A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 13, 05:45 UTC
[ci][amd] fix basic models extra init test (#28676)
Signed-off-by: Bradley Davis <bradleyhd@meta.com>
Bradley D committed
Nov 13, 2025 at 18:44 UTC
b39a5026ebac9242740e48debc79ce8db92c868b
1 file changed
+5
-5
.buildkite/scripts/hardware_ci/run-amd-test.sh
+5
-5
@@ -59,7 +59,7 @@ while true; do
59
fi
60
done
61
62
-echo "--- Pulling container"
62
+echo "--- Pulling container"
63
image_name="rocm/vllm-ci:${BUILDKITE_COMMIT}"
64
container_name="rocm_${BUILDKITE_COMMIT}_$(tr -dc A-Za-z0-9 < /dev/urandom | head -c 10; echo)"
65
docker pull "${image_name}"
@@ -177,13 +177,13 @@ if [[ -z "$render_gid" ]]; then
177
exit 1
178
fi
179
180
-# check if the command contains shard flag, we will run all shards in parallel because the host have 8 GPUs.
180
+# check if the command contains shard flag, we will run all shards in parallel because the host have 8 GPUs.
181
if [[ $commands == *"--shard-id="* ]]; then
182
- # assign job count as the number of shards used
183
- commands=${commands//"--num-shards= "/"--num-shards=${PARALLEL_JOB_COUNT} "}
182
+ # assign job count as the number of shards used
183
+ commands=$(echo "$commands" | sed -E "s/--num-shards[[:blank:]]*=[[:blank:]]*[0-9]*/--num-shards=${PARALLEL_JOB_COUNT} /g" | sed 's/ \\ / /g')
184
for GPU in $(seq 0 $(($PARALLEL_JOB_COUNT-1))); do
185
# assign shard-id for each shard
186
- commands_gpu=${commands//"--shard-id= "/"--shard-id=${GPU} "}
186
+ commands_gpu=$(echo "$commands" | sed -E "s/--shard-id[[:blank:]]*=[[:blank:]]*[0-9]*/--shard-id=${GPU} /g" | sed 's/ \\ / /g')
187
echo "Shard ${GPU} commands:$commands_gpu"
188
echo "Render devices: $BUILDKITE_AGENT_META_DATA_RENDER_DEVICES"
189
docker run \