Port GLM-5.3 MI355X TileRT AgentX to srt-slurm

inferencex-e2e/benchmarks/multi_node/amd_utils/env.sh

@@ -2,21 +2,12 @@
 
 source "$(dirname "${BASH_SOURCE[0]}")/../../benchmark_lib.sh" --validation-only
 check_env_vars ENGINE
-# MoRI-IO queue-pair tuning, the UCX RoCE GID index, SGLang router logging and the
-# SGLang decode cuda-graph NCCL workaround. Only the SGLang MoRI KV path
-# below reads these. ENGINE=tilert moves KV over mooncake and starts no SGLang
-# router, so it is neither given nor reads them: validating them there would force
-# the recipe to invent MoRI tuning for a transport it never uses.
-if [[ "$ENGINE" != "tilert" ]]; then
-    check_env_vars \
-        MORI_IO_SQ_BACKOFF_TIMEOUT_US MORI_IO_QP_MAX_SEND_WR MORI_IO_QP_MAX_CQE MORI_IO_QP_MAX_SGE MORI_IO_TC_DISABLE \
-        UCX_IB_GID_INDEX MORI_APP_LOG_LEVEL SGLANG_ROUTER_STDOUT_LOGS TORCH_NCCL_BLOCKING_WAIT NCCL_BLOCKING_WAIT \
-        SGLANG_OPT_USE_AITER_INDEXER
-fi
-# Dual-engine environment setup for multi-node disaggregated serving.
-#
-# ENGINE=sglang-disagg or tilert selects the engine-specific block.
-#
+# SGLang MoRI environment.
+check_env_vars \
+    MORI_IO_SQ_BACKOFF_TIMEOUT_US MORI_IO_QP_MAX_SEND_WR MORI_IO_QP_MAX_CQE MORI_IO_QP_MAX_SGE MORI_IO_TC_DISABLE \
+    UCX_IB_GID_INDEX MORI_APP_LOG_LEVEL SGLANG_ROUTER_STDOUT_LOGS TORCH_NCCL_BLOCKING_WAIT NCCL_BLOCKING_WAIT \
+    SGLANG_OPT_USE_AITER_INDEXER
+
 # REQUIRED ENVIRONMENT VARIABLES:
 #   IBDEVICES - RDMA/InfiniBand device names (e.g., ionic_0,ionic_1,... or mlx5_0,mlx5_1,...)
 #               Set by runner or auto-detected from hostname.
@@ -119,10 +110,6 @@ else
     fi
 fi
 
-if [[ "$ENGINE" == "tilert" ]]; then
-    echo "[INFO] tilert: IBDEVICES=$IBDEVICES  NCCL_SOCKET_IFNAME=$NCCL_SOCKET_IFNAME  NCCL_IB_HCA=$NCCL_IB_HCA"
-
-else
 
     export SGLANG_USE_AITER=1
     export AITER_LOG_LEVEL=ERROR
@@ -237,5 +224,3 @@ else
             export GPU_MAX_HW_QUEUES=2
         fi
     fi
-
-fi

inferencex-e2e/benchmarks/multi_node/amd_utils/job.slurm

@@ -36,8 +36,6 @@ echo ""
 # at runtime, but the CWD remains the submit-time directory (amd_utils/).
 if [[ "$ENGINE" == "atom-disagg" ]]; then
     MODELS_YAML="$(pwd)/models_atom.yaml"
-elif [[ "$ENGINE" == "tilert" ]]; then
-    MODELS_YAML="$(pwd)/models_tilert.yaml"
 else
     MODELS_YAML="$(pwd)/models.yaml"
 fi
@@ -52,16 +50,6 @@ if [[ -z "${DOCKER_IMAGE_NAME:-}" ]]; then
     exit 1
 fi
 
-if [[ "$ENGINE" == "tilert" && -z "${PREFILL_IMAGE:-}" ]]; then
-    echo "Error: ENGINE=tilert requires PREFILL_IMAGE (e.g. PREFILL_IMAGE=vllm/vllm-openai-rocm:nightly-<sha> in prefill.additional-settings)."
-    exit 1
-fi
-if [[ "$ENGINE" == "tilert" ]]; then
-    # server_tilert.sh takes the container-creation barrier timeout from the
-    # recipe (no 300s default as on the SGLang path); fail here, before sbatch
-    # work is done, rather than inside the container.
-    check_env_vars CONTAINER_BARRIER_TIMEOUT
-fi
 
 # Resolve the models.yaml entry the same way server_sglang.sh does: agentic runs
 # (IS_AGENTIC) use the '<model>-AgentX' recipe, non-agentic disaggregated runs use
@@ -564,41 +552,6 @@ if [[ "$ENGINE" == "atom-disagg" ]]; then
         -e EXTRA_SERVER_ARGS=\${EXTRA_SERVER_ARGS:-}
         -e IBDEVICES=${IBDEVICES:-}
     )
-elif [[ "$ENGINE" == "tilert" ]]; then
-    DOCKER_ENV_ENGINE=(
-        -e MODEL_PATH=$DOCKER_MODEL_PATH
-        -e PREFILL_IMAGE=${PREFILL_IMAGE}
-        -e TILERT_VERSION=${TILERT_VERSION}
-        -e TILERT_PROFILE=${TILERT_PROFILE}
-        -e TILERT_MODEL_TYPE=${TILERT_MODEL_TYPE}
-        -e TILERT_MODEL_PKG=${TILERT_MODEL_PKG}
-        -e TILERT_MAX_MODEL_LEN=${TILERT_MAX_MODEL_LEN}
-        -e TILERT_TRANSPORT=${TILERT_TRANSPORT}
-        -e TILERT_PARSER=${TILERT_PARSER}
-        -e TILERT_QUEUE_TIMEOUT=${TILERT_QUEUE_TIMEOUT}
-        -e TILERT_WEIGHTS_DIR=${TILERT_WEIGHTS_DIR}
-        -e TILERT_RDMA_STRICT=${TILERT_RDMA_STRICT}
-        -e TILERT_CONVERT_LOCK_WAIT=${TILERT_CONVERT_LOCK_WAIT}
-        -e TILERT_SIMULATE_ACC_METHOD=${TILERT_SIMULATE_ACC_METHOD}
-        -e \"TILERT_EXTRA_ENV=${TILERT_EXTRA_ENV:-}\"
-        -e SERVED_MODEL_NAME=${SERVED_MODEL_NAME}
-        -e PREFILL_KV_DTYPE=${PREFILL_KV_DTYPE}
-        -e PREFILL_BLOCK_SIZE=${PREFILL_BLOCK_SIZE}
-        -e PREFILL_SPEC_TOKENS=${PREFILL_SPEC_TOKENS}
-        -e DECODE_KV_DTYPE=${DECODE_KV_DTYPE}
-        -e GPU_MEM_UTIL=${GPU_MEM_UTIL}
-        -e PREFILL_PORT=${PREFILL_PORT}
-        -e DECODE_CTRL_PORT=${DECODE_CTRL_PORT}
-        -e DECODE_HTTP_PORT=${DECODE_HTTP_PORT}
-        -e DECODE_WAIT=${DECODE_WAIT}
-        -e PREFILL_WAIT=${PREFILL_WAIT}
-        -e ROUTER_WAIT=${ROUTER_WAIT}
-        -e SKIP_CONTAINER_BARRIER=${SKIP_CONTAINER_BARRIER}
-        # Golden-acceptance selection on agentic MTP runs; unset elsewhere.
-        -e THINKING_MODE=${THINKING_MODE:-}
-        -e IBDEVICES=${IBDEVICES:-}
-        -e PYTHONPYCACHEPREFIX=/tmp/pycache
-    )
 else
     DOCKER_ENV_ENGINE=(
         -e SGLANG_WS_PATH=${WS_PATH}
@@ -751,12 +704,6 @@ else
 fi
 fi  # end: if ENGINE == atom-disagg
 
-RANK_IMAGE=
-if [[ \"$ENGINE\" == \"tilert\" && \"\$SLURM_PROCID\" -lt \"$xP\" ]]; then
-    RANK_IMAGE=\"$PREFILL_IMAGE\"
-    echo \"[tilert] rank \$SLURM_PROCID is a prefill rank; using PREFILL_IMAGE=\$RANK_IMAGE\"
-fi
-
 exec \$DOCKER_CMD run \
     --init \
     --stop-timeout 10 \
@@ -798,7 +745,7 @@ exec \$DOCKER_CMD run \
     ${CLIENT_DOCKER_ENV} \
     --name \"$DOCKER_CONT_NAME\" \
     --entrypoint \"\" \
-    \"\${RANK_IMAGE:-$DOCKER_IMAGE_NAME}\" bash -lc '
+    \"$DOCKER_IMAGE_NAME\" bash -lc '
         set -o pipefail
         mkdir -p /run_logs/slurm_job-'\"\$SLURM_JOB_ID\"'
         '"$RUN_FILE_FULL"' 2>&1 | tee /run_logs/slurm_job-'\"\$SLURM_JOB_ID\"'/server_\$(hostname).log

inferencex-e2e/benchmarks/multi_node/amd_utils/server.sh

@@ -5,7 +5,6 @@ source "$(dirname "${BASH_SOURCE[0]}")/../../benchmark_lib.sh" --validation-only
 # Dispatches to the engine-specific server launcher based on ENGINE env var.
 #   ENGINE=sglang-disagg (default) -> server_sglang.sh (SGLang + MoRI)
 #   ENGINE=atom-disagg             -> server_atom.sh (ATOM + mooncake)
-#   ENGINE=tilert                  -> server_tilert.sh (vLLM prefill + TileRT decode)
 
 check_env_vars ENGINE WS_PATH
 if [[ -f /config/hicache_mc.env ]]; then
@@ -20,8 +19,6 @@ echo "[DISPATCHER] ENGINE=$ENGINE  WS_PATH=$WS_PATH"
 if [[ "$ENGINE" == "atom-disagg" ]]; then
     export ATOM_WS_PATH="$WS_PATH"
     source "$WS_PATH/server_atom.sh"
-elif [[ "$ENGINE" == "tilert" ]]; then
-    source "$WS_PATH/server_tilert.sh"
 else
     source "$WS_PATH/server_sglang.sh"
 fi

inferencex-e2e/benchmarks/multi_node/amd_utils/server_atom.sh

@@ -16,7 +16,6 @@ check_env_vars \
 
 EXTRA_SERVER_ARGS="${EXTRA_SERVER_ARGS:-}"
 
-source $ATOM_WS_PATH/setup_deps.sh
 source $ATOM_WS_PATH/env_atom.sh
 
 # lm-eval with high num_concurrent exhausts the default 1024 FD limit.

inferencex-e2e/benchmarks/multi_node/amd_utils/server_sglang.sh

@@ -20,7 +20,6 @@ BENCH_MAX_CONC_VALUE=$(echo "$BENCH_MAX_CONCURRENCY" | tr 'x' '\n' | sort -n | t
 # can resolve formulas like "BENCH_MAX_CONC_VALUE*2" for max_running_requests.
 export BENCH_MAX_CONC_VALUE
 
-source $SGLANG_WS_PATH/setup_deps.sh
 source $SGLANG_WS_PATH/env.sh
 
 # Install before starting UMBP or serving processes. Early readiness failures must