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

SWDEV-510419: Restore compute partition after memory partition test (#15)

Browse Source 
Memory partition test was changing original compute partiton based
on default compute mode. Corrected this to set back to original
compute partition.

Signed-off-by: Bindhiya Kanangot Balakrishnan <Bindhiya.KanangotBalakrishnan@amd.com>
This commit is contained in:
Kanangot Balakrishnan, Bindhiya
2025-03-05 23:46:17 -06:00
committed by Charis Poag
parent 1dd9ca9df4
commit 1d6b8d9422
1 changed files with 68 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tests/rocm_smi_test/functional/memorypartition_read_write.cc
+68
View File
@@ -110,6 +110,15 @@ mapStringToRSMIMemoryPartitionTypes {
{"NPS8", RSMI_MEMORY_PARTITION_NPS8}
};
static const std::map<std::string, rsmi_compute_partition_type_t>
mapStringToRSMIComputePartitionTypes {
{"DPX", RSMI_COMPUTE_PARTITION_DPX},
{"TPX", RSMI_COMPUTE_PARTITION_TPX},
{"QPX", RSMI_COMPUTE_PARTITION_QPX},
{"CPX", RSMI_COMPUTE_PARTITION_CPX},
{"SPX", RSMI_COMPUTE_PARTITION_SPX}
};
void TestMemoryPartitionReadWrite::Run(void) {
rsmi_status_t ret, err, ret_set;
char orig_memory_partition[255];
@@ -125,6 +134,41 @@ void TestMemoryPartitionReadWrite::Run(void) {
return;
}
// initial_num_devices - keep this value static, due to parition changes
// fluctuating # of devices. We should end up with same # of devices at
// end of test.
uint32_t initial_num_devices = num_monitor_devs();
char orig_char_computePartition[initial_num_devices][255];
char current_char_computePartition[255];
for (uint32_t dv_ind = 0; dv_ind < num_monitor_devs(); ++dv_ind) {
// Run and get compute_partition, so we can reset to later
ret = rsmi_dev_compute_partition_get(dv_ind, orig_char_computePartition[dv_ind],
255);
if(ret == RSMI_STATUS_SETTING_UNAVAILABLE
|| ret== RSMI_STATUS_PERMISSION
|| ret == RSMI_STATUS_BUSY
|| ret == RSMI_STATUS_NOT_SUPPORTED
|| ret == RSMI_STATUS_INVALID_ARGS) {
IF_VERB(STANDARD) {
std::cout << "\t**rsmi_dev_compute_partition_get(): "
<< "Not supported on this device"
<< std::endl;
}
continue;
}
std::cout << "\t**rsmi_dev_compute_partition_get(" << dv_ind
<< ", " << orig_char_computePartition[dv_ind] << ")\n";
ret = rsmi_dev_compute_partition_get(dv_ind, orig_char_computePartition[dv_ind], 255);
IF_VERB(STANDARD) {
std::cout << std::endl << "\t**"
<< "Original compute partition: "
<< orig_char_computePartition[dv_ind] << std::endl;
}
}
for (uint32_t dv_ind = 0; dv_ind < num_monitor_devs(); ++dv_ind) {
bool wasSetSuccess = false;
if (dv_ind != 0) {
@@ -408,4 +452,28 @@ void TestMemoryPartitionReadWrite::Run(void) {
<< current_memory_partition << ")" << std::endl;
}
}
// Return to original compute_partition
for (uint32_t dv_ind = 0; dv_ind < num_monitor_devs(); ++dv_ind) {
ret = rsmi_dev_compute_partition_get(dv_ind, current_char_computePartition, 255);
if(ret == RSMI_STATUS_SETTING_UNAVAILABLE
|| ret== RSMI_STATUS_PERMISSION
|| ret == RSMI_STATUS_BUSY
|| ret == RSMI_STATUS_NOT_SUPPORTED
|| ret == RSMI_STATUS_INVALID_ARGS)
continue;
if (strcmp(orig_char_computePartition[dv_ind], current_char_computePartition) != 0) {
rsmi_compute_partition_type_t newPartition
= mapStringToRSMIComputePartitionTypes.at(
std::string(orig_char_computePartition[dv_ind]));
ret = rsmi_dev_compute_partition_set(dv_ind, newPartition);
IF_VERB(STANDARD) {
std::cout << "\t**"
<< "Returning compute partition to: "
<< std::string(orig_char_computePartition[dv_ind]) << std::endl;
}
}
}
}