[rocm-libraries] ROCm/rocm-libraries#6302 (commit 8d419e8)

CK: Remove 41 commented-out dead code blocks (~200 lines)
 (#6302)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Depends on #6300

## Summary

Remove 41 commented-out code blocks across 33 files in Composable
Kernel, totaling ~200 lines.

Identified using an automated dead code scanning skill (`ck-dead-code`)
with a calibrated two-stage pipeline:
1. **Pre-filter**: Keyword-based scan found 1,338 `//`-commented blocks.
Calibrated heuristics (trained on 50-sample expert classification)
reduced to 89 high-confidence candidates — 93% noise reduction.
2. **Expert triage**: LLM expert classified each block in context as
CODE_REMOVE, CODE_KEEP, or NOT_CODE.

| Classification | Count |
|---------------|-------|
| Removed (this PR) | 41 |
| Kept (debug helpers, alt configs, reference impls) | 32 |
| Not code (false positives) | 16 |

Removed blocks include: superseded implementations, old test data,
abandoned stubs, unreachable code, and buggy dead code.

Author: AviralGoelAMD

codegen/test/grouped_conv_fwd_multiple_d_v1.cpp

@@ -198,10 +198,6 @@ struct Epilogue
                                                               input_left_pads,
                                                               input_right_pads);
 
-        // auto res = rtc::from_gpu(out_dev);
-        // pass &= ck::utils::check_err(res, out_host, "Error: incorrect results!", 1e-5f, 1e-4f);
-        // assert(pass);
-
         // Simple check: this checks that the output from each instance matches the output from the
         // first instance
         CHECK(report(solution, check(rtc::from_gpu(out_dev))));

codegen/test/grouped_conv_fwd_multiple_d_v2.cpp

@@ -198,10 +198,6 @@ struct Epilogue
                                                               input_left_pads,
                                                               input_right_pads);
 
-        // auto res = rtc::from_gpu(out_dev);
-        // pass &= ck::utils::check_err(res, out_host, "Error: incorrect results!", 1e-5f, 1e-4f);
-        // assert(pass);
-
         // Simple check: this checks that the output from each instance matches the output from the
         // first instance
         CHECK(report(solution, check(rtc::from_gpu(out_dev))));

codegen/test/grouped_conv_fwd_multiple_d_v3.cpp

@@ -198,10 +198,6 @@ struct Epilogue
                                                               input_left_pads,
                                                               input_right_pads);
 
-        // auto res = rtc::from_gpu(out_dev);
-        // pass &= ck::utils::check_err(res, out_host, "Error: incorrect results!", 1e-5f, 1e-4f);
-        // assert(pass);
-
         // Simple check: this checks that the output from each instance matches the output from the
         // first instance
         CHECK(report(solution, check(rtc::from_gpu(out_dev))));

codegen/test/grouped_conv_fwd_multiple_d_v4.cpp

@@ -198,10 +198,6 @@ struct Epilogue
                                                               input_left_pads,
                                                               input_right_pads);
 
-        // auto res = rtc::from_gpu(out_dev);
-        // pass &= ck::utils::check_err(res, out_host, "Error: incorrect results!", 1e-5f, 1e-4f);
-        // assert(pass);
-
         // Simple check: this checks that the output from each instance matches the output from the
         // first instance
         CHECK(report(solution, check(rtc::from_gpu(out_dev))));

example/60_gemm_multi_ABD/gemm_multi_ABD_xdl_multiply_bias_fastgelu_bf16_i8.cpp

@@ -238,16 +238,6 @@ int main(int argc, char* argv[])
 
         Tensor<B1DataType> b_k_n(f_host_tensor_descriptor(K, N, StrideB, B0Layout{}));
 
-#if 0
-        for(int n = 0; n < N; ++n)
-        {
-            for(int k = 0; k < K; ++k)
-            {
-                b_element_op(b_k_n(k, n), b0_k_n(k, n), b1_k_n(k, n));
-            }
-        }
-#endif
-
         using ReferenceGemmInstance = ck::tensor_operation::host::ReferenceGemm<A0DataType,
                                                                                 B0DataType,
                                                                                 CShuffleDataType,

example/65_gemm_multiply_multiply/moe_gemm2_xdl_fp8.cpp

@@ -252,9 +252,6 @@ int main(int argc, char* argv[])
     Tensor<ck::index_t> expert_ids(HostTensorDescriptor({sorted_tile_num}, {1}));
     Tensor<ck::index_t> sorted_token_ids(HostTensorDescriptor({sorted_size}, {1}));
     Tensor<ck::index_t> max_token_id(HostTensorDescriptor({1}));
-    // max_token_id.mData[0] = valid_size;
-    // max_token_id.mData = {valid_size, 0, 2, 3, 4, 6, 8, 10, 12, 13};
-    // int eids[]         = {0, 0, 1, 2, 3, 3, 4, 4, 5, 5, 6, 7, 7, 3, 3, 3};
     max_token_id.mData = {valid_size, 0, 1, 2, 3, 4, 5, 6, 7, 8};
     // int eids[]         = {0, 1, 2, 3, 4, 5, 6, 7, 3, 3, 3}; // {2, 1, 1, 2, 2, 2, 1, 2}
     for(int i = 0; i < sorted_tile_num; i++)

example/65_gemm_multiply_multiply/moe_gemm2_xdl_fp8_blockscale.cpp

@@ -261,16 +261,6 @@ int main(int argc, char* argv[])
     Tensor<ck::index_t> max_token_id(HostTensorDescriptor({1}));
 
     max_token_id.mData = {valid_size, 0, 1, 2, 3, 4, 5, 6, 7, 8};
-    // int eids[]         = {0, 1, 3, 3, 3};
-    //  int eids[]         = {0, 1, 2, 3, 4, 5, 6, 7}; //, 3, 3, 3}; // {2, 1, 1, 2, 2, 2, 1, 2}
-    // int eids[] = {0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 3, 3, 3};
-    // int eids[]         = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
-    //                     1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
-    //                     2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
-    //                     3, 3, 3, 3, 3, 3, 3, 3, 4, 4,
-    //                     5, 5, 5, 5, 6, 6, 6, 6, 7, 7,
-    //                     7, 7,
-    //                     3, 3, 3};
     for(int i = 0; i < sorted_tile_num; i++)
     {
         expert_ids.mData[i] = i / ck::math::integer_divide_ceil(valid_tile_num, experts);

example/ck_tile/11_add_rmsnorm2d_rdquant/instances/add_rmsnorm2d_rdquant_fwd_bf16_n1024_instance.cpp

@@ -5,14 +5,6 @@
 
 // clang-format off
 //                                                                rm  rn  tm  tn  vn  pd      x     3p
-#if 0
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::bf16_t, 1,  2,  4,  64, 8,  true , true, false>>(const S&, A);
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::bf16_t, 1,  4,  4,  64, 4,  true , true, false>>(const S&, A);
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::bf16_t, 1,  8,  4,  64, 2,  true , true, false>>(const S&, A);
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::bf16_t, 1, 16,  4,  64, 1,  true , true, false>>(const S&, A);
-
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::bf16_t, 1,  1,  1, 256, 4,  true , true, false>>(const S&, A);
-#endif
 
 template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::bf16_t,  ck_tile::int8_t, 1, 1, 2,  128, 8,  true,  true, false>>(const S&, A);
 template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::bf16_t,  ck_tile::int8_t, 1, 2, 2,  128, 4,  true,  true, false>>(const S&, A);

example/ck_tile/11_add_rmsnorm2d_rdquant/instances/add_rmsnorm2d_rdquant_fwd_fp16_n1024_instance.cpp

@@ -5,14 +5,6 @@
 
 // clang-format off
 //                                                                rm  rn  tm  tn  vn  pd      x     3p
-#if 0
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::fp16_t, 1,  2,  4,  64, 8,  true , true, false>>(const S&, A);
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::fp16_t, 1,  4,  4,  64, 4,  true , true, false>>(const S&, A);
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::fp16_t, 1,  8,  4,  64, 2,  true , true, false>>(const S&, A);
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::fp16_t, 1, 16,  4,  64, 1,  true , true, false>>(const S&, A);
-
-template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::fp16_t, 1,  1,  1, 256, 4,  true , true, false>>(const S&, A);
-#endif
 
 template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::fp16_t,  ck_tile::int8_t, 1, 1, 2,  128, 8,  true,  true, false>>(const S&, A);
 template float add_rmsnorm2d_rdquant_fwd_<trait_<ck_tile::fp16_t,  ck_tile::int8_t, 1, 2, 2,  128, 4,  true,  true, false>>(const S&, A);

example/ck_tile/12_smoothquant/instances/smoothquant_bf16_n1024_instance.cpp

@@ -5,14 +5,6 @@
 
 // clang-format off
 //                                                  rm rn  tm   tn  vn   pd   2p
-#if 0
-template float smoothquant_<trait_<ck_tile::bf16_t, 1,  2,  4,  64, 8,  true, false>>(const S&, A);
-template float smoothquant_<trait_<ck_tile::bf16_t, 1,  4,  4,  64, 4,  true, false>>(const S&, A);
-template float smoothquant_<trait_<ck_tile::bf16_t, 1,  8,  4,  64, 2,  true, false>>(const S&, A);
-template float smoothquant_<trait_<ck_tile::bf16_t, 1, 16,  4,  64, 1,  true, false>>(const S&, A);
-
-template float smoothquant_<trait_<ck_tile::bf16_t, 1,  1,  1, 256, 4,  true, false>>(const S&, A);
-#endif
 
 template float smoothquant_<trait_<ck_tile::bf16_t,  1, 1, 2,  128, 8,  true, false>>(const S&, A);
 template float smoothquant_<trait_<ck_tile::bf16_t,  1, 2, 2,  128, 4,  true, false>>(const S&, A);