| 1 |
On Sun, Dec 18, 2022 at 2:30 PM Frank Steinmetzger <Warp_7@×××.de> wrote: |
| 2 |
> |
| 3 |
> Am Sun, Dec 18, 2022 at 01:07:43PM -0600 schrieb Dale: |
| 4 |
> |
| 5 |
> > Mostly, I need a better CPU. If I encrypt anyway. |
| 6 |
> |
| 7 |
> Did you ever tell us the exact CPU you have in there? All I can remember is |
| 8 |
> it has 4 cores. And some AMD processor with a II in its name, but that was |
| 9 |
> you main rig, right? |
| 10 |
|
| 11 |
What encryption algorithm are you using? You should see if this is |
| 12 |
hardware-accelerated in the kernel for your CPU, or if not if there is |
| 13 |
another strong algorithm which is. Most newer CPUs will tend to have |
| 14 |
hardware support for algorithms like AES, and the kernel will use |
| 15 |
this. This will greatly improve CPU performance. |
| 16 |
|
| 17 |
I've run into this issue with zfs on Raspberry Pis. ZFS does the |
| 18 |
encryption internally, and the openzfs code didn't have support for |
| 19 |
ARM hardware encryption the last time I checked (this could have |
| 20 |
changed). I found that dm-crypt works MUCH better on Pis as a result, |
| 21 |
as the kernel does have ARM encryption hardware support. |
| 22 |
|
| 23 |
Again, this all depends on the algorithm. If you're using something |
| 24 |
exotic odds are the hardware won't handle it natively. |
| 25 |
|
| 26 |
-- |
| 27 |
Rich |
Archives