modify kBatch value

host/driver_offline/include/device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_atomic_nchw_kcyx_nkhw.hpp

@@ -76,7 +76,7 @@ void device_convolution_backward_weight_implicit_gemm_v4r4r2_xdlops_atomic_nchw_
 
     constexpr index_t GemmCThreadTransferDstScalarPerVector = 1;
 
-    constexpr index_t KBatch = 96;
+    constexpr index_t KBatch = 64;
 #elif 1
     // [M, N, K0, K1] = [128, 128, 4, 8] for fp16
     constexpr index_t BlockSize = 256;

host/driver_offline/src/conv_wrw_driver_offline.cpp

@@ -253,8 +253,7 @@ int main(int argc, char* argv[])
                           in_left_pads_dev,
                           in_right_pads_dev);
     };
-    
-    
+
     // set zero to wei_device
     wei_device.GenerateTensorValue(GeneratorTensor_0{}, num_thread);
 #if USE_CONV_WRW_V4R4R2_XDL_NCHW
@@ -284,7 +283,6 @@ int main(int argc, char* argv[])
     }
 #endif
 
-
 #if USE_CONV_WRW_V4R4R4_XDL_NHWC
     if(algo == ConvBackwardWeightAlgo::V4R4R4XDLNHWC)
     {