add support for GPUs using wavefront size of 32 (#285)

* add gfx1100 support

Add support for Radeon 7900 GPUs (RX and PRO), and 7800 PRO.

I was contemplating to add gfx1101 and gfx1102 GPUs as well, but those are the lower end models that are more unlikely to be used for compute intensive jobs. In addition, I do not have access to them to test the support.

* update WF_SIZe for different options

Radeon systems use a WarpSize of 32, unlike current Instinct systems,
which use a warp size of 64. For the device side, a gfx specific ifdef
is sufficient. For the host side, we need to query the device
properties.

* adjust functional tests to wf_size of 32

* update unit tests to handle wf_size of 32

* address reviewer comments

[ROCm/rocshmem commit: d0c2845031]
This commit is contained in:
Edgar Gabriel
2025-10-22 16:04:58 -05:00
committed by GitHub
vanhempi b771a26916
commit d37af80d7e
19 muutettua tiedostoa jossa 192 lisäystä ja 56 poistoa
@@ -118,6 +118,15 @@ ExecTest() {
NUM_THREADS=$4
MAX_MSG_SIZE=$5
if command -v amd-smi >/dev/null
then
NUM_GPUS=$(amd-smi list | grep GPU | wc -l)
elif command -v rocm-smi >/dev/null
then
NUM_GPUS=$(rocm-smi --showserial | grep GPU | wc -l)
else
NUM_GPUS=64
fi
TIMEOUT=$((5 * 60)) # Timeout in seconds
TEST_NUM=${TEST_NUMBERS[$TEST_NAME]}
@@ -159,9 +168,13 @@ ExecTest() {
CMD+=" >> $LOG_DIR/$TEST_LOG_NAME.log 2>&1"
# Run Test
echo $TEST_LOG_NAME
echo "# $CMD" >"$LOG_DIR/$TEST_LOG_NAME.log"
eval $CMD
if [ $NUM_RANKS -le $NUM_GPUS ] && [[ "" == "$HOSTFILE" ]]; then
echo $TEST_LOG_NAME
echo "# $CMD" >"$LOG_DIR/$TEST_LOG_NAME.log"
eval $CMD
else
echo "Skipping test $TEST_LOG_NAME"
fi
# Validate Test
if [ $? -ne 0 ]