gahow/aituner
0
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Profile FA3 KV-cache updates separately

Browse Source
This commit is contained in:
Gahow Wang
2026-07-16 21:37:01 +08:00
parent a9aed87518
commit 9bc38d8851
5 changed files with 100 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
runs/frontier-qwen30-vllm020-profile-v1/jobs_flashattn_full.toml
View File

@@ -1,55 +1,55 @@
version = 1
[[jobs]]
name = "qwen30-vllm020-flashattn-full-tp1-20260716-v1"
name = "qwen30-vllm020-flashattn-kv-full-tp1-20260716-v2"
gpus = 1
gpu_model = "H20"
hosts = ["dash0"]
command = "cd /home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-v1/runs/frontier-qwen30-vllm020-profile-v1 && timeout --signal=TERM --kill-after=30s 1320 bash run_flashattn_full.sh"
artifacts = ["artifacts/flashattn-full-tp1"]
artifacts = ["artifacts/flashattn-kv-full-v2-tp1"]
[jobs.env]
HOME = "/tmp/wjh"
XDG_CACHE_HOME = "/tmp/wjh/.cache"
VLLM_CACHE_ROOT = "/tmp/wjh/.cache/vllm"
TP = "1"
OUTPUT_ROOT = "/home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-fleet/artifacts/flashattn-full-tp1"
OUTPUT_ROOT = "/home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-fleet/artifacts/flashattn-kv-full-v2-tp1"
VENV_ROOT = "/tmp/wjh/venvs/vllm-0.20.0-cu129-profiler-v1"
VLLM_SOURCE = "/home/admin/cpfs/wjh/agentic-kv/third_party/vllm_v20_build"
MODEL = "/home/admin/cpfs/wjh/models/Qwen/Qwen3-30B-A3B"
[[jobs]]
name = "qwen30-vllm020-flashattn-full-tp2-20260716-v1"
name = "qwen30-vllm020-flashattn-kv-full-tp2-20260716-v2"
gpus = 1
gpu_model = "H20"
hosts = ["dash0"]
command = "cd /home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-v1/runs/frontier-qwen30-vllm020-profile-v1 && timeout --signal=TERM --kill-after=30s 1320 bash run_flashattn_full.sh"
artifacts = ["artifacts/flashattn-full-tp2"]
artifacts = ["artifacts/flashattn-kv-full-v2-tp2"]
[jobs.env]
HOME = "/tmp/wjh"
XDG_CACHE_HOME = "/tmp/wjh/.cache"
VLLM_CACHE_ROOT = "/tmp/wjh/.cache/vllm"
TP = "2"
OUTPUT_ROOT = "/home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-fleet/artifacts/flashattn-full-tp2"
OUTPUT_ROOT = "/home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-fleet/artifacts/flashattn-kv-full-v2-tp2"
VENV_ROOT = "/tmp/wjh/venvs/vllm-0.20.0-cu129-profiler-v1"
VLLM_SOURCE = "/home/admin/cpfs/wjh/agentic-kv/third_party/vllm_v20_build"
MODEL = "/home/admin/cpfs/wjh/models/Qwen/Qwen3-30B-A3B"
[[jobs]]
name = "qwen30-vllm020-flashattn-full-tp4-20260716-v1"
name = "qwen30-vllm020-flashattn-kv-full-tp4-20260716-v2"
gpus = 1
gpu_model = "H20"
hosts = ["dash0"]
command = "cd /home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-v1/runs/frontier-qwen30-vllm020-profile-v1 && timeout --signal=TERM --kill-after=30s 1320 bash run_flashattn_full.sh"
artifacts = ["artifacts/flashattn-full-tp4"]
artifacts = ["artifacts/flashattn-kv-full-v2-tp4"]
[jobs.env]
HOME = "/tmp/wjh"
XDG_CACHE_HOME = "/tmp/wjh/.cache"
VLLM_CACHE_ROOT = "/tmp/wjh/.cache/vllm"
TP = "4"
OUTPUT_ROOT = "/home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-fleet/artifacts/flashattn-full-tp4"
OUTPUT_ROOT = "/home/admin/cpfs/wjh/aituner/aituner-qwen30-vllm020-profile-fleet/artifacts/flashattn-kv-full-v2-tp4"
VENV_ROOT = "/tmp/wjh/venvs/vllm-0.20.0-cu129-profiler-v1"
VLLM_SOURCE = "/home/admin/cpfs/wjh/agentic-kv/third_party/vllm_v20_build"
MODEL = "/home/admin/cpfs/wjh/models/Qwen/Qwen3-30B-A3B"