From fab09198e4685b8d1b2b5d885b8f47e101ff2abf Mon Sep 17 00:00:00 2001
From: jthomson04 <jothomson@nvidia.com>
Date: Mon, 2 Feb 2026 11:48:33 -0800
Subject: [PATCH 1/5] b200 fp8

Signed-off-by: jthomson04 <jothomson@nvidia.com>
---
 .github/configs/nvidia-master.yaml | 435 +++++++++++++++++++++++++++++
 1 file changed, 435 insertions(+)

diff --git a/.github/configs/nvidia-master.yaml b/.github/configs/nvidia-master.yaml
index e99282490..a35f25c15 100644
--- a/.github/configs/nvidia-master.yaml
+++ b/.github/configs/nvidia-master.yaml
@@ -383,6 +383,441 @@ dsr1-fp4-b200-dynamo-trt:
         ep: 8
         dp-attn: true
 
+dsr1-fp8-b200-dynamo-trt:
+  image: nvcr.io/nvidia/ai-dynamo/tensorrtllm-runtime:0.8.1.post2
+  model: deepseek-ai/DeepSeek-R1-0528
+  model-prefix: dsr1
+  runner: b200-multinode-slurm
+  precision: fp8
+  framework: dynamo-trt
+  multinode: true
+  disagg: true
+  seq-len-configs:
+  - isl: 1024
+    osl: 1024
+    search-space:
+    # MTP configurations - Low latency (TP attention)
+    - spec-decoding: "mtp"
+      conc-list: [8]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen8_tp8_batch1_eplb0_mtp3_8.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen8_tp8_batch1_eplb0_mtp3_8.yaml"
+      decode:
+        num-worker: 8
+        tp: 8
+        dp-attn: false
+    - spec-decoding: "mtp"
+      conc-list: [32]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen8_tp8_batch4_eplb0_mtp3_32.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen8_tp8_batch4_eplb0_mtp3_32.yaml"
+      decode:
+        num-worker: 8
+        tp: 8
+        dp-attn: false
+    - spec-decoding: "mtp"
+      conc-list: [64]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen8_tp8_batch8_eplb0_mtp3_64.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen8_tp8_batch8_eplb0_mtp3_64.yaml"
+      decode:
+        num-worker: 8
+        tp: 8
+        dp-attn: false
+    - spec-decoding: "mtp"
+      conc-list: [256]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen8_tp8_batch32_eplb0_mtp3_256.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen8_tp8_batch32_eplb0_mtp3_256.yaml"
+      decode:
+        num-worker: 8
+        tp: 8
+        dp-attn: false
+    # MTP configurations - High throughput (DP attention)
+    - spec-decoding: "mtp"
+      conc-list: [896]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen7_dep8_batch128_eplb0_mtp3_896.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen7_dep8_batch128_eplb0_mtp3_896.yaml"
+      decode:
+        num-worker: 7
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - spec-decoding: "mtp"
+      conc-list: [1024]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen4_dep8_batch256_eplb0_mtp3_1024.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen4_dep8_batch256_eplb0_mtp3_1024.yaml"
+      decode:
+        num-worker: 4
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - spec-decoding: "mtp"
+      conc-list: [1184]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen3_dep8_batch384_eplb0_mtp3_1184.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen3_dep8_batch384_eplb0_mtp3_1184.yaml"
+      decode:
+        num-worker: 3
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - spec-decoding: "mtp"
+      conc-list: [1600]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen2_dep8_batch768_eplb0_mtp2_1600.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/mtp/ctx1_gen2_dep8_batch768_eplb0_mtp2_1600.yaml"
+      decode:
+        num-worker: 2
+        tp: 8
+        ep: 8
+        dp-attn: true
+
+    # Non-MTP (STP) configurations - Low latency (TP attention)
+    - conc-list: [4]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen3_tp8_batch1024_eplb0_mtp0_4.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen3_tp8_batch1024_eplb0_mtp0_4.yaml"
+      decode:
+        num-worker: 3
+        tp: 8
+        dp-attn: false
+    - conc-list: [32]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen3_tp8_batch1024_eplb0_mtp0_32.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen3_tp8_batch1024_eplb0_mtp0_32.yaml"
+      decode:
+        num-worker: 3
+        tp: 8
+        dp-attn: false
+    - conc-list: [128]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen3_tp8_batch1024_eplb0_mtp0_128.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen3_tp8_batch1024_eplb0_mtp0_128.yaml"
+      decode:
+        num-worker: 3
+        tp: 8
+        dp-attn: false
+    # Non-MTP (STP) configurations - High throughput (DP attention)
+    - conc-list: [1920]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen5_dep8_batch48_eplb0_mtp0_1920.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen5_dep8_batch48_eplb0_mtp0_1920.yaml"
+      decode:
+        num-worker: 5
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - conc-list: [4096]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen1_dep8_batch512_eplb0_mtp0_4096.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/stp/ctx1_gen1_dep8_batch512_eplb0_mtp0_4096.yaml"
+      decode:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - conc-list: [5152]
+      prefill:
+        num-worker: 2
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/1k1k/stp/ctx2_gen5_dep8_batch128_eplb0_mtp0_5152.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/1k1k/stp/ctx2_gen5_dep8_batch128_eplb0_mtp0_5152.yaml"
+      decode:
+        num-worker: 5
+        tp: 8
+        ep: 8
+        dp-attn: true
+
+  - isl: 8192
+    osl: 1024
+    search-space:
+    # MTP configurations - Low latency (TP attention)
+    - spec-decoding: "mtp"
+      conc-list: [8]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/mtp/ctx1_gen6_tp8_batch8_eplb0_mtp3_8.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/mtp/ctx1_gen6_tp8_batch8_eplb0_mtp3_8.yaml"
+      decode:
+        num-worker: 6
+        tp: 8
+        dp-attn: false
+    - spec-decoding: "mtp"
+      conc-list: [8]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/mtp/ctx1_gen2_tp8_batch32_eplb0_mtp3_8.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/mtp/ctx1_gen2_tp8_batch32_eplb0_mtp3_8.yaml"
+      decode:
+        num-worker: 2
+        tp: 8
+        dp-attn: false
+    - spec-decoding: "mtp"
+      conc-list: [48]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/mtp/ctx1_gen6_tp8_batch8_eplb0_mtp3_48.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/mtp/ctx1_gen6_tp8_batch8_eplb0_mtp3_48.yaml"
+      decode:
+        num-worker: 6
+        tp: 8
+        dp-attn: false
+    - spec-decoding: "mtp"
+      conc-list: [64]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/mtp/ctx1_gen4_tp8_batch16_eplb0_mtp3_64.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/mtp/ctx1_gen4_tp8_batch16_eplb0_mtp3_64.yaml"
+      decode:
+        num-worker: 4
+        tp: 8
+        dp-attn: false
+    # MTP configurations - High throughput (DP attention)
+    - spec-decoding: "mtp"
+      conc-list: [224]
+      prefill:
+        num-worker: 2
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/mtp/ctx2_gen3_dep8_batch8_eplb0_mtp3_224.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/mtp/ctx2_gen3_dep8_batch8_eplb0_mtp3_224.yaml"
+      decode:
+        num-worker: 3
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - spec-decoding: "mtp"
+      conc-list: [288]
+      prefill:
+        num-worker: 2
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/mtp/ctx2_gen1_dep8_batch32_eplb0_mtp3_288.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/mtp/ctx2_gen1_dep8_batch32_eplb0_mtp3_288.yaml"
+      decode:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - spec-decoding: "mtp"
+      conc-list: [1088]
+      prefill:
+        num-worker: 4
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/mtp/ctx4_gen1_dep8_batch128_eplb0_mtp2_1088.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/mtp/ctx4_gen1_dep8_batch128_eplb0_mtp2_1088.yaml"
+      decode:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+
+    # Non-MTP (STP) configurations - Low latency (TP attention)
+    - conc-list: [1]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen1_tp8_batch1_eplb0_mtp0_1.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen1_tp8_batch1_eplb0_mtp0_1.yaml"
+      decode:
+        num-worker: 1
+        tp: 8
+        dp-attn: false
+    - conc-list: [32]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen4_tp8_batch32_eplb0_mtp0_32.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen4_tp8_batch32_eplb0_mtp0_32.yaml"
+      decode:
+        num-worker: 4
+        tp: 8
+        dp-attn: false
+    - conc-list: [128]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen4_tp8_batch32_eplb0_mtp0_128.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen4_tp8_batch32_eplb0_mtp0_128.yaml"
+      decode:
+        num-worker: 4
+        tp: 8
+        dp-attn: false
+    - conc-list: [96]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen6_tp8_batch16_eplb0_mtp0_96.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen6_tp8_batch16_eplb0_mtp0_96.yaml"
+      decode:
+        num-worker: 6
+        tp: 8
+        dp-attn: false
+    # Non-MTP (STP) configurations - High throughput (DP attention)
+    - conc-list: [128]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen1_dep8_batch128_eplb0_mtp0_128.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen1_dep8_batch128_eplb0_mtp0_128.yaml"
+      decode:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - conc-list: [128]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen2_dep8_batch64_eplb0_mtp0_128.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen2_dep8_batch64_eplb0_mtp0_128.yaml"
+      decode:
+        num-worker: 2
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - conc-list: [256]
+      prefill:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen1_dep8_batch256_eplb0_mtp0_256.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/stp/ctx1_gen1_dep8_batch256_eplb0_mtp0_256.yaml"
+      decode:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+    - conc-list: [640]
+      prefill:
+        num-worker: 2
+        tp: 8
+        ep: 8
+        dp-attn: true
+        additional-settings:
+        # https://github.com/ishandhanani/srt-slurm/blob/sa-submission-q1-2026/recipes/trtllm/b200-fp8/8k1k/stp/ctx2_gen1_dep8_batch640_eplb0_mtp0_640.yaml
+        - "CONFIG_FILE=recipes/trtllm/b200-fp8/8k1k/stp/ctx2_gen1_dep8_batch640_eplb0_mtp0_640.yaml"
+      decode:
+        num-worker: 1
+        tp: 8
+        ep: 8
+        dp-attn: true
+
+
 dsr1-fp4-b300-dynamo-trt:
   image: nvcr.io/nvidia/ai-dynamo/tensorrtllm-runtime:0.8.1.post1
   model: deepseek-r1-fp4

From d63b5a4fbf2d04d5b9d529408268f84537fee423 Mon Sep 17 00:00:00 2001
From: jthomson04 <jothomson@nvidia.com>
Date: Thu, 5 Feb 2026 17:04:57 -0800
Subject: [PATCH 2/5] updates

Signed-off-by: jthomson04 <jothomson@nvidia.com>
---
 perf-changelog.yaml               | 6 ++++++
 runners/launch_b200-dgxc-slurm.sh | 6 ++++--
 2 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 20fd381fc..00de1286e 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -400,3 +400,9 @@
     - "Update GPT-OSS FP4 B200 TRT pareto configurations and new container image"
     - "Extend maximum concurrency to 256 across all sequence lengths"
   pr-link: https://github.com/InferenceMAX/InferenceMAX/pull/639
+
+- config-keys:
+    - dsr1-fp8-b200-dynamo-trt
+  description:
+    - "Introduce new DSR1 FP8 B200 Dynamo TRT configurations for 8k1k and 1k1k"
+  pr-link: https://github.com/InferenceMAX/InferenceMAX/pull/647
\ No newline at end of file
diff --git a/runners/launch_b200-dgxc-slurm.sh b/runners/launch_b200-dgxc-slurm.sh
index 7a6d31288..b77ca390c 100644
--- a/runners/launch_b200-dgxc-slurm.sh
+++ b/runners/launch_b200-dgxc-slurm.sh
@@ -34,8 +34,10 @@ export SLURM_ACCOUNT="root"
 if [[ $MODEL_PREFIX == "dsr1" ]]; then
     if [[ $PRECISION == "fp4" ]]; then
         export MODEL_PATH="/lustre/fsw/models/dsr1-0528-nvfp4-v2"
+        export SRT_SLURM_MODEL_PREFIX="dsr1"
     elif [[ $PRECISION == "fp8" ]]; then
-        export MODEL_PATH="/lustre/fsw/models/dsr1-0528-fp8"
+        export MODEL_PATH="/raid/models/dsr1-0528-fp8"
+        export SRT_SLURM_MODEL_PREFIX="dsr1-fp8"
     else
         echo "Unsupported precision: $PRECISION. Supported precisions are: fp4, fp8"
         exit 1
@@ -72,7 +74,7 @@ network_interface: ""
 srtctl_root: "${GITHUB_WORKSPACE}/${SRT_REPO_DIR}"
 # Model path aliases
 model_paths:
-  "${MODEL_PREFIX}": "${MODEL_PATH}"
+  "${SRT_SLURM_MODEL_PREFIX}": "${MODEL_PATH}"
 # Container aliases
 containers:
   dynamo-trtllm: "${SQUASH_FILE}"

From fc9912138b4e096c96cf1fef76c408f2fd93e867 Mon Sep 17 00:00:00 2001
From: jthomson04 <jothomson@nvidia.com>
Date: Thu, 5 Feb 2026 17:07:53 -0800
Subject: [PATCH 3/5] newline

Signed-off-by: jthomson04 <jothomson@nvidia.com>
---
 perf-changelog.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 00de1286e..13d7b2f25 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -405,4 +405,4 @@
     - dsr1-fp8-b200-dynamo-trt
   description:
     - "Introduce new DSR1 FP8 B200 Dynamo TRT configurations for 8k1k and 1k1k"
-  pr-link: https://github.com/InferenceMAX/InferenceMAX/pull/647
\ No newline at end of file
+  pr-link: https://github.com/InferenceMAX/InferenceMAX/pull/647

From 3ae74afcfa71da11484db7b81112a2ba10b0c652 Mon Sep 17 00:00:00 2001
From: jthomson04 <jothomson@nvidia.com>
Date: Thu, 5 Feb 2026 17:13:09 -0800
Subject: [PATCH 4/5] Add ep config field

Signed-off-by: jthomson04 <jothomson@nvidia.com>
---
 .github/configs/nvidia-master.yaml | 15 +++++++++++++++
 1 file changed, 15 insertions(+)

diff --git a/.github/configs/nvidia-master.yaml b/.github/configs/nvidia-master.yaml
index ce8ecae79..63925846f 100644
--- a/.github/configs/nvidia-master.yaml
+++ b/.github/configs/nvidia-master.yaml
@@ -410,6 +410,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 8
         tp: 8
+        ep: 1
         dp-attn: false
     - spec-decoding: "mtp"
       conc-list: [32]
@@ -424,6 +425,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 8
         tp: 8
+        ep: 1
         dp-attn: false
     - spec-decoding: "mtp"
       conc-list: [64]
@@ -438,6 +440,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 8
         tp: 8
+        ep: 1
         dp-attn: false
     - spec-decoding: "mtp"
       conc-list: [256]
@@ -452,6 +455,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 8
         tp: 8
+        ep: 1
         dp-attn: false
     # MTP configurations - High throughput (DP attention)
     - spec-decoding: "mtp"
@@ -528,6 +532,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 3
         tp: 8
+        ep: 1
         dp-attn: false
     - conc-list: [32]
       prefill:
@@ -541,6 +546,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 3
         tp: 8
+        ep: 1
         dp-attn: false
     - conc-list: [128]
       prefill:
@@ -554,6 +560,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 3
         tp: 8
+        ep: 1
         dp-attn: false
     # Non-MTP (STP) configurations - High throughput (DP attention)
     - conc-list: [1920]
@@ -616,6 +623,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 6
         tp: 8
+        ep: 1
         dp-attn: false
     - spec-decoding: "mtp"
       conc-list: [8]
@@ -630,6 +638,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 2
         tp: 8
+        ep: 1
         dp-attn: false
     - spec-decoding: "mtp"
       conc-list: [48]
@@ -644,6 +653,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 6
         tp: 8
+        ep: 1
         dp-attn: false
     - spec-decoding: "mtp"
       conc-list: [64]
@@ -658,6 +668,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 4
         tp: 8
+        ep: 1
         dp-attn: false
     # MTP configurations - High throughput (DP attention)
     - spec-decoding: "mtp"
@@ -719,6 +730,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 1
         tp: 8
+        ep: 1
         dp-attn: false
     - conc-list: [32]
       prefill:
@@ -732,6 +744,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 4
         tp: 8
+        ep: 1
         dp-attn: false
     - conc-list: [128]
       prefill:
@@ -745,6 +758,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 4
         tp: 8
+        ep: 1
         dp-attn: false
     - conc-list: [96]
       prefill:
@@ -758,6 +772,7 @@ dsr1-fp8-b200-dynamo-trt:
       decode:
         num-worker: 6
         tp: 8
+        ep: 1
         dp-attn: false
     # Non-MTP (STP) configurations - High throughput (DP attention)
     - conc-list: [128]

From 8afce6b8ac4df8608cf1eb41f86cc91007df5a82 Mon Sep 17 00:00:00 2001
From: jthomson04 <jothomson@nvidia.com>
Date: Tue, 10 Feb 2026 08:52:42 -0800
Subject: [PATCH 5/5] fix perf changelog

Signed-off-by: jthomson04 <jothomson@nvidia.com>
---
 perf-changelog.yaml | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/perf-changelog.yaml b/perf-changelog.yaml
index 129bf5a88..99de5649d 100644
--- a/perf-changelog.yaml
+++ b/perf-changelog.yaml
@@ -496,4 +496,4 @@
     - dsr1-fp8-b200-dynamo-trt
   description:
     - "Introduce new DSR1 FP8 B200 Dynamo TRT configurations for 8k1k and 1k1k"
-  pr-link: https://github.com/InferenceMAX/InferenceMAX/pull/647
+  pr-link: https://github.com/InferenceMAX/InferenceMAX/pull/616