tuned

driver/device_convolution_implicit_gemm_v4_nchw_kcyx_nkhw.hpp

@@ -75,14 +75,14 @@ void device_convolution_implicit_gemm_v4_nchw_kcyx_nkhw(InDesc,
     constexpr index_t InBlockCopySrcDataPerRead_B   = 1;
     constexpr index_t InBlockCopyDstDataPerWrite_N2 = 4;
 
-    using WeiBlockCopySubLengths_E_K            = Sequence<1, 4>;
-    using WeiBlockCopyClusterLengths_E_K        = Sequence<8, 32>;
-    using WeiBlockCopyThreadClusterArrangeOrder = Sequence<1, 0>; // [E, K]
-    using WeiBlockCopySrcAccessOrder            = Sequence<1, 0>; // [E, K]
-    using WeiBlockCopyDstAccessOrder            = Sequence<0, 1>; // [K, E]
-
-    constexpr index_t WeiBlockCopySrcDataPerRead_E  = 1;
-    constexpr index_t WeiBlockCopyDstDataPerWrite_K = 4;
+    using WeiBlockCopySubLengths_E_K            = Sequence<4, 1>;
+    using WeiBlockCopyClusterLengths_E_K        = Sequence<2, 128>;
+    using WeiBlockCopyThreadClusterArrangeOrder = Sequence<1, 0>; // [K, E]
+    using WeiBlockCopySrcAccessOrder            = Sequence<1, 0>; // [K, E]
+    using WeiBlockCopyDstAccessOrder            = Sequence<0, 1>; // [E, K]
+
+    constexpr index_t WeiBlockCopySrcDataPerRead_E  = 4;
+    constexpr index_t WeiBlockCopyDstDataPerWrite_K = 1;
 #endif
 
     constexpr index_t GridSize =