AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Support GPU memory test and compute queue test using Rocr

Browse Source 
A new diagnostic module librdc_rocr.so is created. The
module uses Rocr to test the memory allocation, memory access
and compute queue ready status.

Change-Id: I9098f4fc3209bf381b7cb3658a4e94c2e22f2fe9
This commit is contained in:
Bill(Shuzhou) Liu
2021-09-16 15:38:37 -04:00
committed by Shuzhou Liu
parent 6ab71e1a4a
commit 78e2f2486b
74 changed files with 4811 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files
example/diagnostic_example.cc
+2 -3
View File
@@ -31,9 +31,7 @@ THE SOFTWARE.
static std::string get_test_name(rdc_diag_test_cases_t test_case) {
const std::map<rdc_diag_test_cases_t, std::string> test_desc = {
{RDC_DIAG_COMPUTE_PROCESS, "No compute process"},
{RDC_DIAG_SDMA_QUEUE, "SDMA Queue ready"},
{RDC_DIAG_COMPUTE_QUEUE, "Compute Queue ready"},
{RDC_DIAG_VRAM_CHECK, "VRAM check"},
{RDC_DIAG_SYS_MEM_CHECK, "System memory check"},
{RDC_DIAG_NODE_TOPOLOGY, "Node topology check"},
{RDC_DIAG_GPU_PARAMETERS, "GPU parameters check"},
@@ -160,7 +158,8 @@ int main(int, char **) {
std::cout << std::setw(22) << std::left
<< get_test_name(RDC_DIAG_COMPUTE_PROCESS) + ":"
<< rdc_diagnostic_result_string(test_result.status) << "\n";
<< test_result.info << "\n";
// Cleanup consists of shutting down RDC.
cleanup: