[wasm] Add Minimum, Maximum kernels

tfjs-backend-wasm/src/cc/BUILD

@@ -128,6 +128,8 @@ tfjs_cc_library(
     ":Conv2D",
     ":DepthwiseConv2dNative",
     ":FloorDiv",
+    ":Minimum",
+    ":Maximum",
     ":FusedConv2D",
     ":FusedDepthwiseConv2D",
     ":Div",
@@ -289,6 +291,26 @@ tfjs_cc_library(
   ],
 )
 
+tfjs_cc_library(
+  name = "Minimum",
+  srcs = ["kernels/Minimum.cc"],
+  deps = [
+    ":backend",
+    ":binary",
+    ":util",
+  ],
+)
+
+tfjs_cc_library(
+  name = "Maximum",
+  srcs = ["kernels/Maximum.cc"],
+  deps = [
+    ":backend",
+    ":binary",
+    ":util",
+  ],
+)
+
 tfjs_cc_library(
   name = "FusedConv2D",
   srcs = ["kernels/FusedConv2D.cc"],

tfjs-backend-wasm/src/cc/kernels/Maximum.cc

@@ -0,0 +1,53 @@
+/* Copyright 2019 Google Inc. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * ===========================================================================*/
+
+#ifdef __EMSCRIPTEN__
+#include <emscripten.h>
+#endif
+
+#include <algorithm>
+
+#include "src/cc/binary.h"
+#include "src/cc/util.h"
+
+namespace tfjs {
+namespace wasm {
+// We use C-style API to interface with Javascript.
+extern "C" {
+
+#ifdef __EMSCRIPTEN__
+EMSCRIPTEN_KEEPALIVE
+#endif
+void Maximum(const int a_id, const size_t* a_shape_ptr, const int a_shape_len,
+             const int b_id, const size_t* b_shape_ptr, const int b_shape_len,
+             const DType dtype, const int out_id) {
+  switch (dtype) {
+    case DType::float32:
+      binary_f32(a_id, b_id, out_id,
+        [](float a, float b) { return std::max(a, b); });
+      break;
+    case DType::int32:
+      binary_i32(a_id, b_id, out_id,
+        [](int a, int b) { return std::max(a, b); });
+      break;
+    default:
+      util::warn(
+          "Maximum for tensor ids %d and %d failed. Unsupported dtype %d",
+          a_id, b_id, dtype);
+  }
+}
+
+}  // extern "C"
+}  // namespace wasm
+}  // namespace tfjs

tfjs-backend-wasm/src/cc/kernels/Minimum.cc

@@ -0,0 +1,53 @@
+/* Copyright 2019 Google Inc. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * ===========================================================================*/
+
+#ifdef __EMSCRIPTEN__
+#include <emscripten.h>
+#endif
+
+#include <algorithm>
+
+#include "src/cc/binary.h"
+#include "src/cc/util.h"
+
+namespace tfjs {
+namespace wasm {
+// We use C-style API to interface with Javascript.
+extern "C" {
+
+#ifdef __EMSCRIPTEN__
+EMSCRIPTEN_KEEPALIVE
+#endif
+void Minimum(const int a_id, const size_t* a_shape_ptr, const int a_shape_len,
+             const int b_id, const size_t* b_shape_ptr, const int b_shape_len,
+             const DType dtype, const int out_id) {
+  switch (dtype) {
+    case DType::float32:
+      binary_f32(a_id, b_id, out_id,
+        [](float a, float b) { return std::min(a, b); });
+      break;
+    case DType::int32:
+      binary_i32(a_id, b_id, out_id,
+        [](int a, int b) { return std::min(a, b); });
+      break;
+    default:
+      util::warn(
+          "Minimum for tensor ids %d and %d failed. Unsupported dtype %d",
+          a_id, b_id, dtype);
+  }
+}
+
+}  // extern "C"
+}  // namespace wasm
+}  // namespace tfjs

tfjs-backend-wasm/src/kernels/Maximum.ts

@@ -0,0 +1,20 @@
+/**
+ * @license
+ * Copyright 2019 Google Inc. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+
+import { registerBinaryKernel } from './binary_kernel';
+const supportsBroadcast = false;
+registerBinaryKernel('Maximum', supportsBroadcast);

tfjs-backend-wasm/src/kernels/Minimum.ts

@@ -0,0 +1,20 @@
+/**
+ * @license
+ * Copyright 2019 Google Inc. All Rights Reserved.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ * =============================================================================
+ */
+
+import { registerBinaryKernel } from './binary_kernel';
+const supportsBroadcast = false;
+registerBinaryKernel('Minimum', supportsBroadcast);

tfjs-backend-wasm/src/kernels/all_kernels.ts

@@ -37,8 +37,10 @@ import './FusedBatchNorm';
 import './FusedConv2D';
 import './FusedDepthwiseConv2D';
 import './Max';
+import './Maximum';
 import './MaxPool';
 import './Min';
+import './Minimum';
 import './Mul';
 import './NonMaxSuppressionV3';
 import './PadV2';

tfjs-backend-wasm/src/setup_test.ts

@@ -195,6 +195,22 @@ const TEST_FILTERS: TestFilter[] = [
   {include: 'argmax', excludes: ['gradient']},
   {include: 'exp '},
   {include: 'unstack'},
+  {
+    include: 'minimum',
+    excludes: [
+      'gradient',  // Not yet implemented.
+      'broadcasts 2x1 Tensor2D and 2x2 Tensor2D' // Broadcasting along inner
+                                                 // dims not supported yet.
+    ]
+  },
+  {
+    include: 'maximum',
+    excludes: [
+      'gradient',  // Not yet implemented.
+      'broadcasts 2x1 Tensor2D and 2x2 Tensor2D' // Broadcasting along inner
+                                                 // dims not supported yet.
+    ]
+  },
 ];
 
 const customInclude = (testName: string) => {

tfjs-core/src/ops/binary_ops.ts

@@ -673,13 +673,13 @@ function minimum_<T extends Tensor>(
     const [$a, $b] = saved;
     const derA = () => dy.mul($a.lessEqual($b).toFloat());
     const derB = () => dy.mul($a.greater($b).toFloat());
-    return {$a: derA, $b: derB};
+    return {a: derA, b: derB};
   };
   return ENGINE.runKernelFunc((backend, save) => {
     const res = backend.minimum($a, $b);
     save([$a, $b]);
     return res;
-  }, {$a, $b}, der) as T;
+  }, {a: $a, b: $b}, der, 'Minimum') as T;
 }
 
 /**
@@ -738,13 +738,13 @@ function maximum_<T extends Tensor>(
     const [$a, $b] = saved;
     const derA = () => dy.mul($a.greaterEqual($b).toFloat());
     const derB = () => dy.mul($a.less($b).toFloat());
-    return {$a: derA, $b: derB};
+    return {a: derA, b: derB};
   };
   return ENGINE.runKernelFunc((backend, save) => {
     const res = backend.maximum($a, $b);
     save([$a, $b]);
     return res;
-  }, {$a, $b}, der) as T;
+  }, {a: $a, b: $b}, der, 'Maximum') as T;
 }
 
 /**