# Raspberry Pi Zero 2

**URL:** https://community.openconversational.ai/t/raspberry-pi-zero-2/11459
**Category:** Picroft
**Created:** [November 3, 2021, 2:39pm UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459 "2021-11-03T14:39:43Z")
**Posts on this page:** 8
**Page:** 1

<div class="post-metadata">

### Author: ![GiantLameChips](https://community.openconversational.ai/user_avatar/community.openconversational.ai/giantlamechips/32/3377_2.png) [@GiantLameChips](https://community.openconversational.ai/u/GiantLameChips)
#### Post date: [November 3, 2021, 2:39pm UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459/1 "2021-11-03T14:39:43Z")

</div>

Has anyone tested Picroft on the zero 2?

---

<div class="post-metadata">

### Author: ![goldyfruit](https://community.openconversational.ai/user_avatar/community.openconversational.ai/goldyfruit/32/2741_2.png) [@goldyfruit](https://community.openconversational.ai/u/goldyfruit)
#### Post date: [November 3, 2021, 5:17pm UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459/2 "2021-11-03T17:17:55Z")

</div>

I think it could be a bit short in resources, 512MB of memory will not be enough in my opinion.

---

<div class="post-metadata">

### Author: ![baconator](https://community.openconversational.ai/letter_avatar_proxy/v4/letter/b/d07c76/32.png) [@baconator](https://community.openconversational.ai/u/baconator)
#### Post date: [November 4, 2021, 9:36pm UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459/3 "2021-11-04T21:36:08Z")

</div>

It’s possible to run it, but without more ram you’d likely be limited on skills and/or speed.

---

<div class="post-metadata">

### Author: ![goldyfruit](https://community.openconversational.ai/user_avatar/community.openconversational.ai/goldyfruit/32/2741_2.png) [@goldyfruit](https://community.openconversational.ai/u/goldyfruit)
#### Post date: [November 8, 2021, 7:30pm UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459/4 "2021-11-08T19:30:18Z")

</div>

I ran some tests, not on a RPi0 2 but on a RPi 4B. I’m able to run Raspberry Pi OS Bullseye 64-bit without X server.

Mycroft AI is running within Docker containers and it consumes `518MB` of RAM, I guess if you remove Docker from the equation you would be able to run it as @baconator mentioned.

```auto
pi@rpi4b01:~ $ free -m
               total used free shared buff/cache available
Mem: 7812 518 6452 1 841 7172
Swap: 99 0 99

```

---

<div class="post-metadata">

### Author: ![goldyfruit](https://community.openconversational.ai/user_avatar/community.openconversational.ai/goldyfruit/32/2741_2.png) [@goldyfruit](https://community.openconversational.ai/u/goldyfruit)
#### Post date: [November 8, 2021, 8:34pm UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459/5 "2021-11-08T20:34:12Z")

</div>

I rebooted everything and ask many thing to Mycroft, here is the memory usage after 40 minutes.

```auto
pi@rpi4b01:~ $ uptime
 20:33:53 up 40 min, 1 user, load average: 0.74, 0.53, 0.47

pi@rpi4b01:~ $ free -m
               total used free shared buff/cache available
Mem: 7812 498 6597 1 716 7190
Swap: 99 0 99

```

---

<div class="post-metadata">

### Author: ![ChanceNCounter](https://community.openconversational.ai/user_avatar/community.openconversational.ai/chancencounter/32/1965_2.png) [@ChanceNCounter](https://community.openconversational.ai/u/ChanceNCounter)
#### Post date: [November 9, 2021, 1:00am UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459/6 "2021-11-09T01:00:38Z")

</div>

Boy, that’s cutting it close. Remember you’ve still got whatever Skills want to do.

What’s your wake word engine there?

---

<div class="post-metadata">

### Author: ![goldyfruit](https://community.openconversational.ai/user_avatar/community.openconversational.ai/goldyfruit/32/2741_2.png) [@goldyfruit](https://community.openconversational.ai/u/goldyfruit)
#### Post date: [November 9, 2021, 1:06am UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459/7 "2021-11-09T01:06:39Z")

</div>

I’m running `precise-lite`. These are the skills installed for now:

| Skill | Status |
| --- | --- |
| mycroft-joke | [installed] |
| fallback-query | [installed] |
| mycroft-audio-record | [installed] |
| mycroft-fallback-duck-duck-go | [installed] |
| mycroft-playback-control | [installed] |
| mycroft-version-checker | [installed] |
| mycroft-speak | [installed] |
| mycroft-wiki | [installed] |
| mycroft-reminder | [installed] |
| mycroft-installer | [installed] |
| mycroft-hello-world | [installed] |
| mycroft-ip | [installed] |
| mycroft-personal | [installed] |
| mycroft-naptime | [installed] |
| mycroft-support-helper | [installed] |
| mycroft-pairing | [installed] |
| mycroft-sonos-controller-skill.smartgic | [installed] |
| mycroft-stop | [installed] |
| mycroft-spelling | [installed] |
| mycroft-singing | [installed] |
| mycroft-configuration | [installed] |
| mycroft-volume | [installed] |
| homeassistant | [installed] |
| mycroft-timer | [installed] |
| mycroft-wakeword-led-gpio-skill.smartgic | [installed] |
| mycroft-npr-news | [installed] |
| fallback-unknown | [installed] |
| mycroft-weather | [installed] |
| mycroft-date-time | [installed] |
| mycroft-alarm | [installed] |
| fallback-wolfram-alpha | [installed] |

---

<div class="post-metadata">

### Author: ![ChanceNCounter](https://community.openconversational.ai/user_avatar/community.openconversational.ai/chancencounter/32/1965_2.png) [@ChanceNCounter](https://community.openconversational.ai/u/ChanceNCounter)
#### Post date: [November 9, 2021, 1:27am UTC](https://community.openconversational.ai/t/raspberry-pi-zero-2/11459/8 "2021-11-09T01:27:23Z")

</div>

Well, there goes my idea 😉

The Pi02’s a great candidate for a client device. I’m excited to use it as a HiveMind dev platform! But I think you’ll find you’re skating too close to its memory limit with relatively few, very light skills going.

On the other hand, it’s pretty neat to learn that the whole stack can run inside of 500MB RAM, at least with online voice. An even lighter version running as a kiosk could definitely live in there.

Of course, that assumes the Zero doesn’t present additional surprises in practice. Maybe it lacks the Mazda factor (no _zoom zoom_)
