zhyang-dev/openfoam
1
0
Fork 0
mirror of https://develop.openfoam.com/Development/openfoam.git synced 2025-11-28 03:28:01 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

ENH: globalIndex with direct gather/broadcast

Browse Source 
- less communication than gatherList/scatterList

ENH: refine send granularity in Pstream::exchange

STYLE: ensure PstreamBuffers and defaultCommsType agree

- simpler loops for lduSchedule
This commit is contained in:
Mark Olesen
2022-03-07 09:40:13 +01:00
parent b8c3dc4e49
commit 0cf02eb384
18 changed files with 275 additions and 196 deletions
Download Patch File Download Diff File Expand all files Collapse all files

51
applications/test/parallel-chunks/Test-parallel-chunks.C
View File

@ -210,7 +210,7 @@ void broadcast_chunks
<< "Contiguous data only." << sizeof(T) << Foam::abort(FatalError);
}
if (UPstream::maxCommsSize <= int(sizeof(T)))
if (UPstream::maxCommsSize <= 0)
{
// Do in one go
Info<< "send " << sendData.size() << " elements in one go" << endl;
@ -230,48 +230,61 @@ void broadcast_chunks
// guaranteed that some processor's sending size is some other
// processor's receive size. Also we can ignore any local comms.
// We need to send bytes so the number of iterations:
// We need to send chunks so the number of iterations:
// maxChunkSize iterations
// ------------ ----------
// 0 0
// 1..maxChunkSize 1
// maxChunkSize+1..2*maxChunkSize 2
// etc.
// ...
const label maxChunkSize(UPstream::maxCommsSize/sizeof(T));
const label maxChunkSize
(
max
(
static_cast<label>(1),
static_cast<label>(UPstream::maxCommsSize/sizeof(T))
)
);
label nIter(0);
label nChunks(0);
{
label nSendMax = 0;
// Get max send count (elements)
// forAll(sendBufs, proci)
// {
// if (proci != Pstream::myProcNo(comm))
// {
// nSendMax = max(nSendMax, sendBufs[proci].size());
// nChunks = max(nChunks, sendBufs[proci].size());
// }
// }
nSendMax = sendSize;
nChunks = sendSize;
if (nSendMax)
// Convert from send count (elements) to number of chunks.
// Can normally calculate with (count-1), but add some safety
if (nChunks)
{
nIter = 1 + ((nSendMax-1)/maxChunkSize);
nChunks = 1 + (nChunks/maxChunkSize);
}
reduce(nIter, maxOp<label>(), tag, comm);
reduce(nChunks, maxOp<label>(), tag, comm);
Info
<< "send " << nSendMax << " elements ("
<< (nSendMax*sizeof(T)) << " bytes) in " << nIter
<< " iterations of " << maxChunkSize << " chunks ("
<< "send " << sendSize << " elements ("
<< (sendSize*sizeof(T)) << " bytes) in " << nChunks
<< " chunks of " << maxChunkSize << " elements ("
<< (maxChunkSize*sizeof(T)) << " bytes) for maxCommsSize:"
<< Pstream::maxCommsSize
<< endl;
}
// stress-test with shortened sendSize
// will produce useless loops, but no calls
// sendSize /= 2;
label nSend(0);
label startSend(0);
char* charPtrSend;
for (label iter = 0; iter < nIter; ++iter)
for (label iter = 0; iter < nChunks; ++iter)
{
nSend = min
(
@ -297,6 +310,8 @@ void broadcast_chunks
startSend += nSend;
}
}
Info<< "final: " << startSend << endl;
}
@ -305,8 +320,9 @@ void broadcast_chunks
int main(int argc, char *argv[])
{
argList::noCheckProcessorDirectories();
argList::addOption("comms-size", "int", "override Pstream::maxCommsSize");
#include "setRootCase.H"
#include "createTime.H"
if (!Pstream::parRun())
{
@ -322,6 +338,9 @@ int main(int argc, char *argv[])
broadcast_chunks<labelList, label>(input1);
Pstream::maxCommsSize = 33;
args.readIfPresent("comms-size", Pstream::maxCommsSize);
broadcast_chunks<labelList, label>(input1);
// Mostly the same with PstreamBuffers