feat: H20 support

LyleLuo · LyleLuo · commit 0f5e88eb8bae · 2025-10-16T11:18:13.000+08:00
Signed-off-by: Weile Luo &lt;weilel@nvidia.com&gt;
diff --git a/README.md b/README.md
@@ -298,6 +298,8 @@ To go through the process, refer to the [guidance](collector/README.md) under th
 |--------|-------------------|--------|
 | h100_sxm | TRTLLM(0.20.0, 1.0.0rc3) | ✅ |
 | h200_sxm | TRTLLM(0.20.0, 1.0.0rc3) | ✅ |
+| h20_3e | TRTLLM(1.0.0) | ✅ |
+| A100 | TRTLLM(1.0.0) | ✅ |
 | b200_sxm | TRTLLM(1.0.0rc6) | ✅ |
 | gb200_sxm | TRTLLM(1.0.0rc6) | ✅ |
 
diff --git a/src/aiconfigurator/systems/data/h20_3e/nccl/2.27.3/nccl_perf.txt b/src/aiconfigurator/systems/data/h20_3e/nccl/2.27.3/nccl_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:babed36b408253a4ba9ea5031e9314f5cea2f7dd0b849f475d758bf36a45c9fa
+size 35171
diff --git a/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/context_attention_perf.txt b/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/context_attention_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f0fe8ca8f79fd0b68d567b9cd1576599bb9c53eb5239a9161ca7c6efc56444a
+size 6374897
diff --git a/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/context_mla_perf.txt b/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/context_mla_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:abdc05a5446f6f6aca47ab284237e1556cc0543469a5d18e74cb67b2405ec73e
+size 459376
diff --git a/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/custom_allreduce_perf.txt b/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/custom_allreduce_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ffc0b4cd20fd98e79b9d455f3f4a5f7322fec8b8f68ff0ed62a9040afb6eefe6
+size 5787
diff --git a/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/gemm_perf.txt b/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/gemm_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29ae0a32a2e37223971db4dc7bb93212787e520117ceea4b1751b0ee58e5ff48
+size 9486670
diff --git a/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/generation_attention_perf.txt b/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/generation_attention_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12cacb85ed0375d42a95af8a7ee5679e7071cc27dcabd8df81bba676b89f6e8c
+size 4543889
diff --git a/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/generation_mla_perf.txt b/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/generation_mla_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1824c4fa7a9dfdefa0d63d714081a1ad49bff81b65b064f739697b9bc518ee36
+size 993378
diff --git a/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/mla_bmm_perf.txt b/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/mla_bmm_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c843b2e7088a9db80f4a054215456861056731fd02bb6193004ef00d3e6ec660
+size 136602
diff --git a/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/moe_perf.txt b/src/aiconfigurator/systems/data/h20_3e/trtllm/1.0.0/moe_perf.txt
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:358c2b6ed8e4e1a47be248a582d8a7e61dd16c8f8a3c4fb76e34714cfdb98ff1
+size 667000
diff --git a/src/aiconfigurator/systems/h20_3e.yaml b/src/aiconfigurator/systems/h20_3e.yaml
@@ -0,0 +1,30 @@
+# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+data_dir: data/h20_3e # relative to systems_dir
+gpu:
+  mem_bw: 4917000000000 # 4917GB/s
+  mem_bw_empirical_scaling_factor: 0.8 # some nonofficial correction based on observations, you should try to modify based on your own observations
+  mem_empirical_constant_latency: 0.000003 # 3us some nonofficial correction based on observations, you should try to modify based on your own observations
+  mem_capacity: 151397597184 # 141GiB
+  float16_tc_flops: 148000000000000 # 148TFLOPS
+  int8_tc_flops: 296000000000000 # 296TFLOPS
+  fp8_tc_flops: 296000000000000 # 296TFLOPS
+  power: 500  # Watt
+  sm_version: 90
+
+node:
+  num_gpus_per_node: 8
+  inter_node_bw: 25000000000  # Byte/s per GPU, single direction, 1:1 CX7 per node
+  intra_node_bw: 450000000000  # Byte/s per gpu, single direction
+  pcie_bw: 64000000000  # Byte/s, single direction, pcie 5.0
+  p2p_latency: 0.00001  # 10us some nonofficial correction based on observations, you should try to modify based on your own observations
+
+misc:
+  nccl_mem: # some nonofficial correction based on observations, you should try to modify based on your own observations
+    1: 0
+    2: 358612992 # 342MB
+    4: 411041792 # 392MB
+    8: 411041792 # 392MB
+  other_mem: 3758096384 # increase from 551MB to 3.5GB for safer deployment, this will cover part of the inaccurate mem calc.
+  nccl_version: '2.27.3'

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:babed36b408253a4ba9ea5031e9314f5cea2f7dd0b849f475d758bf36a45c9fa`
	`3`	`+size 35171`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:4f0fe8ca8f79fd0b68d567b9cd1576599bb9c53eb5239a9161ca7c6efc56444a`
	`3`	`+size 6374897`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:abdc05a5446f6f6aca47ab284237e1556cc0543469a5d18e74cb67b2405ec73e`
	`3`	`+size 459376`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:ffc0b4cd20fd98e79b9d455f3f4a5f7322fec8b8f68ff0ed62a9040afb6eefe6`
	`3`	`+size 5787`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:29ae0a32a2e37223971db4dc7bb93212787e520117ceea4b1751b0ee58e5ff48`
	`3`	`+size 9486670`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:12cacb85ed0375d42a95af8a7ee5679e7071cc27dcabd8df81bba676b89f6e8c`
	`3`	`+size 4543889`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:1824c4fa7a9dfdefa0d63d714081a1ad49bff81b65b064f739697b9bc518ee36`
	`3`	`+size 993378`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:c843b2e7088a9db80f4a054215456861056731fd02bb6193004ef00d3e6ec660`
	`3`	`+size 136602`
Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+version https://git-lfs.github.com/spec/v1`
	`2`	`+oid sha256:358c2b6ed8e4e1a47be248a582d8a7e61dd16c8f8a3c4fb76e34714cfdb98ff1`
	`3`	`+size 667000`