Add support for CUDA 12.0, drop Kepler (sm_35).
Support for H100 features.
Make socket code more robust and protected. Solves #555.
Improve performance on large CUDA graphs, reducing dependencies.
Reduce inter-socket bandwidth on AMD CPUs to favor better paths.
Various fixes to ncclCommAbort.
Make service thread polling resistant to EINTR.
Compile with profiling API by default.
Extend NVTX instrumentation with call arguments.


[ROCm/rccl commit: 28189e2df8]
This commit is contained in:
Sylvain Jeaugey
2022-11-29 04:27:46 -08:00
rodzic 1bcb7a7cb6
commit 2ce8946622
46 zmienionych plików z 3325 dodań i 1037 usunięć
+14
Wyświetl plik
@@ -27,6 +27,7 @@ ncclResult_t getHostName(char* hostname, int maxlen, const char delim);
uint64_t getHash(const char* string, int n);
uint64_t getHostHash();
uint64_t getPidHash();
ncclResult_t getRandomData(void* buffer, size_t bytes);
struct netIf {
char prefix[64];
@@ -48,6 +49,19 @@ inline uint64_t clockNano() {
return uint64_t(ts.tv_sec)*1000*1000*1000 + ts.tv_nsec;
}
/* get any bytes of random data from /dev/urandom, return 0 if it succeeds; else
* return -1 */
inline ncclResult_t getRandomData(void* buffer, size_t bytes) {
ncclResult_t ret = ncclSuccess;
if (bytes > 0) {
const size_t one = 1UL;
FILE* fp = fopen("/dev/urandom", "r");
if (buffer == NULL || fp == NULL || fread(buffer, bytes, one, fp) != one) ret = ncclSystemError;
if (fp) fclose(fp);
}
return ret;
}
////////////////////////////////////////////////////////////////////////////////
template<typename Int>