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

## Benchmarks e velocidade { #benchmarks-and-speed }

Ao verificar os _benchmarks_, é comum observar algumas ferramentas de diferentes tipos comparadas como equivalentes.

Especificamente, observa-se Uvicorn, Starlette e FastAPI comparados juntos (entre muitas outras ferramentas).

Al agregar una estrella, otros usuarios podrán encontrarlo más fácilmente y ver que ya ha sido útil para otros.

## Observa el repositorio de GitHub para lanzamientos { #watch-the-github-repository-for-releases }

Puedes "observar" FastAPI en GitHub (haciendo clic en el botón "watch" en la parte superior derecha): [https://github.com/fastapi/fastapi](https://github.com/fastapi/fastapi). 👀

If a synchronizing read-like memory operation <i>r</i>
observes a synchronizing write-like memory operation <i>w</i>
(that is, if <i>W</i>(<i>r</i>) = <i>w</i>),
then <i>w</i> is synchronized before <i>r</i>.
Informally, the synchronized before relation is a subset of the implied total order
mentioned in the previous paragraph,
limited to the information that <i>W</i> directly observes.
</p>

<p>

      assertThat(striped.getAt(i)).isSameInstanceAs(object); // idempotent
      observed.add(object);
    }
    assertTrue("All stripes observed", observed.size() == striped.size());

    // this uses #get(key), makes sure an already observed stripe is returned
    for (int i = 0; i < striped.size() * 100; i++) {
      assertTrue(observed.contains(striped.get(new Object())));
    }

    try {
      striped.getAt(-1);

      assertThat(striped.getAt(i)).isSameInstanceAs(object); // idempotent
      observed.add(object);
    }
    assertTrue("All stripes observed", observed.size() == striped.size());

    // this uses #get(key), makes sure an already observed stripe is returned
    for (int i = 0; i < striped.size() * 100; i++) {
      assertTrue(observed.contains(striped.get(new Object())));
    }

    try {
      striped.getAt(-1);

        int falseCount = 0;
        for (int i = 0; i < readerThreadCount; i++) {
            for (Boolean observed : observations[i]) {
                if (!observed) {
                    falseCount++;
                }
            }
        }
        assertTrue("Some threads should have observed alive=false", falseCount > 0);
    }

    /**

        }
      }
      return;
    }

    /*
     * This is an unsynchronized read! After the read, the function returns immediately or acquires
     * the lock to check again. Since an IDLE state was observed inside the preceding synchronized
     * block, and reference field assignment is atomic, this may save reacquiring the lock when
     * another thread or the worker task has cleared the count and set the state.
     *

    }

    /**
     * Checks if `expected` and `observed` are equal when viewed as a set and headers are
     * deduped.
     *
     * TODO: See if duped headers should be preserved on decode and verify.
     */
    private fun assertSetEquals(
      message: String,
      expected: List<Header>,
      observed: List<Header>,
    ) {
      assertThat(LinkedHashSet(observed), message)
        .isEqualTo(LinkedHashSet(expected))

            }

            // Pre-store the checksums as any future puts will overwrite them
            for (String extension : checksums.keySet()) {
                ChecksumObserver observer = checksums.get(extension);
                sums.put(extension, observer.getActualChecksum());
            }

            // We do this in here so we can checksum the artifact metadata too, otherwise it could be metadata itself

   * table is internally partitioned to try to permit the indicated number of concurrent updates
   * without contention. Because assignment of entries to these partitions is not necessarily
   * uniform, the actual concurrency observed may vary. Ideally, you should choose a value to
   * accommodate as many threads as will ever concurrently modify the table. Using a significantly