mirror of
https://github.com/microsoft/mscclpp.git
synced 2026-04-19 22:39:11 +00:00
Add support for multicast reduce insruction (#316)
This commit is contained in:
@@ -41,8 +41,8 @@ extern "C" __global__ void __launch_bounds__(1024, 1)
|
||||
|
||||
for (int idx = my_st + my_offset; idx < my_en; idx += my_step) {
|
||||
uint4 val;
|
||||
DeviceMulticastPointerDeviceHandle::multimemLoad(val, mc_ptr + idx);
|
||||
DeviceMulticastPointerDeviceHandle::multimemStore(val, mc_ptr + idx);
|
||||
mscclpp::DeviceMulticastPointerDeviceHandle::multimemLoadReduce(val, mc_ptr + idx);
|
||||
mscclpp::DeviceMulticastPointerDeviceHandle::multimemStore(val, mc_ptr + idx);
|
||||
}
|
||||
|
||||
deviceSyncer.sync(gridDim.x);
|
||||
|
||||
Reference in New Issue
Block a user