FW v1.0.32-test2 uses less WATTS

[Ligaman11]

Hello,

After spending a lot of time downgrading and upgrading firmware I have noticed that only v1.0.32-test2 firmware for some reason uses around 10 watts less than any other other firmware, even the most recent one, now keep in mind that it does take longer for the hashrate to settle but non the less it ends up at the same hashrate as any other firmware, Example, NerdQaxe++ Rev-6 is running around 117 watts on any firmware i throw at it, now if i install v1.0.32-test2 firmware it drops my watts to 107 watts on the same overclocks and the hashrate stays the same as running 117 watts, Can someone please let me know if this has been discussed before or why all the new firmware including betas are just magically adding 10 watts to the miner

Thank you

[alucine]

If you know real watts, buy a wall socket with energy meter. In Amazon 10 €, in AliExpress 4€.

By example my nerdqaxe++ with firmware 1.33, in web 88.4w in socket 94w.

Always use wall socket to compare 😀

[Ligaman11]

Let me be a bit more specific, I'm the type of person that never comments anywhere unless if I have done my research or testing, 10 watts are coming from the firmware alone, yes sir I have tested thru the socket and the difference is around 10 watts, 3 out of 4 miners are 10 watts, the 4th miner is around 7 watts difference thru the outlet, my apologies for not stating it beforehand

117 watts showing at the miner, at the sockets 126-128

Test 2 firmware = at the miner 107 watts at the outlet 116-118 watts

If firmware was faulty in calculating then the wall outlet would still register 126-128 watts as I'm using the same over clocks, test it yourself and let me know if I'm wrong,

[kukulle-hood]

@Ligaman11 Thank you for your eagle-eye on comparing old and new firmwares, love it.
I meanwhile also downgraded to the v1.0.32-test2 version to compare it with the latest, brand new v1.0.34-rc1 firmware.

I can confirm your statement, but not to the extent you described.

I a using a NerdQAxe++ (I dont know which revision) running on 750 MHz

New Firmware --» Indicated in the Browser 113,2W --» Measured at the Socket 132,2W
Old Firmware --» Indicated in the Browser 111,8W --» Measured at the Socket 131,1W

All measured after approx. 15 minutes when reaching "full power"

Now I go back again to the new Firmware.

Robert

[Ligaman11]

@kukulle-hood

im assuming your running revision 6 with the yellow xt30 connector, since i don't think you should run any previous Revisions at those overclocks since the old revisions with the barrel connector shouldn't run higher than 96 watts since the port cant handle it, now yes you can run more watts thru that barrel port but your risking a fuse pop,

i just tested again using a different outlet tester and yeah it showing 7-9 watts in difference between firmwares, now im running these miners about 2 hours for the power to settle. NOW since for me firmware is causing more energy to be used at the same hash, can the developers some how trick the miner/pool to report/use less watts, and increase more hash, I dont know anymore, ( still running the v1.0.34-rc1 firmware )

I also forgot to mention that i have been using the meanwell 12v 650 powersupply since day one and running 4 miners per power suppply, keeps me under the 80% rule

[kukulle-hood]

Hello @Ligaman11, Thanx to your feedback.
No, my system has unfortunately no xt30 connector, still have this barrel connector.
I am facing no problem at or in the barrel connector itselve with heat. I see more problems with the cable meanwhile what is gettinng a bit warm, even that I have thicker cross section. At the given little fuse I have already changed from 8 Ampere to 10 Ampere when I was at 100 Watts.
I am also running a meanwell 12V, 150W, now at about 75%

Cheers Robert

[shufps]

Hmm no idea where this should come from 🤷‍♂️

For example 1.0.32-test2 to 1.0.32-test1 is mainly just esp-idf downgrade from 5.5 to 5.3, but the same esp-idf 5.3 is now in the 1.0.34-rc1 🤔

[gruve-p]

I have the same behaviour on 2 NerdQaxe++ Rev 6 with 800/1250
1.0.32-test2 gives me 106W / Asic temp at 60C / VR temp at 61
1.0.32.1 (and 1.0.33 and 1.0.34) gives me 112W / Asic temp at 64C / VR temp at 65

[Ligaman11]

I have it on all 4 rev6 miners, so yeah it's something on the firmware that's tricking us lol

[gruve-p]

I am keeping it at 1.0.32-test2 to give the stock PSU more headroom

[gruve-p]

@Ligaman11 Thank you for your eagle-eye on comparing old and new firmwares, love it. I meanwhile also downgraded to the v1.0.32-test2 version to compare it with the latest, brand new v1.0.34-rc1 firmware.

I can confirm your statement, but not to the extent you described.

The impact is more noticable on the rev 6 (with xt30 connector) than with rev 5 (barrel connector)

[Ligaman11]

It's mainly when it's maxed out in overclocked, 800/1250 but it is kinda weird that's it's close to 10 watts at the outlet

[gruve-p]

It's mainly when it's maxed out in overclocked, 800/1250 but it is kinda weird that's it's close to 10 watts at the outlet

Yeah I run it on 800/1250. Did you try v1.0.34-rc2.1 ?

[Ligaman11]

I will in just a bit, didn't know it came out, thank you again

[shufps]

I have the same behaviour on 2 NerdQaxe++ Rev 6 with 800/1250 1.0.32-test2 gives me 106W / Asic temp at 60C / VR temp at 61 1.0.32.1 (and 1.0.33 and 1.0.34) gives me 112W / Asic temp at 64C / VR temp at 65

It's because in earlier versions the 800MHz were not reachable even when selectable in the UI. So it run always at 750MHz max.

This has been fixed since the 1.0.32-beta1 and now it runs at 50MHz more and consumes more power.

#247

So instead of using an old firmware you could switch the Asic clock to 750MHz and have the same power consumption.