# CPU WHEEL
test:linux_cpu_wheel_test_filters --test_tag_filters=-no_oss,-tf_tosa,-oss_excluded,-oss_serial,-gpu,-tpu,-benchmark-test,-v1only,-no_oss_py310,-no_oss_py313
test:linux_cpu_wheel_test_filters --build_tag_filters=-no_oss,-tf_tosa,-oss_excluded,-oss_serial,-gpu,-tpu,-benchmark-test,-v1only,-no_oss_py310,-no_oss_py313
test:linux_cpu_wheel_test_filters --test_lang_filters=py --test_size_filters=small,medium

            uv.lock
      - name: Install Dependencies
        run: uv sync --no-dev --group tests --extra all
      - name: CodSpeed benchmarks
        uses: CodSpeedHQ/action@v4
        with:
          mode: simulation
          run: uv run --no-sync pytest tests/benchmarks --codspeed

  coverage-combine:
    needs:
      - test
    runs-on: ubuntu-latest
    steps:
      - name: Dump GitHub context

import static java.lang.Long.parseLong;

import java.util.Random;
import org.jspecify.annotations.NullUnmarked;

/**
 * Utility class for being able to seed a {@link Random} value with a passed in seed from a
 * benchmark parameter.
 *
 * <p>TODO: Remove this class once Caliper has a better way.
 *
 * @author Nicholaus Shupe
 */
@NullUnmarked
public final class SpecialRandom extends Random {

        assertTrue(queryFieldConfig.searchFieldSet.contains("field3"));
    }

    /**
     * Test performance improvement of Set-based lookup over array-based lookup.
     * This is a micro-benchmark to demonstrate the O(1) vs O(n) improvement.
     */
    @Test
    public void test_lookupPerformance_SetFasterThanArray() {
        // Create a large array of fields
        int fieldCount = 1000;

### Migration

*   Python 3.3+ support via changes to python codebase and ability to specify
    python version via ./configure.

*   Some improvements to GPU performance and memory usage:
    [convnet benchmarks](https://github.com/soumith/convnet-benchmarks/issues/66)
    roughly equivalent with native cudnn v2 performance. Improvements mostly due
    to moving to 32-bit indices, faster shuffling kernels. More improvements to
    come in later releases.

# Benchmarks { #benchmarks }

Independent TechEmpower benchmarks show **FastAPI** applications running under Uvicorn as [one of the fastest Python frameworks available](https://www.techempower.com/benchmarks/#section=test&runid=7464e520-0dc2-473d-bd34-dbdfd7e85911&hw=ph&test=query&l=zijzen-7), only below Starlette and Uvicorn themselves (used internally by FastAPI).

But when checking benchmarks and comparisons you should keep the following in mind.

# Benchmarks { #benchmarks }

Los benchmarks independientes de TechEmpower muestran aplicaciones de **FastAPI** ejecutándose bajo Uvicorn como [uno de los frameworks de Python más rápidos disponibles](https://www.techempower.com/benchmarks/#section=test&runid=7464e520-0dc2-473d-bd34-dbdfd7e85911&hw=ph&test=query&l=zijzen-7), solo por debajo de Starlette y Uvicorn en sí mismos (utilizados internamente por FastAPI).

# Benchmarks { #benchmarks }

Benchmarks independentes da TechEmpower mostram as aplicações **FastAPI** rodando com Uvicorn como [um dos frameworks Python mais rápidos disponíveis](https://www.techempower.com/benchmarks/#section=test&runid=7464e520-0dc2-473d-bd34-dbdfd7e85911&hw=ph&test=query&l=zijzen-7), somente atrás dos próprios Starlette e Uvicorn (utilizados internamente pelo FastAPI).

Mas quando se checa _benchmarks_ e comparações você deveria ter o seguinte em mente.

# Benchmarks { #benchmarks }

Unabhängige TechEmpower-Benchmarks zeigen **FastAPI**-Anwendungen, die unter Uvicorn ausgeführt werden, als [eines der schnellsten verfügbaren Python-Frameworks](https://www.techempower.com/benchmarks/#section=test&runid=7464e520-0dc2-473d-bd34-dbdfd7e85911&hw=ph&test=query&l=zijzen-7), nur unterhalb von Starlette und Uvicorn selbst (die intern von FastAPI verwendet werden).