- devicereset would lose track of default stream and thus subsequent
synchronization calls might not actually sychronize.
- Also deviceReset now correctly frees streams.
- fix waits in P2P staging copy - first phase (Device0-to-Staging) must
wait for second phase (Staging to Device1) to finish draining the
buffer.
[ROCm/hip commit: e020d68309]
- add P2P staging buffer copy.
- If copy device does not have sufficient access permissions, fall back
to staging buffer.
- improve docs for which copy device is used.
[ROCm/hip commit: dcabc9dbf7]
- set USE_PEER_TO_PEER=3 (requires HCC "am_memtracker_update_peers")
- when enabling peer, turn it on for previously allocated memory.
- hipDeviceCanAccessPeer is no longer self-ware (self does not qualify
as a peer)
- device peerlist always includes self, so when we call allow_access
we never remove self access.
- hipDeviceReset() removes old peer mappings.
[ROCm/hip commit: 1f53c55d3e]
- set USE_PEER_TO_PEER=3 (requires HCC "am_memtracker_update_peers")
- when enabling peer, turn it on for previously allocated memory.
- hipDeviceCanAccessPeer is no longer self-ware (self does not qualify
as a peer)
- device peerlist always includes self, so when we call allow_access
we never remove self access.
- hipDeviceReset() removes old peer mappings.
[ROCm/hip commit: 83f0de7806]
introduce LockedAccessor option so destructor does not unlock.
Allows locks to exist across function boundaries, required
for hipLaunchKernel macro which has several unusual requirements.
(including C comppatibility, must use variadic macro, more).
[ROCm/hip commit: 8635863724]
- Remove "call-to-call" for hipStreamCreate and hipEventCreate.
These now call an internal functions rather than calling through
hipStreamCreateWithFalgs and hipEventCreateWithFlags.
- Add HIP_INIT_API for more functions so they trace correctly.
- Use stream#DEVICE.STREAMID in debug messages via new specialization in
tace_helper.
[ROCm/hip commit: 7934cf620d]
Move critical data into separate class and protect with LockAccessor
wrapper class.
For device, the streams list is the critical data since it is modified when
streams are created or destroyed. The streams list is accessed in
several places including when synchronizing across all streams on the
device (ie from the default stream).
Other device data is set once by the device cosntructor and is not critical
so
All functions which acquire the LockAccessor now named with "locked_" prefix.
[ROCm/hip commit: 530ab9434a]
Ben Sander
Aditya Atluri
pensun