net.ipv4.tcp_mtu_probing=1
net.ipv4.tcp_base_mss=1280

# Disable ipv6
net.ipv6.conf.all.disable_ipv6=1
net.ipv6.conf.default.disable_ipv6=1
net.ipv6.conf.lo.disable_ipv6=1

[bootloader]
# Avoid firing timers for all CPUs at the same time. This is irrelevant for
# full nohz systems

shared state, and the operations that mutate that state. It maps the nodes of a dataflow graph across many machines in a cluster, and within a machine across multiple computational devices, including multicore CPUs, general purpose GPUs, and custom-designed ASICs known as Tensor Processing Units (TPUs). This architecture gives flexibility to the application developer, whereas in previous “parameter server” designs the management of shared state is built into the system, TensorFlow enables developers to...

## Resumen

Puedes usar múltiples worker processes con la opción CLI `--workers` con los comandos `fastapi` o `uvicorn` para aprovechar los **CPUs de múltiples núcleos**, para ejecutar **múltiples procesos en paralelo**.

## Recap { #recap }

You can use multiple worker processes with the `--workers` CLI option with the `fastapi` or `uvicorn` commands to take advantage of **multi-core CPUs**, to run **multiple processes in parallel**.

You could use these tools and ideas if you are setting up **your own deployment system** while taking care of the other deployment concepts yourself.

## Recapitular

Você pode usar vários processos de trabalho com a opção CLI `--workers` com os comandos `fastapi` ou `uvicorn` para aproveitar as vantagens de **CPUs multi-core** e executar **vários processos em paralelo**.

Você pode usar essas ferramentas e ideias se estiver configurando **seu próprio sistema de implantação** enquanto cuida dos outros conceitos de implantação.

	sysCPUAvgIdleMD   = NewGaugeMD(sysCPUAvgIdle, "Average CPU idle time")
	sysCPUAvgIOWaitMD = NewGaugeMD(sysCPUAvgIOWait, "Average CPU IOWait time")
	sysCPULoadMD      = NewGaugeMD(sysCPULoad, "CPU load average 1min")
	sysCPULoadPercMD  = NewGaugeMD(sysCPULoadPerc, "CPU load average 1min (percentage)")
	sysCPUNiceMD      = NewGaugeMD(sysCPUNice, "CPU nice time")
	sysCPUStealMD     = NewGaugeMD(sysCPUSteal, "CPU steal time")

### Hardware attacks

Physical GPUs or TPUs can also be the target of attacks. [Published
research](https://scholar.google.com/scholar?q=gpu+side+channel) shows that it
might be possible to use side channel attacks on the GPU to leak data from other
running models or processes in the same system. GPUs can also have
implementation bugs that might allow attackers to leave malicious code running

-   Different Python versions
-   Linux, MacOS, and Windows machines (these pool definitions are internal)
-   x86 and arm64
-   CPU-only, or with NVIDIA CUDA support (Linux only), or with TPUs

## How to Test Your Changes to TensorFlow

You may check how your changes will affect TensorFlow by:

1. Creating a PR and observing the presubmit test results

load(
    "@rules_ml_toolchain//third_party/gpus/cuda/hermetic:cuda_json_init_repository.bzl",
    "cuda_json_init_repository",
)

cuda_json_init_repository()

load(
    "@cuda_redist_json//:distributions.bzl",
    "CUDA_REDISTRIBUTIONS",
    "CUDNN_REDISTRIBUTIONS",
)
load(
    "@rules_ml_toolchain//third_party/gpus/cuda/hermetic:cuda_redist_init_repositories.bzl",

- Nightly builds:
  - [Linux CPU-only](https://storage.googleapis.com/tensorflow-nightly/github/tensorflow/lib_package/libtensorflow-cpu-linux-x86_64.tar.gz)
  - [Linux GPU](https://storage.googleapis.com/tensorflow-nightly/github/tensorflow/lib_package/libtensorflow-gpu-linux-x86_64.tar.gz)