# 基准测试 { #benchmarks }

第三方机构 TechEmpower 的基准测试表明在 Uvicorn 下运行的 **FastAPI** 应用程序是 [可用的最快的 Python 框架之一](https://www.techempower.com/benchmarks/#section=test&runid=7464e520-0dc2-473d-bd34-dbdfd7e85911&hw=ph&test=query&l=zijzen-7)，仅次于 Starlette 和 Uvicorn 本身（由 FastAPI 内部使用）。

但是在查看基准得分和对比时，请注意以下几点。

## 基准测试和速度 { #benchmarks-and-speed }

当你查看基准测试时，几个不同类型的工具被等效地做比较是很常见的情况。

具体来说，是将 Uvicorn，Starlette 和 FastAPI 一起比较（在许多其它工具中）。

# 基準測試 { #benchmarks }

由第三方機構 TechEmpower 的基準測試表明在 Uvicorn 下運行的 **FastAPI** 應用程式是 [最快的 Python 可用框架之一](https://www.techempower.com/benchmarks/#section=test&runid=7464e520-0dc2-473d-bd34-dbdfd7e85911&hw=ph&test=query&l=zijzen-7)，僅次於 Starlette 和 Uvicorn 本身（於 FastAPI 內部使用）。

但是在查看基準得分和對比時，請注意以下幾點。

## 基準測試和速度 { #benchmarks-and-speed }

當你查看基準測試時，常見到不同類型的多個工具被視為等同來比較。

具體來說，是將 Uvicorn、Starlette 和 FastAPI 同時進行比較（以及許多其他工具）。

// A simple logging device to test custom device registration.
#include <memory>

#include "tensorflow/c/c_api.h"
#include "tensorflow/c/eager/c_api.h"
#include "tensorflow/c/eager/c_api_experimental.h"
#include "tensorflow/c/eager/c_api_test_util.h"
#include "tensorflow/c/tf_status.h"
#include "tensorflow/core/lib/gtl/cleanup.h"
#include "tensorflow/core/platform/test.h"

namespace {

struct LoggingDevice {

sleep 200

# Test whether most recent tag update on minio2 is replicated to minio1
val=$(./mc tag list minio1/newbucket --json | jq -r .tagset | jq -r .key)
if [ "${val}" != "val2" ]; then
	echo "expected bucket tag to have replicated, exiting..."
	exit_1
fi

# Test if bucket created/deleted when minio1 is down healed

# Gradle
# ------
.gradle
/build
/*/build
/*/*/build
/*/*/*/build
/*/*/*/*/*/build
/**/docs/src/samples/**/build
/**/docs/src/snippets/**/build
/*/internal-android-performance-testing/build-android-libs
test-splits/
/gradle/verification-keyring.gpg

# Kotlin
# ------
.kotlin

# IDEA
# ----
!/.idea
/.idea/*

import static com.google.common.collect.testing.Helpers.copyToList;

import com.google.common.annotations.GwtIncompatible;
import java.util.List;

/**
 * Creates, based on your criteria, a JUnit test suite that exhaustively tests a
 * ConcurrentNavigableMap implementation.
 *
 * @author Louis Wasserman
 */
@GwtIncompatible
public class ConcurrentNavigableMapTestSuiteBuilder<K, V>

  <modelVersion>4.0.0</modelVersion>
  <groupId>com.example.group</groupId>
  <artifactId>testvalidpom</artifactId>
  <version>0.0.1-SNAPSHOT</version>

  <description>
    This will test if the module validator recognized that this dependency with classifier
    is not the same as the module itself.
  </description>
  <dependencies>
    <dependency>
      <groupId>com.example.group</groupId>

  <modelVersion>4.0.0</modelVersion>
  <groupId>com.example.group</groupId>
  <artifactId>testinvalidpom</artifactId>
  <version>0.0.1-SNAPSHOT</version>

  <description>
    This will test if the module validator recognized that this
    dependency is the same as the module itself.
  </description>
  <dependencies>
    <dependency>
      <groupId>com.example.group</groupId>

    assert response.status_code == status.HTTP_200_OK
    actual_schema = response.json()
    assert (
        len(actual_schema["paths"]["/"]["get"]["parameters"]) == 1
    )  # primary goal of this test
    assert actual_schema == snapshot(
        {
            "components": {
                "schemas": {
                    "HTTPValidationError": {
                        "properties": {

   * Returns an early guess of the next response, used for policy on how an incoming request should
   * be received. The default implementation returns an empty response. Mischievous implementations
   * can return other values to test HTTP edge cases, such as unhappy socket policies or throttled
   * request bodies.
   */
  public open fun peek(): MockResponse = MockResponse()

  /**