Writes data to a GM address without going through DCache. When multiple cores operate on GM addresses, if the data cannot be aligned to a Cache Line, using DCache will cause random data overwriting across cores because reads and writes are performed in Cache Line sizes. In this case, you can read and write GM addresses directly without going through DCache, thereby avoiding the random overwriting issue described above.
[object Object]
None
PIPE_S
None
[object Object]