You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Allocates the memory as write-combined. On some system configurations, write-combined allocation may be transferred faster across the PCI Express bus, however, could have low read efficiency by most CPUs. It’s a good option for data transfer from host to device via mapped pinned memory.
However, it looks like it's not ever used by the rocm runtime. Its value is stored for later query with hipPointerGetAttributes, but I see no other uses of this value.
I'd like to see a clarification of this in the documentation. If I'm correct and it is indeed unused by the runtime, perhaps we could have some wording along the lines of "for CUDA compatibility; ignored"
The text was updated successfully, but these errors were encountered:
https://rocm.docs.amd.com/projects/HIP/en/docs-develop/reference/hip_runtime_api/global_defines_enums_structs_files/global_enum_and_defines.html#c.hipHostMallocWriteCombined suggests that the
hipHostMallocWriteCombined
flagHowever, it looks like it's not ever used by the rocm runtime. Its value is stored for later query with
hipPointerGetAttributes
, but I see no other uses of this value.I'd like to see a clarification of this in the documentation. If I'm correct and it is indeed unused by the runtime, perhaps we could have some wording along the lines of "for CUDA compatibility; ignored"
The text was updated successfully, but these errors were encountered: