Power Consumption Thread

[Sean Ho]

Doesn't the D4040 still suffer from the AVR54 LPC clock bug?

 

[fohdeesha]

Doesn't the D4040 still suffer from the AVR54 LPC clock bug?

yup would not touch

 

[AveryFreeman]

They have a hollow/tinny whine like the DX010. Not quite as loud as the ICX6110. But you can patch the fan configuration although I haven't done it for SONiC. The problem with using SONiC on the D4040 is the switch was never officially supported. @okrasit shared his deb modules and while it generally worked the LEDs didn't and it was for an old version -- I recall a 2018 snapshot. Newer versions won't load the modules because they were compiled for an older kernel. He indicated he was going to update it but hasn't logged on here for over a year. Maybe he has moved to better platforms; I hope he's doing well.

It looks like there's an official build guide that's been archived here: GitHub - celestica-Inc/sonic-buildimage: Scripts which perform an installable binary image build for SONiC

It has over 1,000 forks, so it seems probable that someone is continuing it. The trick is finding out who.

Usually to figure out who's continuing a project, I'll search for the repository name in "all of github" and sort by "recently updated". Let's see how that turns out... : GitHub - ShoneWu/sonic-buildimage

Looks like ShoneWu updated his repo 14 minutes ago ... README.md says:

A good choice of OS for building SONiC is currently Ubuntu 20.04.

... sounds promising.

 

[AveryFreeman]

Doesn't the D4040 still suffer from the AVR54 LPC clock bug?

Yep, if built before 2018 (or has B0 stepping C2358 CPU)

 

[NablaSquaredG]

SN2700 Rev A2 in IDLE (no ports connected, only interface), one power supply: 51W (no ports connected, only mgmt, after fans ramp down to 8.6k/7.3k RPM)

 

[NablaSquaredG]

SN2100 Rev B1 in IDLE (no ports connected, only console + mgmt), one power supply: 36W (no ports connected, fans ramped down)

SN2100 Rev B1 2 ports linked, one power supply: 42W

 

[NablaSquaredG]

Did some power scaling tests:

240V @ 50Hz, Mellanox MSN2100-BB2F Rev B1, 8x0.5m fs.com QSFP28 DAC im loopback, 1 PSU
Fans spun down to 6500RPM
"according to switch" -> averaged report form show power
NO TRAFFIC on ports, just link up


16x100G, 2 PSUs: 73.4W wall, 61W according to switch
16x40G, 2 PSUs: 63.8W wall, 52W according to switch
16x100G, 1 PSU: 72.8W wall, 61W according to switch
16x40G, 1 PSU: 63.1W wall, 52W according to switch

12x100G, 2 PSUs: 64.1W wall, 53W according to switch
12x40G, 2 PSUs: 57.1W wall, 46W according to switch
12x100G, 1 PSU: 63.4W wall, 53W according to switch
12x40G, 1 PSU: 56.5W wall, 46W according to switch

8x100G, 2 PSUs: 54.5W wall, 43W according to switch
8x40G, 2 PSUs: 50W wall, 38W according to switch
8x100G, 1 PSU: 53.7W wall, 43W according to switch
8x40G, 1 PSU: 49.1W wall, 38W according to switch


4x100G, 2 PSUs: 45.9W wall, 35W according to switch
4x40G, 2 PSUs: 43.8W wall, 33W according to switch
4x100G, 1 PSU: 45.1W wall, 35W according to switch
4x40G, 1 PSU: 43.1W wall, 33W according to switch

No Ports, 2 PSUs: 37W wall, 27W according to switch
No Ports, 1 PSU: 36.1W wall, 27W according to switch

EDIT:
So in total, that gives you a relatively constant 2.25W,per 100G port or 1.5W per 40G port in addition to the 36-37W IDLE power (not including additional power consumption from transceivers or active cables)

 

[NablaSquaredG]

ICX7450-24, 1x ICX7400-4x10GF, 2x ICX7400-1x40GQ, no ports linked
2x RPS15-E, 2x ICX-FAN10-E: 54W
1x RPS15-E, 2x ICX-FAN10-E: 47.5W
1x RPS15-E, 1x ICX-FAN10-E: 45W

ICX7450-24, 2x ICX7400-1x40GQ (10G module removed)
2x RPS15-E, 2x ICX-FAN10-E: 46.5W

ICX7450-24, no modules
2x RPS15-E, 2x ICX-FAN10-E: 44W

 

[NablaSquaredG]

ICX6450-48P @ 240V, no ports linked: 55.2W

 

[NablaSquaredG]

ICX7150-C12P: 17.4W IDLE (240V, no ports linked, 2 samples)
ICX6450-C12-PD: 12.2W IDLE (240V, no ports linked, 2 samples)

I should really create a public google sheet...

 

[NablaSquaredG]

ICX6450-24P: 35.3W IDLE (240V, no ports linked, 7 samples)

So, after testing 7 different samples, I can say that I'm 99% certain that the "25w power draw for the 24-port models with or without PoE" in the Brocade ICX thread https://forums.servethehome.com/ind...s-cheap-powerful-10gbe-40gbe-switching.21107/ is not correct for the PoE models.

 

[fohdeesha]

ICX6450-24P: 35.3W IDLE (240V, no ports linked, 7 samples)

So, after testing 7 different samples, I can say that I'm 99% certain that the "25w power draw for the 24-port models with or without PoE" in the Brocade ICX thread https://forums.servethehome.com/ind...s-cheap-powerful-10gbe-40gbe-switching.21107/ is not correct for the PoE models.

were they all 240v fed? My samples are from 120v power feed through a wattnode revenue meter

edit: granted I only tested two now that I think about it, and this was years ago at this point

 

[NablaSquaredG]

were they all 240v fed? My samples are from 120v power feed through a wattnode revenue meter

edit: granted I only tested two now that I think about it, and this was years ago at this point

yep, all 240V

are you sure your power meter was accurate? Because usually, power supplies are more efficient at 240V than 120V (even if it's just in the low single digit range). There should definitely not be a 10W difference for 120 vs 240V, especially not to the disfavor of 240V

If I remember correctly, the non-P sat at 25W..

 

[fohdeesha]

yep, all 240V

are you sure your power meter was accurate? Because usually, power supplies are more efficient at 240V than 120V (even if it's just in the low single digit range). There should definitely not be a 10W difference for 120 vs 240V, especially not to the disfavor of 240V

If I remember correctly, the non-P sat at 25W..

yeah all those readings were through a WattNode Revenue meter, NIST traceable to +/- 0.5%. it's entirely possible I swapped the poe and non poe around though, I think that was almost 6 years ago now. Ill see if I should update them