michael/AI-Profile-Router
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Adopt tested MTP2 for Large; retain Medium MTP3 after warmup failure

Browse Source
This commit is contained in:
Mikei386
2026-09-20 22:59:50 +02:00
parent 9e836cf5fd
commit b352e29c89
29 changed files with 2340 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
experiments/profile-mtp2-20260920/large-mtp3/config.json
+11
View File
@@ -0,0 +1,11 @@
{
"label": "large-mtp3",
"profile": "large",
"ubatch": 256,
"split": "86,14",
"mtp": 3,
"prompts": [
4096
],
"minimum_headroom_mib": 448
}
experiments/profile-mtp2-20260920/large-mtp3/gpu.json
+278
View File
@@ -0,0 +1,278 @@
[
{
"time": 1789937847.9726973,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "6964",
"total": "12288",
"temp": "50",
"util": "100",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "11272",
"total": "16303",
"temp": "48",
"util": "0",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937850.0014837,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "8350",
"total": "12288",
"temp": "50",
"util": "4",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15682",
"total": "16303",
"temp": "49",
"util": "0",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937852.0301683,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10722",
"total": "12288",
"temp": "50",
"util": "0",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15684",
"total": "16303",
"temp": "48",
"util": "0",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937854.061566,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10726",
"total": "12288",
"temp": "56",
"util": "59",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15700",
"total": "16303",
"temp": "56",
"util": "47",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937856.0915275,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10726",
"total": "12288",
"temp": "57",
"util": "53",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15700",
"total": "16303",
"temp": "58",
"util": "52",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937858.1227798,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10726",
"total": "12288",
"temp": "56",
"util": "47",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15700",
"total": "16303",
"temp": "57",
"util": "52",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937860.1527982,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10726",
"total": "12288",
"temp": "58",
"util": "56",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15700",
"total": "16303",
"temp": "58",
"util": "44",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937862.1820376,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10728",
"total": "12288",
"temp": "56",
"util": "50",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15702",
"total": "16303",
"temp": "69",
"util": "82",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937864.2163124,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10728",
"total": "12288",
"temp": "58",
"util": "54",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15702",
"total": "16303",
"temp": "60",
"util": "39",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937866.2504282,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10728",
"total": "12288",
"temp": "59",
"util": "42",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15702",
"total": "16303",
"temp": "60",
"util": "46",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937868.2800074,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10728",
"total": "12288",
"temp": "59",
"util": "48",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15702",
"total": "16303",
"temp": "60",
"util": "44",
"pcie_gen": "4",
"pcie_width": "16"
}
]
},
{
"time": 1789937870.3127642,
"gpus": [
{
"name": "NVIDIA GeForce RTX 3060",
"used": "10728",
"total": "12288",
"temp": "59",
"util": "59",
"pcie_gen": "3",
"pcie_width": "4"
},
{
"name": "NVIDIA GeForce RTX 5080",
"used": "15702",
"total": "16303",
"temp": "60",
"util": "40",
"pcie_gen": "4",
"pcie_width": "16"
}
]
}
]
experiments/profile-mtp2-20260920/large-mtp3/loaded.json
+133
View File
File diff suppressed because one or more lines are too long
experiments/profile-mtp2-20260920/large-mtp3/result.json
+302
View File
File diff suppressed because one or more lines are too long
experiments/profile-mtp2-20260920/large-mtp3/server-args.json
+73
View File
@@ -0,0 +1,73 @@
[
"--model",
"/models/qwen3.8-27b-iq4-xs-pure/qwen3.8-27b-IQ4_XS-pure.gguf",
"--mmproj",
"/models/qwen/mmproj-BF16.gguf",
"--mmproj-offload",
"--mmproj-device",
"CUDA1",
"--alias",
"qwen-large",
"--ctx-size",
"192000",
"--flash-attn",
"on",
"--cache-type-k",
"q4_0",
"--cache-type-v",
"q4_0",
"--cache-prompt",
"--cache-ram",
"32768",
"--threads",
"6",
"--threads-batch",
"6",
"--batch-size",
"2048",
"--ubatch-size",
"256",
"--parallel",
"1",
"--kv-unified",
"--jinja",
"--reasoning",
"auto",
"--reasoning-preserve",
"--host",
"127.0.0.1",
"--port",
"5005",
"--metrics",
"--fit",
"off",
"--n-gpu-layers",
"all",
"--load-mode",
"none",
"--no-ui",
"--temperature",
"0.2",
"--top-p",
"0.8",
"--top-k",
"20",
"--device",
"CUDA0,CUDA1",
"--main-gpu",
"0",
"--split-mode",
"layer",
"--tensor-split",
"86,14",
"--spec-type",
"draft-mtp",
"--spec-draft-n-max",
"3",
"--spec-draft-type-k",
"f16",
"--spec-draft-type-v",
"f16",
"--verbosity",
"3"
]
experiments/profile-mtp2-20260920/large-mtp3/server.log.gz
BIN
View File
Binary file not shown.