From 4e53a588c366fa7258c5647b8ced4e738a9b603b Mon Sep 17 00:00:00 2001
From: aska-0096 <haocwang@amd.com>
Date: Tue, 3 Jun 2025 02:44:31 +0000
Subject: [PATCH] re-enable more instances for a8w8

---
 .../gpu/gemm_multiply_multiply_wp.hpp         | 154 ++++++++++++++++++
 1 file changed, 154 insertions(+)

diff --git a/library/include/ck/library/tensor_operation_instance/gpu/gemm_multiply_multiply_wp.hpp b/library/include/ck/library/tensor_operation_instance/gpu/gemm_multiply_multiply_wp.hpp
index 90a9fa381d..a42584b679 100644
--- a/library/include/ck/library/tensor_operation_instance/gpu/gemm_multiply_multiply_wp.hpp
+++ b/library/include/ck/library/tensor_operation_instance/gpu/gemm_multiply_multiply_wp.hpp
@@ -181,6 +181,137 @@ void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma1
                                                                      MultiplyMultiply>>>&
         instances);
 
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p1_default_instances_v2(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p2_default_instances_v2(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p3_default_instances_v2(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p4_default_instances_v2(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p5_default_instances_v2(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p1_default_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p2_default_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p3_default_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p4_default_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p5_default_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     BF16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>&
+        instances);
 #endif
 
 template <typename ADataType,
@@ -262,6 +393,29 @@ struct DeviceOperationInstanceFactory<
                     op_ptrs);
                 add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma16x16_mn_compute_default_instances_p6(
                     op_ptrs);
+                
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p1_default_instances_v2(
+                    op_ptrs);
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p2_default_instances_v2(
+                    op_ptrs);
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p3_default_instances_v2(
+                    op_ptrs);
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p4_default_instances_v2(
+                    op_ptrs);
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p5_default_instances_v2(
+                    op_ptrs);
+                
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p1_default_instances(
+                    op_ptrs);
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p2_default_instances(
+                    op_ptrs);
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p3_default_instances(
+                    op_ptrs);
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p4_default_instances(
+                    op_ptrs);
+                add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_bf16_mk_mfma_mn_p5_default_instances(
+                    op_ptrs);
+                
             }
         }
 #endif