- [PRF](#prf): HMAC-SHA-256
- [AEAD](#aead): AES-256-GCM if the CPU supports AES-NI, ChaCha20-Poly1305 otherwise. More specifically AES-256-GCM is only selected for X86-64 CPUs with AES-NI extension.

A smaller CPU-only package is also available:

```
$ pip install tensorflow-cpu
```

To update TensorFlow to the latest version, add `--upgrade` flag to the above
commands.

*Nightly binaries are available for testing using the
[tf-nightly](https://pypi.python.org/pypi/tf-nightly) and
[tf-nightly-cpu](https://pypi.python.org/pypi/tf-nightly-cpu) packages on PyPi.*

- 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)

Setting `max_sleep` to a *lower* value and setting `max_io` to a *higher* value would make heal go faster.

Each node is responsible of healing its local drives; Each drive will have multiple heal workers which is the quarter of the number of CPU cores of the node or the quarter of the configured nr_requests of the drive (https://www.kernel.org/doc/Documentation/block/queue-sysfs.txt). It is also possible to provide a custom number of workers by using this command: `mc admin config set alias/...

```yaml
apiVersion: install.istio.io/v1alpha1
kind: IstioOperator
spec:
  components:
    pilot:
      k8s:
        resources:
          requests:
            cpu: 1000m # override from default 500m
            memory: 4096Mi # ... default 2048Mi
        hpaSpec:
          maxReplicas: 10 # ... default 5
          minReplicas: 2  # ... default 1

To perform bi-directional replication, repeat the above process on the target site - this time setting the source bucket as the replication target. It is recommended that replication be run in a system with at least two CPU's available to the process, so that replication can run in its own thread.

![put](https://raw.githubusercontent.com/minio/minio/master/docs/bucket/replication/PUT_bucket_replication.png)

streaming compression due to its stability and performance.

This algorithm is specifically optimized for machine generated content.
Write throughput is typically at least 500MB/s per CPU core,
and scales with the number of available CPU cores.
Decompression speed is typically at least 1GB/s.

This means that in cases where raw IO is below these numbers
compression will not only reduce disk usage but also help increase system throughput.

These "env" files match up with an environment matrix that roughly covers:

-   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

Example:

```sh
minio server /data{1...4}
```

The command takes no flags

```sh
mc support diagnostics myminio/
```

The output printed will be of the form

```sh
● Admin Info ... ✔ 
● CPU ... ✔ 
● Disk Hardware ... ✔ 
● Os Info ... ✔ 
● Mem Info ... ✔ 
● Process Info ... ✔ 
● Config ... ✔ 
● Drive ... ✔ 
● Net ... ✔ 
*********************************************************************************