NASP/cuda-samples
60
0
Fork 0
mirror of https://github.com/NVIDIA/cuda-samples.git synced 2026-04-02 15:45:41 +08:00
Code Issues Packages Projects Releases Wiki Activity

Bug 5809031: Explicitly adding P2H2P fallback as cudaMemcpyPeerAsync automatically fallback not supported with P2P off and CC mode on

Browse Source
This commit is contained in:
Shawn Zeng 2026-01-21 18:44:45 +08:00
parent 6c4d183ba3
commit 5bbae96fd3
1 changed files with 149 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

171
Samples/5_Domain_Specific/p2pBandwidthLatencyTest/p2pBandwidthLatencyTest.cu
View File

@ -101,6 +101,54 @@ void printHelp(void)
printf("--numElems=<NUM_OF_INT_ELEMS> Number of integer elements to be used in p2p copy.\n");
}
///////////////////////////////////////////////////////////////////////////
// Detect if cudaMemcpyPeerAsync will automatically fall back to
// host-staged copies when P2P is disabled.
//
// We probe a single representative pair (device 0 -> device 1).
// On a given system, confidential-computing (CC) and security
// policies are uniform across GPUs, so if this pair is blocked
// with cudaErrorNotSupported in P2P-off mode, it is reasonable
// to assume all cross-GPU pairs behave the same.
//
// For a production application that must handle heterogeneous
// environments, users may want to probe all device pairs.
///////////////////////////////////////////////////////////////////////////
bool detectFallback(int numGPUs)
{
if (numGPUs <= 1)
return false;
cudaSetDevice(0);
int *tmp0 = nullptr, *tmp1 = nullptr;
cudaStream_t s;
cudaStreamCreateWithFlags(&s, cudaStreamNonBlocking);
size_t testElems = 1; // 比如 1M ints (4MB)
cudaMalloc(&tmp0, testElems * sizeof(int));
cudaSetDevice(1);
cudaMalloc(&tmp1, testElems * sizeof(int));
cudaCheckError();
bool needsFallback = false;
cudaError_t testErr = cudaMemcpyPeerAsync(tmp1, 1, tmp0, 0, testElems * sizeof(int), s);
if (testErr == cudaErrorNotSupported) {
needsFallback = true;
printf("Note: cudaMemcpyPeerAsync reported '%s' - will use host-mediated copy when P2P is disabled\n",
cudaGetErrorString(testErr));
cudaGetLastError();
}
cudaStreamSynchronize(s);
cudaFree(tmp0);
cudaFree(tmp1);
cudaStreamDestroy(s);
cudaCheckError();
return needsFallback;
}
void checkP2Paccess(int numGPUs)
{
for (int i = 0; i < numGPUs; i++) {
@ -128,7 +176,9 @@ void performP2PCopy(int *dest,
int num_elems,
int repeat,
bool p2paccess,
cudaStream_t streamToRun)
cudaStream_t streamToRun,
bool useFallback,
int *hostBuffer)
{
int blockSize = 0;
int numBlocks = 0;
@ -140,15 +190,33 @@ void performP2PCopy(int *dest,
for (int r = 0; r < repeat; r++) {
copyp2p<<<numBlocks, blockSize, 0, streamToRun>>>((int4 *)dest, (int4 *)src, num_elems / 4);
}
cudaCheckError();
}
else if (useFallback && srcDevice != destDevice) {
// Use host-mediated copy for cross-GPU transfers when cudaMemcpyPeerAsync is not supported
for (int r = 0; r < repeat; r++) {
cudaMemcpyAsync(hostBuffer, src, sizeof(int) * num_elems, cudaMemcpyDeviceToHost, streamToRun);
cudaMemcpyAsync(dest, hostBuffer, sizeof(int) * num_elems, cudaMemcpyHostToDevice, streamToRun);
}
cudaCheckError();
}
else if (useFallback && srcDevice == destDevice) {
// Same device copy
for (int r = 0; r < repeat; r++) {
cudaMemcpyAsync(dest, src, sizeof(int) * num_elems, cudaMemcpyDeviceToDevice, streamToRun);
}
cudaCheckError();
}
else {
// Use cudaMemcpyPeerAsync (original behavior)
for (int r = 0; r < repeat; r++) {
cudaMemcpyPeerAsync(dest, destDevice, src, srcDevice, sizeof(int) * num_elems, streamToRun);
}
cudaCheckError();
}
}
void outputBandwidthMatrix(int numElems, int numGPUs, bool p2p, P2PDataTransfer p2p_method)
void outputBandwidthMatrix(int numElems, int numGPUs, bool p2p, P2PDataTransfer p2p_method, bool needsFallback)
{
int repeat = 5;
volatile int *flag = NULL;
@ -178,6 +246,16 @@ void outputBandwidthMatrix(int numElems, int numGPUs, bool p2p, P2PDataTransfer
cudaCheckError();
}
// Decide if fallback is needed based on global detection result and current test scenario
bool useFallback = (!p2p && needsFallback);
int *hostBuffer = nullptr;
if (useFallback) {
// Allocate hostBuffer for this function's numElems
cudaHostAlloc((void **)&hostBuffer, sizeof(int) * numElems, cudaHostAllocDefault);
cudaCheckError();
}
vector<double> bandwidthMatrix(numGPUs * numGPUs);
for (int i = 0; i < numGPUs; i++) {
@ -215,16 +293,20 @@ void outputBandwidthMatrix(int numElems, int numGPUs, bool p2p, P2PDataTransfer
if (i == j) {
// Perform intra-GPU, D2D copies
performP2PCopy(buffers[i], i, buffersD2D[i], i, numElems, repeat, access, stream[i]);
performP2PCopy(
buffers[i], i, buffersD2D[i], i, numElems, repeat, access, stream[i], useFallback, hostBuffer);
}
else {
if (p2p_method == P2P_WRITE) {
performP2PCopy(buffers[j], j, buffers[i], i, numElems, repeat, access, stream[i]);
performP2PCopy(
buffers[j], j, buffers[i], i, numElems, repeat, access, stream[i], useFallback, hostBuffer);
}
else {
performP2PCopy(buffers[i], i, buffers[j], j, numElems, repeat, access, stream[i]);
performP2PCopy(
buffers[i], i, buffers[j], j, numElems, repeat, access, stream[i], useFallback, hostBuffer);
}
}
cudaCheckError();
cudaEventRecord(stop[i], stream[i]);
cudaCheckError();
@ -284,11 +366,16 @@ void outputBandwidthMatrix(int numElems, int numGPUs, bool p2p, P2PDataTransfer
cudaCheckError();
}
if (hostBuffer) {
cudaFreeHost(hostBuffer);
cudaCheckError();
}
cudaFreeHost((void *)flag);
cudaCheckError();
}
void outputBidirectionalBandwidthMatrix(int numElems, int numGPUs, bool p2p)
void outputBidirectionalBandwidthMatrix(int numElems, int numGPUs, bool p2p, bool needsFallback)
{
int repeat = 5;
volatile int *flag = NULL;
@ -319,6 +406,16 @@ void outputBidirectionalBandwidthMatrix(int numElems, int numGPUs, bool p2p)
cudaCheckError();
}
// Decide if fallback is needed based on global detection result and current test scenario
bool useFallback = (!p2p && needsFallback);
int *hostBuffer = nullptr;
if (useFallback) {
// Allocate hostBuffer for this function's numElems
cudaHostAlloc((void **)&hostBuffer, sizeof(int) * numElems, cudaHostAllocDefault);
cudaCheckError();
}
vector<double> bandwidthMatrix(numGPUs * numGPUs);
for (int i = 0; i < numGPUs; i++) {
@ -362,18 +459,22 @@ void outputBidirectionalBandwidthMatrix(int numElems, int numGPUs, bool p2p)
if (i == j) {
// For intra-GPU perform 2 memcopies buffersD2D <-> buffers
performP2PCopy(buffers[i], i, buffersD2D[i], i, numElems, repeat, access, stream0[i]);
performP2PCopy(buffersD2D[i], i, buffers[i], i, numElems, repeat, access, stream1[i]);
performP2PCopy(
buffers[i], i, buffersD2D[i], i, numElems, repeat, access, stream0[i], useFallback, hostBuffer);
performP2PCopy(
buffersD2D[i], i, buffers[i], i, numElems, repeat, access, stream1[i], useFallback, hostBuffer);
}
else {
if (access && p2p_mechanism == SM) {
cudaSetDevice(j);
}
performP2PCopy(buffers[i], i, buffers[j], j, numElems, repeat, access, stream1[j]);
performP2PCopy(
buffers[i], i, buffers[j], j, numElems, repeat, access, stream1[j], useFallback, hostBuffer);
if (access && p2p_mechanism == SM) {
cudaSetDevice(i);
}
performP2PCopy(buffers[j], j, buffers[i], i, numElems, repeat, access, stream0[i]);
performP2PCopy(
buffers[j], j, buffers[i], i, numElems, repeat, access, stream0[i], useFallback, hostBuffer);
}
// Notify stream0 that stream1 is complete and record the time of
@ -439,11 +540,16 @@ void outputBidirectionalBandwidthMatrix(int numElems, int numGPUs, bool p2p)
cudaCheckError();
}
if (hostBuffer) {
cudaFreeHost(hostBuffer);
cudaCheckError();
}
cudaFreeHost((void *)flag);
cudaCheckError();
}
void outputLatencyMatrix(int numGPUs, bool p2p, P2PDataTransfer p2p_method)
void outputLatencyMatrix(int numGPUs, bool p2p, P2PDataTransfer p2p_method, bool needsFallback)
{
int repeat = 100;
int numElems = 4; // perform 1-int4 transfer.
@ -478,6 +584,16 @@ void outputLatencyMatrix(int numGPUs, bool p2p, P2PDataTransfer p2p_method)
cudaCheckError();
}
// Decide if fallback is needed based on global detection result and current test scenario
bool useFallback = (!p2p && needsFallback);
int *hostBuffer = nullptr;
if (useFallback) {
// Allocate hostBuffer for this function's numElems
cudaHostAlloc((void **)&hostBuffer, sizeof(int) * numElems, cudaHostAllocDefault);
cudaCheckError();
}
vector<double> gpuLatencyMatrix(numGPUs * numGPUs);
vector<double> cpuLatencyMatrix(numGPUs * numGPUs);
@ -513,14 +629,17 @@ void outputLatencyMatrix(int numGPUs, bool p2p, P2PDataTransfer p2p_method)
sdkResetTimer(&stopWatch);
if (i == j) {
// Perform intra-GPU, D2D copies
performP2PCopy(buffers[i], i, buffersD2D[i], i, numElems, repeat, access, stream[i]);
performP2PCopy(
buffers[i], i, buffersD2D[i], i, numElems, repeat, access, stream[i], useFallback, hostBuffer);
}
else {
if (p2p_method == P2P_WRITE) {
performP2PCopy(buffers[j], j, buffers[i], i, numElems, repeat, access, stream[i]);
performP2PCopy(
buffers[j], j, buffers[i], i, numElems, repeat, access, stream[i], useFallback, hostBuffer);
}
else {
performP2PCopy(buffers[i], i, buffers[j], j, numElems, repeat, access, stream[i]);
performP2PCopy(
buffers[i], i, buffers[j], j, numElems, repeat, access, stream[i], useFallback, hostBuffer);
}
}
float cpu_time_ms = sdkGetTimerValue(&stopWatch);
@ -597,6 +716,11 @@ void outputLatencyMatrix(int numGPUs, bool p2p, P2PDataTransfer p2p_method)
sdkDeleteTimer(&stopWatch);
if (hostBuffer) {
cudaFreeHost(hostBuffer);
cudaCheckError();
}
cudaFreeHost((void *)flag);
cudaCheckError();
}
@ -645,6 +769,9 @@ int main(int argc, char **argv)
checkP2Paccess(numGPUs);
// Environment detection: One-time check if cudaMemcpyPeerAsync is supported when P2P is disabled
bool needsFallback = detectFallback(numGPUs);
// Check peer-to-peer connectivity
printf("P2P Connectivity Matrix\n");
printf(" D\\D");
@ -671,25 +798,25 @@ int main(int argc, char **argv)
}
printf("Unidirectional P2P=Disabled Bandwidth Matrix (GB/s)\n");
outputBandwidthMatrix(numElems, numGPUs, false, P2P_WRITE);
outputBandwidthMatrix(numElems, numGPUs, false, P2P_WRITE, needsFallback);
printf("Unidirectional P2P=Enabled Bandwidth (P2P Writes) Matrix (GB/s)\n");
outputBandwidthMatrix(numElems, numGPUs, true, P2P_WRITE);
outputBandwidthMatrix(numElems, numGPUs, true, P2P_WRITE, needsFallback);
if (p2p_method == P2P_READ) {
printf("Unidirectional P2P=Enabled Bandwidth (P2P Reads) Matrix (GB/s)\n");
outputBandwidthMatrix(numElems, numGPUs, true, p2p_method);
outputBandwidthMatrix(numElems, numGPUs, true, p2p_method, needsFallback);
}
printf("Bidirectional P2P=Disabled Bandwidth Matrix (GB/s)\n");
outputBidirectionalBandwidthMatrix(numElems, numGPUs, false);
outputBidirectionalBandwidthMatrix(numElems, numGPUs, false, needsFallback);
printf("Bidirectional P2P=Enabled Bandwidth Matrix (GB/s)\n");
outputBidirectionalBandwidthMatrix(numElems, numGPUs, true);
outputBidirectionalBandwidthMatrix(numElems, numGPUs, true, needsFallback);
printf("P2P=Disabled Latency Matrix (us)\n");
outputLatencyMatrix(numGPUs, false, P2P_WRITE);
outputLatencyMatrix(numGPUs, false, P2P_WRITE, needsFallback);
printf("P2P=Enabled Latency (P2P Writes) Matrix (us)\n");
outputLatencyMatrix(numGPUs, true, P2P_WRITE);
outputLatencyMatrix(numGPUs, true, P2P_WRITE, needsFallback);
if (p2p_method == P2P_READ) {
printf("P2P=Enabled Latency (P2P Reads) Matrix (us)\n");
outputLatencyMatrix(numGPUs, true, p2p_method);
outputLatencyMatrix(numGPUs, true, p2p_method, needsFallback);
}
printf("\nNOTE: The CUDA Samples are not meant for performance measurements. "