Convert benchmarks to ruff format (#18068)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2025-05-13 13:43:29 +00:00
committed by GitHub
parent b922c2ebd2
commit 009d9e7590
41 changed files with 3980 additions and 2938 deletions
+46 -37
View File
@@ -6,8 +6,7 @@ from typing import Optional
import nvtx
import torch
from vllm.model_executor.layers.rotary_embedding import (RotaryEmbedding,
get_rope)
from vllm.model_executor.layers.rotary_embedding import RotaryEmbedding, get_rope
from vllm.platforms import current_platform
from vllm.utils import FlexibleArgumentParser
@@ -32,40 +31,49 @@ def benchmark_rope_kernels_multi_lora(
# silulating serving 4 LoRAs
scaling_factors = [1, 2, 4, 8]
# batched RoPE can take multiple scaling factors
batched_rope = get_rope(head_size, rotary_dim, max_position, base,
is_neox_style, {
"rope_type": "linear",
"factor": tuple(scaling_factors)
})
batched_rope = get_rope(
head_size,
rotary_dim,
max_position,
base,
is_neox_style,
{"rope_type": "linear", "factor": tuple(scaling_factors)},
)
# non-batched RoPE takes only one scaling factor, we create multiple
# instances to simulate the same behavior
non_batched_ropes: list[RotaryEmbedding] = []
for scaling_factor in scaling_factors:
non_batched_ropes.append(
get_rope(head_size, rotary_dim, max_position, base, is_neox_style,
{
"rope_type": "linear",
"factor": (scaling_factor, )
}))
get_rope(
head_size,
rotary_dim,
max_position,
base,
is_neox_style,
{"rope_type": "linear", "factor": (scaling_factor,)},
)
)
positions = torch.randint(0, max_position, (batch_size, seq_len))
query = torch.randn(batch_size,
seq_len,
num_heads * head_size,
dtype=dtype)
query = torch.randn(batch_size, seq_len, num_heads * head_size, dtype=dtype)
key = torch.randn_like(query)
# create query offsets for batched RoPE, we concat multiple kv cache
# together and each query needs to find the right kv cache of its type
offset_map = torch.tensor(
list(
accumulate([0] + [
max_position * scaling_factor * 2
for scaling_factor in scaling_factors[:-1]
])))
query_types = torch.randint(0,
len(scaling_factors), (batch_size, seq_len),
device=device)
accumulate(
[0]
+ [
max_position * scaling_factor * 2
for scaling_factor in scaling_factors[:-1]
]
)
)
)
query_types = torch.randint(
0, len(scaling_factors), (batch_size, seq_len), device=device
)
# map query types to offsets
query_offsets = offset_map[query_types]
# the kernel takes flattened offsets
@@ -86,27 +94,28 @@ def benchmark_rope_kernels_multi_lora(
torch.cuda.synchronize()
if __name__ == '__main__':
if __name__ == "__main__":
parser = FlexibleArgumentParser(
description="Benchmark the rotary embedding kernels.")
description="Benchmark the rotary embedding kernels."
)
parser.add_argument("--is-neox-style", type=bool, default=True)
parser.add_argument("--batch-size", type=int, default=16)
parser.add_argument("--seq-len", type=int, default=512)
parser.add_argument("--num-heads", type=int, default=8)
parser.add_argument("--head-size",
type=int,
choices=[64, 80, 96, 112, 120, 128, 192, 256],
default=128)
parser.add_argument(
"--head-size",
type=int,
choices=[64, 80, 96, 112, 120, 128, 192, 256],
default=128,
)
parser.add_argument("--rotary-dim", type=int, choices=[16, 32], default=32)
parser.add_argument("--dtype",
type=str,
choices=["bfloat16", "float"],
default="float")
parser.add_argument(
"--dtype", type=str, choices=["bfloat16", "float"], default="float"
)
parser.add_argument("--seed", type=int, default=0)
parser.add_argument("--device",
type=str,
choices=["cuda:0", "cuda:1"],
default="cuda:0")
parser.add_argument(
"--device", type=str, choices=["cuda:0", "cuda:1"], default="cuda:0"
)
args = parser.parse_args()
print(args)