[Gemmini]add Gemmini-MiniLM-L6.

README.md

@@ -202,7 +202,7 @@ Before building the benchmark target, please see the following table and ensure
 | Cases | Hardware Configuration |
 | -------------- | ------------- |
 | Gemmini-ResNet-101  | defaultFpConfig ([link](./docs/GemminiConfig.md#using-default-float-point-configuration)) |
-
+| Gemmini-MiniLM-L6   | defaultFpConfig
 We assume you have already built all the components in the Gemmini README file. Now, let's build and run the cases. 
 
 ```
@@ -216,6 +216,7 @@ $ cmake -G Ninja .. \
 $ ninja
 $ cd bin
 $ spike --extension=gemmini pk Gemmini-ResNet-101
+$ spike --extension=gemmini pk Gemmini-MiniLM-L6
 ```
 
 ## Operation Optimization Benchmark

benchmarks/Gemmini/CMakeLists.txt

@@ -1 +1,2 @@
 add_subdirectory(ResNet-101)
+add_subdirectory(MiniLM-L6)

benchmarks/Gemmini/MiniLM-L6/CMakeLists.txt

@@ -0,0 +1,46 @@
+set(BUDDY_MLIR_BINARY_DIR ${BUDDY_MLIR_BUILD_DIR}/bin)
+set(CMAKE_CXX_COMPILER riscv64-unknown-linux-gnu-g++)
+set(GEMMINI_CONFIG "dim=4 acc_rows=4096 acc_t=f32 elem_t=f32")
+set(MINILM-L6-DIR ${CMAKE_CURRENT_SOURCE_DIR}/../../DeepLearning/Models/MiniLM-L6)
+
+if (NOT DEFINED ENV{RISCV})
+  message(FATAL_ERROR "Can't find RISCV environment variable(missing: RISCV_TOOLCHAIN)")
+endif()
+
+add_custom_command(
+  OUTPUT 
+    minilm-l6.o
+  COMMAND 
+    ${BUDDY_MLIR_BINARY_DIR}/buddy-opt ${MINILM-L6-DIR}/MiniLM-L6-12.mlir
+    --linalg-bufferize
+    --llvm-request-c-wrappers
+    --convert-linalg-to-gemmini="acc_t=f32"
+    --convert-linalg-to-loops
+    --func-bufferize
+    --arith-bufferize
+    --tensor-bufferize
+    --finalizing-bufferize
+    --lower-affine
+    --convert-scf-to-cf
+    --expand-strided-metadata
+    --convert-vector-to-llvm
+    --memref-expand
+    --arith-expand
+    --lower-gemmini=${GEMMINI_CONFIG}
+    --convert-arith-to-llvm
+    --finalize-memref-to-llvm
+    --convert-math-to-llvm
+    --reconcile-unrealized-casts | 
+    ${BUDDY_MLIR_BINARY_DIR}/buddy-translate -buddy-to-llvmir |
+    ${BUDDY_MLIR_BINARY_DIR}/buddy-llc -filetype=obj -mtriple=riscv64 
+    -mattr=+buddyext,+D -float-abi=hard -o minilm-l6.o
+)
+
+add_library(MINILM-L6 STATIC minilm-l6.o)
+set_target_properties(MINILM-L6 PROPERTIES LINKER_LANGUAGE CXX)
+add_executable(Gemmini-MiniLM-L6 Main.cpp)
+target_link_libraries(Gemmini-MiniLM-L6
+  -static
+  MINILM-L6
+  CRunnerUtils
+)

benchmarks/Gemmini/MiniLM-L6/Main.cpp

@@ -0,0 +1,70 @@
+//===- Main.cpp -----------------------------------------------------------===//
+//
+// 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.
+//
+//===----------------------------------------------------------------------===//
+//
+// This is the main file of the Gemmini MiniLM-L6 benchmark.
+//
+//===----------------------------------------------------------------------===//
+
+#include <buddy/Core/Container.h>
+#include <buddy/LLM/TextContainer.h>
+#include <cmath>
+#include <iostream>
+#include <string>
+
+namespace {
+
+// Declare the model C interface.
+extern "C" {
+void _mlir_ciface_forward(MemRef<float, 2> *output,
+                          buddy::Text<long long, 2> *input);
+}
+
+// Softmax function.
+void softmax(float *input, size_t size) {
+  assert(size > 0);
+  float m = input[0]; // Find the maximum value
+  for (size_t i = 1; i < size; ++i) {
+    if (input[i] > m) {
+      m = input[i];
+    }
+  }
+  float sum = 0.0;
+  for (size_t i = 0; i < size; ++i) {
+    sum += expf(input[i] - m);
+  }
+  float constant = m + logf(sum);
+  for (size_t i = 0; i < size; ++i) {
+    input[i] = expf(input[i] - constant);
+  }
+}
+} // namespace
+
+int main() {
+  std::string str = "buddy compiler is a domain specific compiler!";
+  buddy::Text<long long, 2> input(str);
+  intptr_t sizesOutput[2] = {1, 2};
+  MemRef<float, 2> output(sizesOutput);
+  input.tokenizeLlama(
+      "../../benchmarks/DeepLearning/Models/MiniLM-L6/Vocab.txt", 200);
+  _mlir_ciface_forward(&output, &input);
+  auto out = output.getData();
+  softmax(out, 2);
+  std::cout << std::string(53, '-') << std::endl;
+  std::cout << "Input: " << str << std::endl;
+  printf("The probability of positive label: %.2lf\n", out[1]);
+  printf("The probability of negative label: %.2lf\n", out[0]);
+  return 0;
+}