anodyine/vllm-omni
1
0 Fork 0
Code Issues 4 Pull Requests 2 Actions Packages Projects Releases Wiki Activity

[CI] Gate qwen3-omni no-async-chunk perf on mean_audio_rtf (#6743)

Browse Source 
Signed-off-by: Inesh Reddy <ineshreddy249@gmail.com>
Co-authored-by: amy-why-3459 <wuhaiyan17@huawei.com>
This commit is contained in:
Inesh Reddy Chappidiandamy-why-3459 authored and GitHub committed 2026-08-28 20:28:58 -07:00
1 parent 46cfa8435b
commit bbb3ae01a1
2 files changed
-25
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

tests/dfx/perf/tests/test_qwen3_omni_async_chunk.json
-9
View File
@@ -46,9 +46,6 @@
"percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"mean_e2el_ms": [
3646.797
],
"mean_audio_rtf": [
0.1098
]
@@ -82,9 +79,6 @@
"percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"mean_e2el_ms": [
4059.0878
],
"mean_audio_rtf": [
0.1335
]
@@ -120,9 +114,6 @@
"percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"mean_e2el_ms": [
5639.5769
],
"mean_audio_rtf": [
0.1798
]
tests/dfx/perf/tests/test_qwen3_omni_no_async_chunk.json
-16
View File
@@ -45,13 +45,6 @@
"percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"mean_e2el_ms": [
15793.1209,
30658.6632,
37463.4943,
47990.5839,
64980.2199
],
"mean_audio_rtf": [
0.169,
0.2041,
@@ -87,9 +80,6 @@
"percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"mean_e2el_ms": [
4556.4347
],
"mean_audio_rtf": [
0.138
]
@@ -123,9 +113,6 @@
"percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"mean_e2el_ms": [
4768.0972
],
"mean_audio_rtf": [
0.1609
]
@@ -161,9 +148,6 @@
"percentile-metrics": "ttft,tpot,itl,e2el,audio_rtf,audio_ttfp,audio_duration",
"baseline": {
"H100": {
"mean_e2el_ms": [
6395.3517
],
"mean_audio_rtf": [
0.2048
]