PR openxla#19342: [ROCm] Skip unsupported tests in dot_algorithms_test

Imported from GitHub PR openxla#19342

Triton is currently disabled on ROCm. Skipping the following subtests in `dot_algorithms_test`:
- TritonAlgorithmTest.Algorithm_BF16_BF16_F32_X3
- TritonAlgorithmTest.Algorithm_BF16_BF16_F32_X6
- TritonAlgorithmTest.Algorithm_TF32_TF32_F32
- TritonAlgorithmTest.Algorithm_TF32_TF32_F32_X3
- TritonAlgorithmTest.Algorithm_BF16_BF16_F32
Copybara import of the project:

--
32bd775 by Milica Makevic <Milica.Makevic@amd.com>:

Disable unsupported Triton subtests

Merging this change closes openxla#19342

COPYBARA_INTEGRATE_REVIEW=openxla#19342 from ROCm:disable_triton_tests 32bd775
PiperOrigin-RevId: 696740956

xla/service/gpu/fusions/triton/dot_algorithms_test.cc

@@ -1146,6 +1146,9 @@ CHECK-NOT: mma.sync.aligned.{{.*}}.row.col.f32.tf32.tf32.f32
 }
 
 TEST_F(TritonAlgorithmTest, Algorithm_BF16_BF16_F32_X3) {
+  if (std::holds_alternative<se::RocmComputeCapability>(GpuComputeComp())) {
+    GTEST_SKIP() << "Triton currently disabled on ROCM.";
+  }
   const std::string kHloText = R"(
     HloModule Algorithm_BF16_BF16_F32_X3
 
@@ -1166,6 +1169,9 @@ TEST_F(TritonAlgorithmTest, Algorithm_BF16_BF16_F32_X3) {
 }
 
 TEST_F(TritonAlgorithmTest, Algorithm_BF16_BF16_F32_X6) {
+  if (std::holds_alternative<se::RocmComputeCapability>(GpuComputeComp())) {
+    GTEST_SKIP() << "Triton currently disabled on ROCM.";
+  }
   const std::string kHloText = R"(
     HloModule Algorithm_BF16_BF16_F32_X6
 
@@ -1185,7 +1191,33 @@ TEST_F(TritonAlgorithmTest, Algorithm_BF16_BF16_F32_X6) {
   EXPECT_TRUE(ok);
 }
 
+TEST_F(TritonAlgorithmTest, Algorithm_TF32_TF32_F32) {
+  if (std::holds_alternative<se::RocmComputeCapability>(GpuComputeComp())) {
+    GTEST_SKIP() << "Triton currently disabled on ROCM.";
+  }
+  const std::string kHloText = R"(
+    HloModule Algorithm_TF32_TF32_F32
+
+    ENTRY main {
+      lhs = f32[128,1]{1,0} parameter(0)
+      rhs = f32[1,128]{1,0} parameter(1)
+      ROOT dot = f32[128,128]{1,0} dot(lhs, rhs),
+          algorithm=dot_tf32_tf32_f32,
+          lhs_contracting_dims={1},
+          rhs_contracting_dims={0}
+    }
+  )";
+  const std::string pattern =
+      R"(CHECK: "kind":"__triton_gemm","triton_gemm_config")";
+  TF_ASSERT_OK_AND_ASSIGN(auto module, GetOptimizedModule(kHloText));
+  TF_ASSERT_OK_AND_ASSIGN(auto ok, RunFileCheck(module->ToString(), pattern));
+  EXPECT_TRUE(ok);
+}
+
 TEST_F(TritonAlgorithmTest, Algorithm_TF32_TF32_F32_X3) {
+  if (std::holds_alternative<se::RocmComputeCapability>(GpuComputeComp())) {
+    GTEST_SKIP() << "Triton currently disabled on ROCM.";
+  }
   const std::string kHloText = R"(
     HloModule Algorithm_TF32_TF32_F32_X3
 
@@ -1209,6 +1241,9 @@ TEST_F(TritonAlgorithmTest, Algorithm_BF16_BF16_F32) {
   if (!SupportsBF16(GpuComputeComp())) {
     GTEST_SKIP() << "BF16 not supported.";
   }
+  if (std::holds_alternative<se::RocmComputeCapability>(GpuComputeComp())) {
+    GTEST_SKIP() << "Triton currently disabled on ROCM.";
+  }
   const std::string kHloText = R"(
     HloModule Algorithm_BF16_BF16_F32