Just built a home security gateway using a #RaspberryPi and #WireGuard (via ProtonVPN) 🛡️
-
@nickbearded Barely using CPU, but how much power is it sipping? I know it’s not much but always looking for quantifiable data. You’ll need a kill-a-watt or energy monitoring plug.
@dhry my Raspberry Pi 4 sips 5V DC via USB-C connector.
-
@dhry my Raspberry Pi 4 sips 5V DC via USB-C connector.
@nickbearded I have three (gathering dust for years) and know that part - but how much energy does it use? kWh. Question might be unclear so attaching a screenshot showing energy usage for the month for my windows machines. Tapo P110M used. Hint - you can’t get this from htop.
-
@nickbearded I have three (gathering dust for years) and know that part - but how much energy does it use? kWh. Question might be unclear so attaching a screenshot showing energy usage for the month for my windows machines. Tapo P110M used. Hint - you can’t get this from htop.
@dhry Fair point! While htop doesn't show kWh, the physics of a Pi 4 are well-documented. At the idle state shown (near 0% CPU), it draws ~3W.
Doing the math: 3W * 24h * 30 days = 2.16 kWh per month.
Comparing that to your 55 kWh Windows setup? My gateway uses about 25 times less energy to do the same job 24/7. That’s the beauty of ARM vs x86 for simple networking tasks
-
@dhry Fair point! While htop doesn't show kWh, the physics of a Pi 4 are well-documented. At the idle state shown (near 0% CPU), it draws ~3W.
Doing the math: 3W * 24h * 30 days = 2.16 kWh per month.
Comparing that to your 55 kWh Windows setup? My gateway uses about 25 times less energy to do the same job 24/7. That’s the beauty of ARM vs x86 for simple networking tasks@nickbearded Ok I’ll retract the question. Not after theoreticals.
-
@nickbearded Ok I’ll retract the question. Not after theoreticals.
@dhry Fair enough! Real-world data is king. Though at these power levels, buying a $20 energy monitor to measure a ~$10/year electricity bill feels like overkill.
-
Just built a home security gateway using a #RaspberryPi and #WireGuard (via ProtonVPN)
️It provides a secure tunnel for all my devices, especially #IoT ones that can't run a VPN. Best part? Efficiency! As you can see in the #htop shot, it’s barely sipping power: only ~135MB of RAM and near 0% CPU.
I can toggle the VPN to manage remote IoT access while keeping a solid layer of protection. Small, silent, and rock solid.
#SelfHosted #Privacy #CyberSecurity #Linux #OpenSource #VPN
@nickbearded could you try again with btop instead? It would give a better idea of the resources. Zero% seems unlikely
-
@nickbearded could you try again with btop instead? It would give a better idea of the resources. Zero% seems unlikely
@GuillaumeRossolini I get why 0.00 load seems 'unlikely', but that’s just a well-optimized headless RPi4.
WireGuard runs in kernel space: it's invisible to the scheduler unless there's heavy traffic.
Load measures the queue: on 4 cores with no tasks waiting, 0.00 is a perfect score.
Irony: btop uses more CPU just to draw its UI than the VPN itself! I'd rather save those cycles for my data than for eye-candy
#RaspberryPi #WireGuard #SelfHosted #Linux #Networking -
@GuillaumeRossolini I get why 0.00 load seems 'unlikely', but that’s just a well-optimized headless RPi4.
WireGuard runs in kernel space: it's invisible to the scheduler unless there's heavy traffic.
Load measures the queue: on 4 cores with no tasks waiting, 0.00 is a perfect score.
Irony: btop uses more CPU just to draw its UI than the VPN itself! I'd rather save those cycles for my data than for eye-candy
#RaspberryPi #WireGuard #SelfHosted #Linux #Networking@nickbearded fair enough for the UI, and thanks for the details of why the tools show no activity
-
@nickbearded fair enough for the UI, and thanks for the details of why the tools show no activity
@nickbearded by the way all your messages show up as Italian, so it may confuse people trying to translate them
-
@GuillaumeRossolini I get why 0.00 load seems 'unlikely', but that’s just a well-optimized headless RPi4.
WireGuard runs in kernel space: it's invisible to the scheduler unless there's heavy traffic.
Load measures the queue: on 4 cores with no tasks waiting, 0.00 is a perfect score.
Irony: btop uses more CPU just to draw its UI than the VPN itself! I'd rather save those cycles for my data than for eye-candy
#RaspberryPi #WireGuard #SelfHosted #Linux #Networking@nickbearded @GuillaumeRossolini love that you’re squeezing every last cycle out of the pi for what actually matters—true efficiency right there.
-
@nickbearded @GuillaumeRossolini love that you’re squeezing every last cycle out of the pi for what actually matters—true efficiency right there.
@NewsGroup @GuillaumeRossolini this one was under Raspberry Pi OS. I did the same on the same Pi but using Armbian: same result, just it was a bit tricky to setup.
-
@NewsGroup @GuillaumeRossolini this one was under Raspberry Pi OS. I did the same on the same Pi but using Armbian: same result, just it was a bit tricky to setup.
@nickbearded @GuillaumeRossolini nice to hear it's consistent across different distros, even if the setup was a bit tricky.
-
R relay@relay.publicsquare.global shared this topic
