fix the builds for gfx950

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_two_stage_xdl_cshuffle.hpp

@@ -111,7 +111,7 @@ __global__ void
             const index_t num_k_per_block)
 {
 #if(!defined(__HIP_DEVICE_COMPILE__) || defined(__gfx908__) || defined(__gfx90a__) || \
-    defined(__gfx940__) || defined(__gfx941__) || defined(__gfx942__))
+    defined(__gfx940__) || defined(__gfx941__) || defined(__gfx942__) || defined(__gfx950__))
     // offset base pointer for each work-group
     const index_t g_idx = __builtin_amdgcn_readfirstlane(blockIdx.z * NumBatchToMerge);
     const index_t k_idx = __builtin_amdgcn_readfirstlane(blockIdx.y * num_k_per_block);

include/ck_tile/core/config.hpp

@@ -4,10 +4,10 @@
 #pragma once
 
 #if defined(__gfx908__) || defined(__gfx90a__) || defined(__gfx940__) || defined(__gfx941__) || \
-    defined(__gfx942__)
+    defined(__gfx942__) || defined(__gfx950__)
 #define __gfx9__
 #endif
-#if defined(__gfx940__) || defined(__gfx941__) || defined(__gfx942__)
+#if defined(__gfx940__) || defined(__gfx941__) || defined(__gfx942__) || defined(__gfx950__)
 #define __gfx94__
 #endif
 #if defined(__gfx1030__) || defined(__gfx1031__) || defined(__gfx1032__) || \