microsoft/mscclpp
1
0
Fork 0
mirror of https://github.com/microsoft/mscclpp.git synced 2026-05-11 17:00:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revised MemoryChannel interfaces (#508)

Browse Source 
* Moved the `MemoryChannel::copy()` method out of the `MemoryChannel` as
a standalone function.
* Renamed `mscclpp::putPackets()` and `mscclpp::getPackets()` to
`mscclpp::copyToPackets()` and `mscclpp::copyFromPackets()` respectively
for consistency.
* Renamed `MemoryChannel::getPackets()` to
`MemoryChannel::unpackPackets()` for clarity. Renamed `getPacketBuffer`
to `packetBuffer`.
* Added the `MemoryChannel::unpackPacket()` method that unpacks one
packet in the buffer.
* Added the `BaseMemoryChannel` class that only contains a semaphore
without memory addresses.
* Removed the `MemoryDevice2DeviceSemaphoreDeviceHandle::signalPacket()`
method that is lacking use cases.
This commit is contained in:
Changho Hwang
2025-04-24 17:02:56 -07:00
committed by GitHub
parent 9df2bdb2bf
commit 710f6686dc
19 changed files with 518 additions and 499 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/test/memory_channel_test.cu
View File

@@ -16,7 +16,7 @@ extern "C" __global__ void __launch_bounds__(1024, 1)
if (bid < nranks && bid != my_rank) {
if (use_packet) {
channels[bid].putPackets(2 * my_offset, my_offset, size_per_rank, tid, blockDim.x, flag);
channels[bid].getPackets(2 * my_nghr_offset, my_nghr_offset, size_per_rank, tid, blockDim.x, flag);
channels[bid].unpackPackets(2 * my_nghr_offset, my_nghr_offset, size_per_rank, tid, blockDim.x, flag);
} else {
channels[bid].put(my_offset, my_offset, size_per_rank, tid, blockDim.x);
__syncthreads();