Enable Random Prefix Caching in Serving Profiling Tool (benchmark_ser…

…ving.py) (vllm-project#8241)

Signed-off-by: Alvant <alvasian@yandex.ru>

benchmarks/benchmark_serving.py

@@ -195,8 +195,16 @@ def sample_sonnet_requests(
 
 
 def sample_random_requests(
-        input_len: int, output_len: int, num_prompts: int, range_ratio: float,
-        tokenizer: PreTrainedTokenizerBase) -> List[Tuple[str, int, int]]:
+    prefix_len: int,
+    input_len: int,
+    output_len: int,
+    num_prompts: int,
+    range_ratio: float,
+    tokenizer: PreTrainedTokenizerBase,
+) -> List[Tuple[str, int, int]]:
+    prefix_token_ids = np.random.randint(0,
+                                         tokenizer.vocab_size,
+                                         size=prefix_len).tolist()
 
     input_lens = np.random.randint(
         int(input_len * range_ratio),
@@ -211,10 +219,12 @@ def sample_random_requests(
     offsets = np.random.randint(0, tokenizer.vocab_size, size=num_prompts)
     input_requests = []
     for i in range(num_prompts):
-        prompt = tokenizer.decode([(offsets[i] + i + j) % tokenizer.vocab_size
+        prompt = tokenizer.decode(prefix_token_ids +
+                                  [(offsets[i] + i + j) % tokenizer.vocab_size
                                    for j in range(input_lens[i])])
+
         input_requests.append(
-            (prompt, int(input_lens[i]), int(output_lens[i])))
+            (prompt, int(prefix_len + input_lens[i]), int(output_lens[i])))
 
     return input_requests
 
@@ -567,6 +577,7 @@ def main(args: argparse.Namespace):
 
     elif args.dataset_name == "random":
         input_requests = sample_random_requests(
+            prefix_len=args.random_prefix_len,
             input_len=args.random_input_len,
             output_len=args.random_output_len,
             num_prompts=args.num_prompts,
@@ -765,6 +776,14 @@ def main(args: argparse.Namespace):
         help="Range of sampled ratio of input/output length, "
         "used only for random sampling.",
     )
+    parser.add_argument(
+        "--random-prefix-len",
+        type=int,
+        default=0,
+        help="Number of fixed prefix tokens before random "
+        " context. The length range of context in a random "
+        " request is [random-prefix-len, "
+        " random-prefix-len + random-prefix-len * random-range-ratio).")
     parser.add_argument(
         "--request-rate",
         type=float,