Skip to content

Commit cab198d

Browse files
authored
[https://nvbugs/6108994][fix] add kv_transfer_timeout_ms to avoid timeout (#15152)
Signed-off-by: Bo Deng <deemod@nvidia.com>
1 parent 0d44f33 commit cab198d

7 files changed

Lines changed: 12 additions & 2 deletions

tests/integration/test_lists/waives.txt

Lines changed: 0 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -322,10 +322,8 @@ perf/test_perf_sanity.py::test_e2e[aggr_upload-super_ad_blackwell-super_ad_ws1_1
322322
perf/test_perf_sanity.py::test_e2e[disagg_upload-e2e-gb200_deepseek-r1-fp4_128k8k_con128_ctx1_pp8_gen1_dep16_eplb0_mtp1_ccb-NIXL] SKIP (https://nvbugs/6215844)
323323
perf/test_perf_sanity.py::test_e2e[disagg_upload-e2e-gb200_deepseek-v32-fp4_1k1k_con2048_ctx1_dep4_gen1_dep4_eplb0_mtp1_ccb-NIXL] SKIP (https://nvbugs/6280649)
324324
perf/test_perf_sanity.py::test_e2e[disagg_upload-e2e-gb200_kimi-k25-thinking-fp4_8k1k_con4096_ctx1_dep4_gen1_dep16_eplb0_mtp0_ccb-NIXL] SKIP (https://nvbugs/6179661)
325-
perf/test_perf_sanity.py::test_e2e[disagg_upload-e2e-gb300_deepseek-r1-fp4_128k8k_con256_ctx1_pp4_gen1_dep8_eplb0_mtp1_ccb-NIXL] SKIP (https://nvbugs/6280649)
326325
perf/test_perf_sanity.py::test_e2e[disagg_upload-e2e-gb300_deepseek-v32-fp4_1k1k_con2048_ctx1_dep4_gen1_dep4_eplb0_mtp1_ccb-NIXL] SKIP (https://nvbugs/6280649)
327326
perf/test_perf_sanity.py::test_e2e[disagg_upload-e2e-gb300_kimi-k25-thinking-fp4_8k1k_con4096_ctx1_dep4_gen1_dep16_eplb0_mtp0_ccb-NIXL] SKIP (https://nvbugs/6280649)
328-
perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-r1-fp4_128k8k_con128_ctx1_pp8_gen1_dep16_eplb0_mtp1_ccb-NIXL] SKIP (https://nvbugs/6221024)
329327
perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-v32-fp4_1k1k_con1024_ctx1_dep4_gen1_dep32_eplb256_mtp3_ccb-NIXL] SKIP (https://nvbugs/6280649)
330328
perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-v32-fp4_1k1k_con1_ctx1_dep4_gen1_tep8_eplb0_mtp3_ccb-NIXL] SKIP (https://nvbugs/6280721)
331329
perf/test_perf_sanity.py::test_e2e[disagg_upload-gen_only-gb200_deepseek-v32-fp4_1k1k_con2048_ctx1_dep4_gen1_dep4_eplb0_mtp1_ccb-NIXL] SKIP (https://nvbugs/6280649)

tests/scripts/perf-sanity/disaggregated/gb200_deepseek-r1-fp4_128k8k_con128_ctx1_pp8_gen1_dep16_eplb0_mtp1_ccb-NIXL.yaml

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -63,6 +63,7 @@ worker_config:
6363
cache_transceiver_config:
6464
max_tokens_in_buffer: 131104
6565
backend: NIXL
66+
kv_transfer_timeout_ms: 600000
6667
disable_overlap_scheduler: true
6768
speculative_config: &id001
6869
decoding_type: MTP
@@ -89,5 +90,6 @@ worker_config:
8990
cache_transceiver_config:
9091
max_tokens_in_buffer: 131104
9192
backend: NIXL
93+
kv_transfer_timeout_ms: 600000
9294
disable_overlap_scheduler: true
9395
speculative_config: *id001

tests/scripts/perf-sanity/disaggregated/gb300_deepseek-v32-fp4_32k4k_con256_ctx1_dep8_gen1_dep8_eplb0_mtp0_ccb-NIXL.yaml

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -70,6 +70,7 @@ worker_config:
7070
cache_transceiver_config:
7171
max_tokens_in_buffer: 120000
7272
backend: NIXL
73+
kv_transfer_timeout_ms: 600000
7374
num_postprocess_workers: 8
7475
stream_interval: 10
7576
ctx:
@@ -97,3 +98,4 @@ worker_config:
9798
cache_transceiver_config:
9899
max_tokens_in_buffer: 120000
99100
backend: NIXL
101+
kv_transfer_timeout_ms: 600000

tests/scripts/perf/disaggregated/gb200_deepseek-r1-fp4_128k8k_con128_ctx1_pp8_gen1_dep16_eplb0_mtp1_ccb-NIXL.yaml

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -74,6 +74,7 @@ worker_config:
7474
cache_transceiver_config:
7575
max_tokens_in_buffer: 131104
7676
backend: NIXL
77+
kv_transfer_timeout_ms: 600000
7778
disable_overlap_scheduler: true
7879
speculative_config: &id001
7980
decoding_type: MTP
@@ -100,5 +101,6 @@ worker_config:
100101
cache_transceiver_config:
101102
max_tokens_in_buffer: 131104
102103
backend: NIXL
104+
kv_transfer_timeout_ms: 600000
103105
disable_overlap_scheduler: true
104106
speculative_config: *id001

tests/scripts/perf/disaggregated/gb300_deepseek-r1-fp4_128k8k_con256_ctx1_pp4_gen1_dep8_eplb0_mtp1_ccb-NIXL.yaml

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -75,6 +75,7 @@ worker_config:
7575
cache_transceiver_config:
7676
max_tokens_in_buffer: 131104
7777
backend: NIXL
78+
kv_transfer_timeout_ms: 600000
7879
disable_overlap_scheduler: true
7980
speculative_config: &id001
8081
decoding_type: MTP
@@ -101,5 +102,6 @@ worker_config:
101102
cache_transceiver_config:
102103
max_tokens_in_buffer: 131104
103104
backend: NIXL
105+
kv_transfer_timeout_ms: 600000
104106
disable_overlap_scheduler: true
105107
speculative_config: *id001

tests/scripts/perf/disaggregated/gb300_deepseek-r1-fp4_128k8k_con256_ctx1_pp4_gen1_dep8_eplb0_mtp1_ccb-UCX.yaml

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -75,6 +75,7 @@ worker_config:
7575
cache_transceiver_config:
7676
max_tokens_in_buffer: 131104
7777
backend: UCX
78+
kv_transfer_timeout_ms: 600000
7879
disable_overlap_scheduler: true
7980
speculative_config: &id001
8081
decoding_type: MTP
@@ -101,5 +102,6 @@ worker_config:
101102
cache_transceiver_config:
102103
max_tokens_in_buffer: 131104
103104
backend: UCX
105+
kv_transfer_timeout_ms: 600000
104106
disable_overlap_scheduler: true
105107
speculative_config: *id001

tests/scripts/perf/disaggregated/gb300_deepseek-v32-fp4_32k4k_con256_ctx1_dep8_gen1_dep8_eplb0_mtp0_ccb-NIXL.yaml

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -70,6 +70,7 @@ worker_config:
7070
cache_transceiver_config:
7171
max_tokens_in_buffer: 120000
7272
backend: NIXL
73+
kv_transfer_timeout_ms: 600000
7374
num_postprocess_workers: 8
7475
stream_interval: 10
7576
ctx:
@@ -97,3 +98,4 @@ worker_config:
9798
cache_transceiver_config:
9899
max_tokens_in_buffer: 120000
99100
backend: NIXL
101+
kv_transfer_timeout_ms: 600000

0 commit comments

Comments
 (0)