Power consumption of the 6100 vs 5100
-
Yes, that's two different hardware configs. I forget if that one had an SSD at the time. The link states make more difference though, especially the 10G NICs.
-
@stephenw10 said in Power consumption of the 6100 vs 5100:
Yes, that's two different hardware configs. I forget if that one had an SSD at the time. The link states make more difference though, especially the 10G NICs.
Thank you @stephenw10 . Actually this phrase from your quote made me assume the 15W measurement involved an SSD:
With the internal devices removed (wifi, modem, SSD), only a single 1G link and powerd enabled I see ~11W.
It’s good to know that the 10 NICs attribute more. It will be years before I’ll be using those.
-
Oh there we go! Yeah I guess it did then.
-
@stephenw10: you mentioned powerd:
powerd enabled I see ~11W
Now my aim is to make sure the internals stay cool for optimal longevity. By selecting a stronger CPU (6100 vs 4100) the CPU usage should stay lower and this cooler.
But at night the unit has less to do, so powerd could save energy. Do I understand this correctly? -
Powerd will lower the CPU clock speed and voltage when it's under low load. It does it dynamically all the time if enabled.
-
My 6100 use 14-16W, 1G WAN and 2 * 1G LAN as LAG, SSD and depends on load.
PowereD is adaptiv, it's a nice low Power high throughput device.
Netgate 6100 & Netgate 2100
-
@NOCling said in Power consumption of the 6100 vs 5100:
My 6100 use 14-16W, 1G WAN and 2 * 1G LAN as LAG, SSD and depends on load.
PowereD is adaptiv, it's a nice low Power high throughput device.
Thank you :-)
So may I assume that these figures are with powerd daemon enabled?Pete
Home: SG-2100 + UniFi + Synology. SG-1100 retired
Parents: SG-1100 + UniFi + Synology
Testing: SG-1100 w/ 120GB SSD via ext USB (eMMC dead). Works great -
@NOCling I thought PowerD didn’t matter with newer Intel based CPUs as they are far better and faster at C-state jumping than software is, and thus has no effect.
-
@Cabledude I don’t think it does matter to be honest. My 6100 measures about 14 watt with two 1Gbe ports active (both using SFPs) and a 512Gb SSD installed. I do not have PowerD enabled.
-
@keyser Thank you for that, very helpful. Just curious: the 6100 Max offers a 128GB SSD but you say yours has 512GB. Is that DIY? I think I read another topic that concludes 6100 DIY SSD upgrade is not possible or at least Netgate discourages it.
-
@Cabledude Yes, I installed a 512Gb SSD myself because I wanted to be sure not kill the built-in eMMC after noticing how hard/much I was writing to the filesystem with pfBlockerNG, NtopNG and Syslog-NG installed. It would have killed my eMMC in less than a year :-)
It’s very easy to install a SSD yourself - they discurage it because you have to be a little carefull when taking it apart - It you just tear it open, you migh break the lightemmiting plastic rods that brings LED light from the MB to the front of the chassis.
-
@keyser said in Power consumption of the 6100 vs 5100:
@Cabledude Yes, I installed a 512Gb SSD myself
So does your 6100 now has two active storage entities? Or can there be only one active?
-
It only boots from one, unless you install as a mirror but I wouldn't recommend that across dissimilar devices.
pfSense only has any built in handling for one drive. Some users have added scripts to allow a separate drive for caching or logs etc but that's all custom stuff.Steve
-
@keyser I want to make sure I purchase the correct/compatible SSD. If you don't mind, can you please provide the brand/model of the 512Gb SSD you've installed? Thanks in advance.
-
@raystn said in Power consumption of the 6100 vs 5100:
@keyser I want to make sure I purchase the correct/compatible SSD. If you don't mind, can you please provide the brand/model of the 512Gb SSD you've installed? Thanks in advance.
Transcend 512GB, M.2 2242, PCIE GEN3X2, B+M KEY (TS512GMTE452T)
-
@keyser Thanks, I appreciate it.
-
I just looked up the TDP for the 4100's CPU C3338R (10,5W) and the 6100's CPU C3558 (16W). I am not a CPU geek so how does this work? Just based on the TDP's the 4100 should consume 5,5W less than the 6100 but that's not the case.
6100 readings below are all in the 14W-15W range.
A buddy of mine has a 4100 and it's consuming 14W on his meter, two 1G ports in use, eMMC and very basic config.So how does TDP translate to real world energy consumption?
@stephenw10 said in Power consumption of the 6100 vs 5100:
With the internal devices removed (wifi, modem, SSD), only an single 1G link and powerd enabled I see ~11W. That's at idle.
So 11W for a 16W TDP, this means that the CPU doesn't necessarily reach its TDP value in real action.
@stephenw10 said in Power consumption of the 6100 vs 5100:
A spot figure for a 6100 I have here is 15W with only 1x1G and 2x2.5G NICs linked but that also has other internal devices.
@keyser said in Power consumption of the 6100 vs 5100:
@Cabledude I don’t think it does matter to be honest. My 6100 measures about 14 watt with two 1Gbe ports active (both using SFPs) and a 512Gb SSD installed. I do not have PowerD enabled.
I hope you don't mind me hammering on about this, but energy consumption is very important to me and if I could save 5W (4100 vs 6100) it would be an important aspect to consider.
You know I think the 4100 is beefy enough for me now, but I am considering the 6100 for only 200 more just in case I need it later. -
Thermal Design Power simply means the power the cooling system must be able to dissipate. So the maximum power it might consume.
So at low loads neither CPU should consume anywhere near that.
Also bare in mind those plug-top style power meters are not that accurate, they are uncalibrated. And that that includes the PSU which will be less efficient at low loads.
-
@stephenw10 said in Power consumption of the 6100 vs 5100:
Thermal Design Power simply means the power the cooling system must be able to dissipate. So the maximum power it might consume.
So at low loads neither CPU should consume anywhere near that.
Okay I see. I just looked up TDP for the Celeron J4125 (my Synology DS720+ NAS) is 10W, but at idle that unit, including two Samsung 250GB SSD’s, consumes 6W total.
Also bare in mind those plug-top style power meters are not that accurate, they are calibrated. And that that includes the PSU which will be less efficient at low loads.
Yes good point. Still, measuring my buddy’s 4100 vs. my (future) 6100 using the same meter should give a pretty good indication.
-
@keyser said in Power consumption of the 6100 vs 5100:
How about getting a SG-2100? It’s the same as 1100, but has 4Gb RAM and an additional NIC with allows a good deal more peak throughput (about 600’ish Mbps).
So it will handle the 500Mbps if you do not enable NTOPNG. If you enable that, expect the throughput to drop to 250-300Mbps peak.
Most importantly: power consumption is very very low,Dear @keyser ,
Have you ever measured the power consumption of your SG-2100? And do you have the Base (eMMC) or the Max (NVMe SSD) version?Thanks so much!
