From 325b902799404f6c9eb07f26cc0a6bf8b95761a1 Mon Sep 17 00:00:00 2001
From: seungrokj <seungrok.jung@amd.com>
Date: Fri, 27 Mar 2026 16:23:20 +0000
Subject: [PATCH 1/6] ATOM support for new models: Kimi-K2.5 FP4, GLM-5 FP8,
 and MiniMax-M2.5 FP8

Signed-off-by: seungrokj <seungrok.jung@amd.com>
---
 .github/configs/amd-master.yaml               | 75 +++++++++++++++++
 .../single_node/glm5_fp8_mi355x_atom.sh       | 79 ++++++++++++++++++
 .../single_node/kimik2.5_fp4_mi355x_atom.sh   | 80 +++++++++++++++++++
 .../minimaxm2.5_fp8_mi355x_atom.sh            | 80 +++++++++++++++++++
 perf-changelog.yaml                           |  9 +++
 5 files changed, 323 insertions(+)
 create mode 100644 benchmarks/single_node/glm5_fp8_mi355x_atom.sh
 create mode 100644 benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
 create mode 100644 benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh

diff --git a/.github/configs/amd-master.yaml b/.github/configs/amd-master.yaml
index b9560803e..2ea215fbe 100644
--- a/.github/configs/amd-master.yaml
+++ b/.github/configs/amd-master.yaml
@@ -293,6 +293,28 @@ glm5-fp8-mi355x-sglang:
     search-space:
     - { tp: 8, conc-start: 4, conc-end: 64 }
 
+glm5-fp8-mi355x-atom:
+  image: TBD
+  model: zai-org/GLM-5-FP8
+  model-prefix: glm5
+  runner: mi355x
+  precision: fp8
+  framework: atom
+  multinode: false
+  seq-len-configs:
+  - isl: 1024
+    osl: 1024
+    search-space:
+    - { tp: 8, conc-start: 4, conc-end: 128 }
+  - isl: 1024
+    osl: 8192
+    search-space:
+    - { tp: 8, conc-start: 4, conc-end: 128 }
+  - isl: 8192
+    osl: 1024
+    search-space:
+    - { tp: 8, conc-start: 4, conc-end: 128 }
+
 kimik2.5-int4-mi355x-vllm:
   image: vllm/vllm-openai-rocm:v0.18.0
   model: moonshotai/Kimi-K2.5
@@ -363,6 +385,31 @@ kimik2.5-fp4-mi355x-vllm:
     - { tp: 8, conc-start: 4, conc-end: 64 }
     - { tp: 4, conc-start: 4, conc-end: 64 }
 
+kimik2.5-fp4-mi355x-atom:
+  image: TBD
+  model: amd/Kimi-K2.5-MXFP4
+  model-prefix: kimik2.5
+  runner: mi355x
+  precision: fp4
+  framework: atom
+  multinode: false
+  seq-len-configs:
+  - isl: 1024
+    osl: 1024
+    search-space:
+    - { tp: 8, conc-start: 4, conc-end: 128 }
+    - { tp: 4, conc-start: 4, conc-end: 128 }
+  - isl: 1024
+    osl: 8192
+    search-space:
+    - { tp: 8, conc-start: 4, conc-end: 128 }
+    - { tp: 4, conc-start: 4, conc-end: 128 }
+  - isl: 8192
+    osl: 1024
+    search-space:
+    - { tp: 8, conc-start: 4, conc-end: 128 }
+    - { tp: 4, conc-start: 4, conc-end: 128 }
+
 minimaxm2.5-fp8-mi355x-vllm:
   image: vllm/vllm-openai-rocm:v0.18.0
   model: MiniMaxAI/MiniMax-M2.5
@@ -391,6 +438,34 @@ minimaxm2.5-fp8-mi355x-vllm:
     - { tp: 4, conc-start: 4, conc-end: 64 }
     - { tp: 8, ep: 8, conc-start: 32, conc-end: 256 }
 
+minimaxm2.5-fp8-mi355x-atom:
+  image: TBD
+  model: MiniMaxAI/MiniMax-M2.5
+  model-prefix: minimaxm2.5
+  runner: mi355x
+  precision: fp8
+  framework: atom
+  multinode: false
+  seq-len-configs:
+  - isl: 1024
+    osl: 1024
+    search-space:
+    - { tp: 2, conc-start: 4, conc-end: 128 }
+    - { tp: 4, conc-start: 4, conc-end: 128 }
+    - { tp: 8, ep: 8, conc-start: 32, conc-end: 256 }
+  - isl: 1024
+    osl: 8192
+    search-space:
+    - { tp: 2, conc-start: 4, conc-end: 128 }
+    - { tp: 4, conc-start: 4, conc-end: 128 }
+    - { tp: 8, ep: 8, conc-start: 32, conc-end: 256 }
+  - isl: 8192
+    osl: 1024
+    search-space:
+    - { tp: 2, conc-start: 4, conc-end: 128 }
+    - { tp: 4, conc-start: 4, conc-end: 128 }
+    - { tp: 8, ep: 8, conc-start: 32, conc-end: 256 }
+
 minimaxm2.5-fp8-mi300x-vllm:
   image: vllm/vllm-openai-rocm:v0.16.0
   model: MiniMaxAI/MiniMax-M2.5
diff --git a/benchmarks/single_node/glm5_fp8_mi355x_atom.sh b/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
new file mode 100644
index 000000000..960363bb6
--- /dev/null
+++ b/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
@@ -0,0 +1,79 @@
+#!/usr/bin/env bash
+
+source "$(dirname "$0")/../benchmark_lib.sh"
+
+check_env_vars \
+    MODEL \
+    TP \
+    CONC \
+    ISL \
+    OSL \
+    RANDOM_RANGE_RATIO \
+    RESULT_FILENAME \
+    EP_SIZE \
+    DP_ATTENTION
+
+if [[ -n "$SLURM_JOB_ID" ]]; then
+  echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
+fi
+
+echo "TP: $TP, CONC: $CONC, ISL: $ISL, OSL: $OSL, EP_SIZE: $EP_SIZE, DP_ATTENTION: $DP_ATTENTION"
+
+SERVER_LOG=/workspace/server.log
+PORT=${PORT:-8888}
+
+export OMP_NUM_THREADS=1
+
+# Calculate max-model-len based on ISL and OSL
+if [ "$ISL" = "1024" ] && [ "$OSL" = "1024" ]; then
+    CALCULATED_MAX_MODEL_LEN=""
+else
+    CALCULATED_MAX_MODEL_LEN=" --max-model-len 10240 "
+fi
+
+if [ "$EP_SIZE" -gt 1 ]; then
+  EP=" --enable-expert-parallel"
+else
+  EP=" "
+fi
+
+# Start GPU monitoring (power, temperature, clocks every second)
+start_gpu_monitor
+
+set -x
+
+python3 -m atom.entrypoints.openai_server \
+    --model $MODEL \
+    --server-port $PORT \
+    -tp $TP \
+    --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
+    --trust-remote-code \
+    > $SERVER_LOG 2>&1 &
+
+SERVER_PID=$!
+
+# Wait for server to be ready
+wait_for_server_ready --port "$PORT" --server-log "$SERVER_LOG" --server-pid "$SERVER_PID"
+
+export PYTHONDONTWRITEBYTECODE=1
+run_benchmark_serving \
+    --model "$MODEL" \
+    --port "$PORT" \
+    --backend vllm \
+    --input-len "$ISL" \
+    --output-len "$OSL" \
+    --random-range-ratio "$RANDOM_RANGE_RATIO" \
+    --num-prompts "$((CONC * 10))" \
+    --max-concurrency "$CONC" \
+    --result-filename "$RESULT_FILENAME" \
+    --result-dir /workspace/
+
+# After throughput, run evaluation only if RUN_EVAL is true
+if [ "${RUN_EVAL}" = "true" ]; then
+    run_eval --framework lm-eval --port "$PORT" --concurrent-requests $CONC
+    append_lm_eval_summary
+fi
+
+# Stop GPU monitoring
+stop_gpu_monitor
+set +x
diff --git a/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh b/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
new file mode 100644
index 000000000..8f94c3829
--- /dev/null
+++ b/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
@@ -0,0 +1,80 @@
+#!/usr/bin/env bash
+
+source "$(dirname "$0")/../benchmark_lib.sh"
+
+check_env_vars \
+    MODEL \
+    TP \
+    CONC \
+    ISL \
+    OSL \
+    RANDOM_RANGE_RATIO \
+    RESULT_FILENAME \
+    EP_SIZE \
+    DP_ATTENTION
+
+if [[ -n "$SLURM_JOB_ID" ]]; then
+  echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
+fi
+
+echo "TP: $TP, CONC: $CONC, ISL: $ISL, OSL: $OSL, EP_SIZE: $EP_SIZE, DP_ATTENTION: $DP_ATTENTION"
+
+SERVER_LOG=/workspace/server.log
+PORT=${PORT:-8888}
+
+export OMP_NUM_THREADS=1
+
+# Calculate max-model-len based on ISL and OSL
+if [ "$ISL" = "1024" ] && [ "$OSL" = "1024" ]; then
+    CALCULATED_MAX_MODEL_LEN=""
+else
+    CALCULATED_MAX_MODEL_LEN=" --max-model-len 10240 "
+fi
+
+if [ "$EP_SIZE" -gt 1 ]; then
+  EP=" --enable-expert-parallel"
+else
+  EP=" "
+fi
+
+# Start GPU monitoring (power, temperature, clocks every second)
+start_gpu_monitor
+
+set -x
+
+python3 -m atom.entrypoints.openai_server \
+    --model $MODEL \
+    --server-port $PORT \
+    -tp $TP \
+    --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
+    --trust-remote-code \
+    > $SERVER_LOG 2>&1 &
+
+SERVER_PID=$!
+
+# Wait for server to be ready
+wait_for_server_ready --port "$PORT" --server-log "$SERVER_LOG" --server-pid "$SERVER_PID"
+
+export PYTHONDONTWRITEBYTECODE=1
+run_benchmark_serving \
+    --model "$MODEL" \
+    --port "$PORT" \
+    --backend vllm \
+    --input-len "$ISL" \
+    --output-len "$OSL" \
+    --random-range-ratio "$RANDOM_RANGE_RATIO" \
+    --num-prompts "$((CONC * 10))" \
+    --max-concurrency "$CONC" \
+    --result-filename "$RESULT_FILENAME" \
+    --result-dir /workspace/ \
+    --trust-remote-code
+
+# After throughput, run evaluation only if RUN_EVAL is true
+if [ "${RUN_EVAL}" = "true" ]; then
+    run_eval --framework lm-eval --port "$PORT" --concurrent-requests $CONC
+    append_lm_eval_summary
+fi
+
+# Stop GPU monitoring
+stop_gpu_monitor
+set +x
diff --git a/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh b/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
new file mode 100644
index 000000000..8f94c3829
--- /dev/null
+++ b/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
@@ -0,0 +1,80 @@
+#!/usr/bin/env bash
+
+source "$(dirname "$0")/../benchmark_lib.sh"
+
+check_env_vars \
+    MODEL \
+    TP \
+    CONC \
+    ISL \
+    OSL \
+    RANDOM_RANGE_RATIO \
+    RESULT_FILENAME \
+    EP_SIZE \
+    DP_ATTENTION
+
+if [[ -n "$SLURM_JOB_ID" ]]; then
+  echo "JOB $SLURM_JOB_ID running on $SLURMD_NODENAME"
+fi
+
+echo "TP: $TP, CONC: $CONC, ISL: $ISL, OSL: $OSL, EP_SIZE: $EP_SIZE, DP_ATTENTION: $DP_ATTENTION"
+
+SERVER_LOG=/workspace/server.log
+PORT=${PORT:-8888}
+
+export OMP_NUM_THREADS=1
+
+# Calculate max-model-len based on ISL and OSL
+if [ "$ISL" = "1024" ] && [ "$OSL" = "1024" ]; then
+    CALCULATED_MAX_MODEL_LEN=""
+else
+    CALCULATED_MAX_MODEL_LEN=" --max-model-len 10240 "
+fi
+
+if [ "$EP_SIZE" -gt 1 ]; then
+  EP=" --enable-expert-parallel"
+else
+  EP=" "
+fi
+
+# Start GPU monitoring (power, temperature, clocks every second)
+start_gpu_monitor
+
+set -x
+
+python3 -m atom.entrypoints.openai_server \
+    --model $MODEL \
+    --server-port $PORT \
+    -tp $TP \
+    --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
+    --trust-remote-code \
+    > $SERVER_LOG 2>&1 &
+
+SERVER_PID=$!
+
+# Wait for server to be ready
+wait_for_server_ready --port "$PORT" --server-log "$SERVER_LOG" --server-pid "$SERVER_PID"
+
+export PYTHONDONTWRITEBYTECODE=1
+run_benchmark_serving \
+    --model "$MODEL" \
+    --port "$PORT" \
+    --backend vllm \
+    --input-len "$ISL" \
+    --output-len "$OSL" \
+    --random-range-ratio "$RANDOM_RANGE_RATIO" \
+    --num-prompts "$((CONC * 10))" \
+    --max-concurrency "$CONC" \
+    --result-filename "$RESULT_FILENAME" \
+    --result-dir /workspace/ \
+    --trust-remote-code
+
+# After throughput, run evaluation only if RUN_EVAL is true
+if [ "${RUN_EVAL}" = "true" ]; then
+    run_eval --framework lm-eval --port "$PORT" --concurrent-requests $CONC
+    append_lm_eval_summary
+fi
+
+# Stop GPU monitoring
+stop_gpu_monitor
+set +x
diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 6a8a6e666..54ec15fc4 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -1108,3 +1108,12 @@
   description:
     - "Update vLLM image from v0.15.1 to v0.18.0 for gptoss H100 and H200 configs"
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/960
+
+- config-keys:
+    - kimik2.5-fp4-mi355x-atom
+    - glm5-fp8-mi355x-atom
+    - minimaxm2.5-fp8-mi355x-atom
+  description:
+    - "New model support on ATOM framework"
+    - "Kimi-K2.5 FP4, GLM-5 FP8, and MiniMax-M2.5 FP8 configs added for MI355X ATOM"
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/954

From 4ed05744f6dc729122fd5265bfd0dea2fc703895 Mon Sep 17 00:00:00 2001
From: seungrokj <seungrok.jung@amd.com>
Date: Fri, 27 Mar 2026 16:28:24 +0000
Subject: [PATCH 2/6] ATOM support for new models: Kimi-K2.5 FP4, GLM-5 FP8,
 and MiniMax-M2.5 FP8

Signed-off-by: seungrokj <seungrok.jung@amd.com>
---
 perf-changelog.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 54ec15fc4..610c4ef96 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -1116,4 +1116,4 @@
   description:
     - "New model support on ATOM framework"
     - "Kimi-K2.5 FP4, GLM-5 FP8, and MiniMax-M2.5 FP8 configs added for MI355X ATOM"
-  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/954
+  pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/963

From 4c4af7892894c96feee98d17b7223b60038e3814 Mon Sep 17 00:00:00 2001
From: seungrokj <seungrok.jung@amd.com>
Date: Mon, 30 Mar 2026 04:32:39 +0000
Subject: [PATCH 3/6] ATOM support for new models: Kimi-K2.5 FP4, GLM-5 FP8,
 and MiniMax-M2.5 FP8

Signed-off-by: seungrokj <seungrok.jung@amd.com>
---
 .github/configs/amd-master.yaml                       | 6 +++---
 benchmarks/single_node/glm5_fp8_mi355x_atom.sh        | 1 +
 benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh    | 1 +
 benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh | 1 +
 4 files changed, 6 insertions(+), 3 deletions(-)
 mode change 100644 => 100755 .github/configs/amd-master.yaml
 mode change 100644 => 100755 benchmarks/single_node/glm5_fp8_mi355x_atom.sh
 mode change 100644 => 100755 benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
 mode change 100644 => 100755 benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh

diff --git a/.github/configs/amd-master.yaml b/.github/configs/amd-master.yaml
old mode 100644
new mode 100755
index 2ea215fbe..e6d9869f2
--- a/.github/configs/amd-master.yaml
+++ b/.github/configs/amd-master.yaml
@@ -294,7 +294,7 @@ glm5-fp8-mi355x-sglang:
     - { tp: 8, conc-start: 4, conc-end: 64 }
 
 glm5-fp8-mi355x-atom:
-  image: TBD
+  image: rocm/atom:rocm7.2.1-ubuntu24.04-pytorch2.9.1-atom0.1.2
   model: zai-org/GLM-5-FP8
   model-prefix: glm5
   runner: mi355x
@@ -386,7 +386,7 @@ kimik2.5-fp4-mi355x-vllm:
     - { tp: 4, conc-start: 4, conc-end: 64 }
 
 kimik2.5-fp4-mi355x-atom:
-  image: TBD
+  image: rocm/atom:rocm7.2.1-ubuntu24.04-pytorch2.9.1-atom0.1.2
   model: amd/Kimi-K2.5-MXFP4
   model-prefix: kimik2.5
   runner: mi355x
@@ -439,7 +439,7 @@ minimaxm2.5-fp8-mi355x-vllm:
     - { tp: 8, ep: 8, conc-start: 32, conc-end: 256 }
 
 minimaxm2.5-fp8-mi355x-atom:
-  image: TBD
+  image: rocm/atom:rocm7.2.1-ubuntu24.04-pytorch2.9.1-atom0.1.2
   model: MiniMaxAI/MiniMax-M2.5
   model-prefix: minimaxm2.5
   runner: mi355x
diff --git a/benchmarks/single_node/glm5_fp8_mi355x_atom.sh b/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
old mode 100644
new mode 100755
index 960363bb6..5377e4041
--- a/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
+++ b/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
@@ -48,6 +48,7 @@ python3 -m atom.entrypoints.openai_server \
     -tp $TP \
     --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
     --trust-remote-code \
+    --no-enable-prefix-caching \
     > $SERVER_LOG 2>&1 &
 
 SERVER_PID=$!
diff --git a/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh b/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
old mode 100644
new mode 100755
index 8f94c3829..d47b2fc74
--- a/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
+++ b/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
@@ -48,6 +48,7 @@ python3 -m atom.entrypoints.openai_server \
     -tp $TP \
     --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
     --trust-remote-code \
+    --no-enable-prefix-caching \
     > $SERVER_LOG 2>&1 &
 
 SERVER_PID=$!
diff --git a/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh b/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
old mode 100644
new mode 100755
index 8f94c3829..d47b2fc74
--- a/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
+++ b/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
@@ -48,6 +48,7 @@ python3 -m atom.entrypoints.openai_server \
     -tp $TP \
     --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
     --trust-remote-code \
+    --no-enable-prefix-caching \
     > $SERVER_LOG 2>&1 &
 
 SERVER_PID=$!

From 371cc4af87057384a391496f00a6305f4a73dc9d Mon Sep 17 00:00:00 2001
From: seungrokj <seungrok.jung@amd.com>
Date: Mon, 30 Mar 2026 04:44:08 +0000
Subject: [PATCH 4/6] ATOM support for new models: Kimi-K2.5 FP4, GLM-5 FP8,
 and MiniMax-M2.5 FP8

Signed-off-by: seungrokj <seungrok.jung@amd.com>
---
 benchmarks/single_node/glm5_fp8_mi355x_atom.sh        | 1 -
 benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh    | 1 -
 benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh | 1 -
 3 files changed, 3 deletions(-)

diff --git a/benchmarks/single_node/glm5_fp8_mi355x_atom.sh b/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
index 5377e4041..960363bb6 100755
--- a/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
+++ b/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
@@ -48,7 +48,6 @@ python3 -m atom.entrypoints.openai_server \
     -tp $TP \
     --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
     --trust-remote-code \
-    --no-enable-prefix-caching \
     > $SERVER_LOG 2>&1 &
 
 SERVER_PID=$!
diff --git a/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh b/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
index d47b2fc74..8f94c3829 100755
--- a/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
+++ b/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
@@ -48,7 +48,6 @@ python3 -m atom.entrypoints.openai_server \
     -tp $TP \
     --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
     --trust-remote-code \
-    --no-enable-prefix-caching \
     > $SERVER_LOG 2>&1 &
 
 SERVER_PID=$!
diff --git a/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh b/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
index d47b2fc74..8f94c3829 100755
--- a/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
+++ b/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
@@ -48,7 +48,6 @@ python3 -m atom.entrypoints.openai_server \
     -tp $TP \
     --kv_cache_dtype fp8 $CALCULATED_MAX_MODEL_LEN $EP \
     --trust-remote-code \
-    --no-enable-prefix-caching \
     > $SERVER_LOG 2>&1 &
 
 SERVER_PID=$!

From 6e47afa206eedce851b7fc4ccf7c443755497fc3 Mon Sep 17 00:00:00 2001
From: seungrokj <seungrok.jung@amd.com>
Date: Tue, 31 Mar 2026 03:19:17 +0000
Subject: [PATCH 5/6] ATOM support for new models: Kimi-K2.5 FP4, GLM-5 FP8,
 and MiniMax-M2.5 FP8

Signed-off-by: seungrokj <seungrok.jung@amd.com>
---
 benchmarks/single_node/glm5_fp8_mi355x_atom.sh        | 2 +-
 benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh    | 2 +-
 benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/benchmarks/single_node/glm5_fp8_mi355x_atom.sh b/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
index 960363bb6..3ab42332f 100755
--- a/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
+++ b/benchmarks/single_node/glm5_fp8_mi355x_atom.sh
@@ -70,7 +70,7 @@ run_benchmark_serving \
 
 # After throughput, run evaluation only if RUN_EVAL is true
 if [ "${RUN_EVAL}" = "true" ]; then
-    run_eval --framework lm-eval --port "$PORT" --concurrent-requests $CONC
+    run_eval --framework lm-eval --port "$PORT" 
     append_lm_eval_summary
 fi
 
diff --git a/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh b/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
index 8f94c3829..ca84f8228 100755
--- a/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
+++ b/benchmarks/single_node/kimik2.5_fp4_mi355x_atom.sh
@@ -71,7 +71,7 @@ run_benchmark_serving \
 
 # After throughput, run evaluation only if RUN_EVAL is true
 if [ "${RUN_EVAL}" = "true" ]; then
-    run_eval --framework lm-eval --port "$PORT" --concurrent-requests $CONC
+    run_eval --framework lm-eval --port "$PORT" 
     append_lm_eval_summary
 fi
 
diff --git a/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh b/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
index 8f94c3829..ca84f8228 100755
--- a/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
+++ b/benchmarks/single_node/minimaxm2.5_fp8_mi355x_atom.sh
@@ -71,7 +71,7 @@ run_benchmark_serving \
 
 # After throughput, run evaluation only if RUN_EVAL is true
 if [ "${RUN_EVAL}" = "true" ]; then
-    run_eval --framework lm-eval --port "$PORT" --concurrent-requests $CONC
+    run_eval --framework lm-eval --port "$PORT" 
     append_lm_eval_summary
 fi
 

From 6ccbd9151c8f09fcc47b5b8445d8711961f94de7 Mon Sep 17 00:00:00 2001
From: seungrokj <seungrok.jung@amd.com>
Date: Tue, 31 Mar 2026 07:04:54 +0000
Subject: [PATCH 6/6] ATOM support for new models: Kimi-K2.5 FP4, GLM-5 FP8,
 and MiniMax-M2.5 FP8

Signed-off-by: seungrokj <seungrok.jung@amd.com>
---
 .github/configs/amd-master.yaml | 15 ---------------
 1 file changed, 15 deletions(-)
 mode change 100755 => 100644 .github/configs/amd-master.yaml

diff --git a/.github/configs/amd-master.yaml b/.github/configs/amd-master.yaml
old mode 100755
new mode 100644
index 280803acc..98c81185f
--- a/.github/configs/amd-master.yaml
+++ b/.github/configs/amd-master.yaml
@@ -252,10 +252,6 @@ glm5-fp8-mi355x-atom:
     osl: 1024
     search-space:
     - { tp: 8, conc-start: 4, conc-end: 128 }
-  - isl: 1024
-    osl: 8192
-    search-space:
-    - { tp: 8, conc-start: 4, conc-end: 128 }
   - isl: 8192
     osl: 1024
     search-space:
@@ -331,11 +327,6 @@ kimik2.5-fp4-mi355x-atom:
     search-space:
     - { tp: 8, conc-start: 4, conc-end: 128 }
     - { tp: 4, conc-start: 4, conc-end: 128 }
-  - isl: 1024
-    osl: 8192
-    search-space:
-    - { tp: 8, conc-start: 4, conc-end: 128 }
-    - { tp: 4, conc-start: 4, conc-end: 128 }
   - isl: 8192
     osl: 1024
     search-space:
@@ -379,12 +370,6 @@ minimaxm2.5-fp8-mi355x-atom:
     - { tp: 2, conc-start: 4, conc-end: 128 }
     - { tp: 4, conc-start: 4, conc-end: 128 }
     - { tp: 8, ep: 8, conc-start: 32, conc-end: 256 }
-  - isl: 1024
-    osl: 8192
-    search-space:
-    - { tp: 2, conc-start: 4, conc-end: 128 }
-    - { tp: 4, conc-start: 4, conc-end: 128 }
-    - { tp: 8, ep: 8, conc-start: 32, conc-end: 256 }
   - isl: 8192
     osl: 1024
     search-space: