Thanks for sending an issue! Please fill in the following template to help quickly solve your problem.

Describe the current behavior / 问题描述 (Mandatory / 必填)

代码检视过程中，发现算子源码可能存在性能问题。
index\apply_top_k_top_p_with_sorted\op_kernel\apply_top_k_top_p_with_sorted.h
107行
DataCopyCustom<T>(x1Local_in, x1Gm[gm_bias], numCol);
108行
DataCopyCustom<T>(x2Local, x2Gm[gm_bias], numCol);
138行：
DataCopyCustom<T>(xGm[gm_bias], x_out, numCol);
291行
DataCopyCustom<T>(yGm[progress], yLocal, numCol);
299行
DataCopyCustom<float>(rstdGm[outer_progress * rowFactor], rstdLocal, num);

norm/add_rms_norm/op_kernel/add_rms_norm_merge_n.h
114行：
DataCopyCustom<T>(x1Local, x1Gm[gm_bias], calc_row_num * numCol);
116行：
DataCopyCustom<T>(x1Local, x1Gm[gm_bias], calc_row_num, numCol);
121行：
DataCopyCustom<T>(x2Local, x2Gm[gm_bias], calc_row_num * numCol);
123行：
DataCopyCustom<T>(x2Local, x2Gm[gm_bias], calc_row_num, numCol);
157行：
DataCopyCustom<T>(xGm[gm_bias], xOut, calc_row_num * numCol);
159行：
DataCopyCustom<T>(xGm[gm_bias], xOut, calc_row_num, numCol);
262行：
DataCopyCustom<T>(yGm[progress], yLocal, calc_row_num, numCol);
271行：
DataCopyCustom<float>(rstdGm[outer_progress * rowFactor], rstdLocal, num);

规则描述：
【描述】：涉及shape大小需要使用int64表示。例如：Memset算子，该算子功能为将tensor所在内存清零，它的切分参数可能会有较大值。Int32的上限为2G，当代模型数据往往会超出该范围，因此必须使用int64表示。

Environment / 环境信息 (Mandatory / 必填)

https://gitcode.com/cann/ops-nn/blob/master/norm/add_rms_norm/op_kernel/add_rms_norm.h
https://gitcode.com/cann/ops-nn/blob/master/norm/add_rms_norm/op_kernel/add_rms_norm_merge_n.h

Steps to reproduce the issue / 重现步骤 (Mandatory / 必填)

检视代码即可

Describe the expected behavior / 预期结果 (Mandatory / 必填)

gm内存偏移使用int64表示

Related log / screenshot / 日志 / 截图 (Mandatory / 必填)

无

Special notes for this issue/备注 (Optional / 选填)