+	MPACTBenchmarkPythonModules
 	MPACTPythonModules
         TorchMLIRPythonModules
         torch-mlir-opt

test/python/file_formats.py

@@ -0,0 +1,61 @@
+# RUN: %PYTHON %s | FileCheck %s
+
+import numpy as np
+
+from mpact_benchmark.utils.tensor_generator import (
+    generate_tensor,
+    print_matrix_market_format,
+    print_extended_frostt_format,
+)
+
+x = generate_tensor(
+    seed=0, shape=(4, 7), sparsity=0.5, dtype=np.float32, drange=(4.0, 4.0)
+)
+
+# CHECK: %%MatrixMarket matrix coordinate real general
+# CHECK: % https://math.nist.gov/MatrixMarket/formats.html
+# CHECK: %
+# CHECK: % density = 50.00%
+# CHECK: %
+# CHECK: 4 7 14
+# CHECK: 1 2 4.0
+# CHECK: 1 3 4.0
+# CHECK: 1 6 4.0
+# CHECK: 2 4 4.0
+# CHECK: 2 5 4.0
+# CHECK: 2 7 4.0
+# CHECK: 3 1 4.0
+# CHECK: 3 3 4.0
+# CHECK: 3 4 4.0
+# CHECK: 3 7 4.0
+# CHECK: 4 2 4.0
+# CHECK: 4 4 4.0
+# CHECK: 4 5 4.0
+# CHECK: 4 7 4.0
+print_matrix_market_format(x)
+
+# CHECK: # Tensor in Extended FROSTT file format
+# CHECK: # http://frostt.io/tensors/file-formats.html
+# CHECK: # extended with two metadata lines:
+# CHECK: #   rank nnz
+# CHECK: #   dims (one per rank)
+# CHECK: #
+# CHECK: # density = 50.00%
+# CHECK: #
+# CHECK: 2 14
+# CHECK: 4 7
+# CHECK: 1 2 4.0
+# CHECK: 1 3 4.0
+# CHECK: 1 6 4.0
+# CHECK: 2 4 4.0
+# CHECK: 2 5 4.0
+# CHECK: 2 7 4.0
+# CHECK: 3 1 4.0
+# CHECK: 3 3 4.0
+# CHECK: 3 4 4.0
+# CHECK: 3 7 4.0
+# CHECK: 4 2 4.0
+# CHECK: 4 4 4.0
+# CHECK: 4 5 4.0
+# CHECK: 4 7 4.0
+print_extended_frostt_format(x)