Skip to content

Commit d806bda

Browse files
committed
[no ci] Add deepseek-ai/DeepSeek-V3.2-Exp
Signed-off-by: Huy Do <[email protected]>
1 parent 64925c5 commit d806bda

File tree

3 files changed

+46
-3
lines changed

3 files changed

+46
-3
lines changed

vllm-benchmarks/benchmarks/cuda/latency-tests.json

Lines changed: 12 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -85,7 +85,7 @@
8585
}
8686
},
8787
{
88-
"test_name": "latency_deepseek_v3_tp8",
88+
"test_name": "latency_deepseek_v3_1_tp8",
8989
"parameters": {
9090
"model": "deepseek-ai/DeepSeek-V3.1",
9191
"tensor_parallel_size": 8,
@@ -95,6 +95,17 @@
9595
"max_model_len": 8192
9696
}
9797
},
98+
{
99+
"test_name": "latency_deepseek_v3_2_tp8",
100+
"parameters": {
101+
"model": "deepseek-ai/DeepSeek-V3.2-Exp",
102+
"tensor_parallel_size": 8,
103+
"load_format": "dummy",
104+
"num_iters_warmup": 5,
105+
"num_iters": 15,
106+
"max_model_len": 8192
107+
}
108+
},
98109
{
99110
"test_name": "latency_deepseek_r1_tp8",
100111
"parameters": {

vllm-benchmarks/benchmarks/cuda/serving-tests.json

Lines changed: 21 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -494,7 +494,7 @@
494494
}
495495
},
496496
{
497-
"test_name": "serving_deepseek_v3_tp8_random_in5k_out8k",
497+
"test_name": "serving_deepseek_v3_1_tp8_random_in5k_out8k",
498498
"qps_list": [1, 4, 16, "inf"],
499499
"server_parameters": {
500500
"model": "deepseek-ai/DeepSeek-V3.1",
@@ -513,6 +513,26 @@
513513
"random_output_len": 8250
514514
}
515515
},
516+
{
517+
"test_name": "serving_deepseek_v3_2_tp8_random_in5k_out8k",
518+
"qps_list": [1, 4, 16, "inf"],
519+
"server_parameters": {
520+
"model": "deepseek-ai/DeepSeek-V3.2-Exp",
521+
"tensor_parallel_size": 8,
522+
"swap_space": 16,
523+
"disable_log_stats": "",
524+
"disable_log_requests": "",
525+
"load_format": "dummy"
526+
},
527+
"client_parameters": {
528+
"model": "deepseek-ai/DeepSeek-V3.2-Exp",
529+
"backend": "vllm",
530+
"dataset_name": "random",
531+
"num_prompts": 200,
532+
"random_input_len": 5250,
533+
"random_output_len": 8250
534+
}
535+
},
516536
{
517537
"test_name": "serving_deepseek_r1_tp8_random_in5k_out8k",
518538
"qps_list": [1, 4, 16, "inf"],

vllm-benchmarks/benchmarks/cuda/throughput-tests.json

Lines changed: 13 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -93,7 +93,7 @@
9393
}
9494
},
9595
{
96-
"test_name": "throughput_deepseek_v3_tp8",
96+
"test_name": "throughput_deepseek_v3_1_tp8",
9797
"parameters": {
9898
"model": "deepseek-ai/DeepSeek-V3.1",
9999
"tensor_parallel_size": 8,
@@ -104,6 +104,18 @@
104104
"max_model_len": 8192
105105
}
106106
},
107+
{
108+
"test_name": "throughput_deepseek_v3_2_tp8",
109+
"parameters": {
110+
"model": "deepseek-ai/DeepSeek-V3.2-Exp",
111+
"tensor_parallel_size": 8,
112+
"load_format": "dummy",
113+
"dataset": "./ShareGPT_V3_unfiltered_cleaned_split.json",
114+
"num_prompts": 200,
115+
"backend": "vllm",
116+
"max_model_len": 8192
117+
}
118+
},
107119
{
108120
"test_name": "throughput_deepseek_r1_tp8",
109121
"parameters": {

0 commit comments

Comments
 (0)