Rename proto flag

This makes it clear this flag is specific to this test, rather than a generic tuner flag

Author: nicolasvasilache

examples/blockdiagperm.cc

@@ -30,7 +30,7 @@
 #include "tc/core/flags.h"
 #include "tc/core/mapping_options.h"
 
-DEFINE_string(tuner_proto, "", "Filename to load and store proto cache ");
+DEFINE_string(proto_path, "", "Filename to load and store proto cache ");
 
 TEST(BlockDiagPerm, SimpleAutotune) {
   // 1. Define and setup the TC compilation unit with CUDA memory
@@ -77,7 +77,7 @@ def blockdiagperm2dfissioned_2(float(B, N) I, int32(N) Idx) -> (O) {
   auto options = tc::CudaMappingOptions::makeNaiveCudaMappingOptions();
   tc::autotune::GeneticAutotunerATen geneticAutotuneATen(tc);
   auto bestOption = geneticAutotuneATen.tune(
-      FLAGS_tuner_proto, "blockdiagperm2dfissioned_1", {I, W}, options);
+      FLAGS_proto_path, "blockdiagperm2dfissioned_1", {I, W}, options);
   auto handle = atCompl.compile(
       "blockdiagperm2dfissioned_1", {I, W}, bestOption.getValue());
   std::vector<at::Tensor> outputs;
@@ -89,7 +89,7 @@ def blockdiagperm2dfissioned_2(float(B, N) I, int32(N) Idx) -> (O) {
   at::Tensor Idx = at::CPU(at::kInt).randperm({500}).toBackend(at::kCUDA);
   tc::autotune::GeneticAutotunerATen geneticAutotuneATen2(tc);
   auto bestOption2 = geneticAutotuneATen.tune(
-      FLAGS_tuner_proto, "blockdiagperm2dfissioned_2", {O, Idx}, options);
+      FLAGS_proto_path, "blockdiagperm2dfissioned_2", {O, Idx}, options);
   auto handle2 = atCompl.compile(
       "blockdiagperm2dfissioned_2", {O, Idx}, bestOption2.getValue());
   std::vector<at::Tensor> outputs2;
@@ -120,7 +120,7 @@ def blockdiagperm2dfissioned_2(float(B, N) I, int32(N) Idx) -> (O) {
 // From root, run with:
 //   ./build/examples/blockdiagperm --tuner_threads=10 --tuner_gen_pop_size=10
 //   --tuner_gen_generations=3 --tuner_gen_number_elites=4
-//   --tuner_proto="/tmp/blockdiagperm"
+//   --proto_path="/tmp/blockdiagperm"
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);
   ::gflags::ParseCommandLineFlags(&argc, &argv, true);

examples/tensordot.cc

@@ -30,7 +30,7 @@
 
 #include "../test/test_harness_aten_cuda.h"
 
-DEFINE_string(tuner_proto, "", "Filename to load and store proto cache ");
+DEFINE_string(proto_path, "", "Filename to load and store proto cache ");
 
 TEST(TensorDot, SimpleAutotune) {
   // 1. Define and setup the TC compilation unit with CUDA memory
@@ -53,7 +53,7 @@ def tensordot(float(N, C1, C2, H, W) I0,
   auto naiveOptions = tc::CudaMappingOptions::makeNaiveCudaMappingOptions();
   tc::autotune::GeneticAutotunerATen geneticAutotuneATen(tc);
   auto bestOption = geneticAutotuneATen.tune(
-      FLAGS_tuner_proto, "tensordot", {I0, I1}, naiveOptions);
+      FLAGS_proto_path, "tensordot", {I0, I1}, naiveOptions);
 
   // 4. Compile and run the TC with the best option.
   // Outputs get allocated; could also be pre-allocated and passed.
@@ -90,7 +90,7 @@ def tensordot(float(N, C1, C2, H, W) I0,
 // From root, run with:
 //   ./build/examples/tensordot --tuner_threads=10 --tuner_gen_pop_size=10
 //   --tuner_gen_generations=3 --tuner_gen_number_elites=4
-//   --tuner_proto="/tmp/tensordot"
+//   --proto_path="/tmp/tensordot"
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);
   ::gflags::ParseCommandLineFlags(&argc, &argv, true);

examples/wavenet.cc

@@ -30,7 +30,7 @@
 
 #include "../test/test_harness_aten_cuda.h"
 
-DEFINE_string(tuner_proto, "", "Filename to load and store proto cache ");
+DEFINE_string(proto_path, "", "Filename to load and store proto cache ");
 
 TEST(WaveNet2Layers, SimpleAutotune) {
   // 1. Define and setup the TC compilation unit with CUDA memory
@@ -121,7 +121,7 @@ def wavenet2layers(
                                      skip_weight1,
                                      skip_bias1};
   auto bestOption = geneticAutotuneATen.tune(
-      FLAGS_tuner_proto, "wavenet2layers", tensors, naiveOptions);
+      FLAGS_proto_path, "wavenet2layers", tensors, naiveOptions);
 
   // 4. Compile and run the TC with the best option.
   // Outputs get allocated; could also be pre-allocated and passed.
@@ -165,7 +165,7 @@ def wavenet2layers(
 // From root, run with:
 //   ./build/examples/wavenet --tuner_threads=10 --tuner_gen_pop_size=10
 //   --tuner_gen_generations=3 --tuner_gen_number_elites=4
-//   --benchmark_iterations=1000 --tuner_proto="/tmp/wavenet"
+//   --benchmark_iterations=1000 --proto_path="/tmp/wavenet"
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);
   ::gflags::ParseCommandLineFlags(&argc, &argv, true);