TensorRT-LLMs/tests/quantization/test_smooth_quant_gemm.py
石晓伟 8f91cff22e
TensorRT-LLM Release 0.15.0 (#2529)
Co-authored-by: Kaiyu Xie <26294424+kaiyux@users.noreply.github.com>
2024-12-04 13:44:56 +08:00

135 lines
5.4 KiB
Python

# SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import os
import sys
import unittest
from itertools import chain, product
import _utils
import numpy as np
import tensorrt as trt
import torch
from parameterized import parameterized
from polygraphy.backend.trt import CreateConfig, EngineFromNetwork, TrtRunner
import tensorrt_llm
from tensorrt_llm import Tensor
from tensorrt_llm.quantization.functional import smooth_quant_gemm
sys.path.append(os.path.join(os.path.dirname(__file__), '..'))
from utils.util import skip_pre_ampere, unittest_name_func
class TestSmoothQuantGemm(unittest.TestCase):
def setUp(self):
tensorrt_llm.logger.set_level('error')
def _sq_gemm(self, m, n, k, dtype, per_token_scaling, per_channel_scaling,
use_plugin):
# Init operands for multiplication in int32
shape1 = (m, k)
mat1 = torch.randint(-128, 128, shape1, dtype=torch.int8)
shape2 = (n, k)
mat2 = torch.randint(-128, 128, shape2, dtype=torch.int8)
# Init scales in fp32
shape_scale_a = (m, 1) if per_token_scaling else (1, 1)
scale_a_torch = torch.ones(shape_scale_a, dtype=torch.float32) * 1e-2
scale_a_torch *= torch.randint(1,
10,
shape_scale_a,
dtype=torch.float32)
shape_scale_b = (1, n) if per_channel_scaling else (1, 1)
scale_b_torch = torch.ones(shape_scale_b, dtype=torch.float32) * 1e-2
scale_b_torch *= torch.randint(1,
10,
shape_scale_b,
dtype=torch.float32)
# Create builder
builder = tensorrt_llm.Builder()
# Create empty network
network = builder.create_network()
# Allow SQ plugin of dtype type
if use_plugin:
network.plugin_config.smooth_quant_gemm_plugin = dtype
with tensorrt_llm.net_guard(network):
# Init TensorRT-LLM tensor for mat1
x = Tensor(name='x',
shape=mat1.shape,
dtype=tensorrt_llm._utils.str_dtype_to_trt("int8"))
# Init TensorRT-LLM tensor for mat2
y = Tensor(name='y',
shape=mat2.shape,
dtype=tensorrt_llm._utils.str_dtype_to_trt("int8"))
# Init TensorRT-LLM tensor for per token scaling
scale_a = tensorrt_llm.functional.constant(scale_a_torch.numpy())
# Init TensorRT-LLM tensor for per channel scaling
scale_b = tensorrt_llm.functional.constant(scale_b_torch.numpy())
# Get output tensor for SQ gemm
output = smooth_quant_gemm(x, y, scale_a, scale_b,
per_token_scaling, per_channel_scaling,
dtype)
output.mark_output('output', dtype)
# TODO: When dtype=int32, per_token_scaling=False, per_channel_scaling=False,
# This test will break using new API on A30, only when running with all other unit tests.
# This is a weird issue, so skip changing this file.
engine = EngineFromNetwork(
(builder.trt_builder, network.trt_network),
config=CreateConfig(
memory_pool_limits={trt.MemoryPoolType.WORKSPACE: 33554432}))
# Infer engine
with TrtRunner(engine) as runner:
outputs = runner.infer(feed_dict={
'x': mat1.numpy(),
'y': mat2.numpy(),
})
ref = _utils.gt_matmul_smooth_quant(mat1,
mat2,
scale_a_torch,
scale_b_torch,
dtype,
bias=None)
np.testing.assert_allclose(ref.cpu().numpy(), outputs['output'])
@parameterized.expand(chain(
product(["float16", "float32", "int32"], [True, False], [True, False],
[True]),
product(["float16", "float32"], [True, False], [True, False], [False])),
name_func=unittest_name_func)
@skip_pre_ampere # SmoothQuant is not supported in pre-Ampere
def test_matmul(self, dtype, per_token_scaling, per_channel_scaling,
use_plugin):
bs = 2
inseq = 16
hidden_size = 768
# qkv_gemm
self._sq_gemm(bs * inseq, 3 * hidden_size, hidden_size, dtype,
per_token_scaling, per_channel_scaling, use_plugin)
# mlp_gemm_1
self._sq_gemm(bs * inseq, 4 * hidden_size, hidden_size, dtype,
per_channel_scaling, per_token_scaling, use_plugin)
if __name__ == '__main__':
unittest.main()