[GUIDE] Undervolt antminer s1 [1.19W/GH at the wall]

[Syke]

Nice mod. Undervolting has a far greater effect than underclocking does on profitability. I'll be looking at doing some undervolting in a few months when summer rolls around.

[1714986028]

1714986028

[1714986028]

1714986028

[1714986028]

1714986028

[1714986028]

1714986028

[1714986028]

1714986028

[1714986028]

1714986028

[crazyates]

In order to grow from 4.9Th to 8.5Th you would have to go from 24 antminers to 45 units if all run at ~200GH overclocked.
If underclocking to ~150GH, you are looking at 57 miners. 

that means paying >32 BTC in order to add 3.6TH  (or 12BTC/TH)

I need 57 miners to reach the 8.5 Th/s , as underclocked.

I already hold 24 miners, so I need 33 pcs more. Which you agree on this.

33 pcs , as underclocked will hash approximately at 5 Th/s ( 33 x 150 = 4,950 Gh/s).

33 pcs normaly cost 33x0.938 = 31 BTC.  So, 31BTC / 5 Th/s = 6.2BTC per Th/s.
Which is lower than any preorder S2 (6.37BTC PREORDER CLOSED)

You have 4 potential mining setups being discussed:

1) 24 OC units running 4.9TH/s @ ~9.9kW (200GH/s + 410W each)
2) 45 OC units running 9.0TH/s @ ~18.5kW (200GH/s + 410W each)
3) 57 UV units running 8.0TH/s @ ~9.5kW ( 140Gh/s + 166W each)
4) 24 OC + 33UV gets 9.4TH/s @ ~15.3kW (200GH/s + 410W for the OC, and 140GH/s + 166W for the UV)

I think he's talking about moving from setup 1) to setup 3), which would increase your hashrate, keep you at your same power, and cost you 31BTC.

You're talking above moving from setup 1) to setup 4). Yes, 33 UV Ants would get you ~5TH/s, but you're not looking at the loss of hashrate that you'd be getting from undervolting the other 24 units. Moving the original 24 Ants from 200 to 140 loses you ~1440GH/s, while you'd be gaining ~4950GH/s, netting an increase of ~3.5TH/s.

Think of it this way: You have 5 Ants that are running 200GH/s @ 410W. You're getting exactly 1TH/s @ 2050W. If you were to add 2 Ants, and undervolt all 7 of them to 140GH/s @ 166W, you're looking at a combined 980GH/s @ 1160W. You paid 1.88BTC (not including coupons/discounts) to stay at the same hashrate, and drop your power consumption.

At current Coinbase, that 1.88BTC = ~$1050 USD. Moving from 2050W to 1160 saves you 890W, which at $0.14/kW, is ~$90 USD/month. You're paying $1050 up front to save $90/month. Might not be worth it.

This same calculation is the same whether you're buying all your Ants now, or buying some ants to compensate for undervolting your existing farm. If you're looking at buying now, and you have to choose between 5 and 7, you're still ordering 2 extra units to save power.

The only scenario this is beneficial is either A) You're running into massive temp issues, and cant seem to get these things cool, and they're constantly being cooked and destroying themselves (aka you live in a desert), or B) you're up against a power limit, and this is the only way to increase your hashrate. I think Tzortz fits into this category, so it might not be an entirely bad idea.

[tzortz]

In order to grow from 4.9Th to 8.5Th you would have to go from 24 antminers to 45 units if all run at ~200GH overclocked.
If underclocking to ~150GH, you are looking at 57 miners. 

that means paying >32 BTC in order to add 3.6TH  (or 12BTC/TH)

I need 57 miners to reach the 8.5 Th/s , as underclocked.

I already hold 24 miners, so I need 33 pcs more. Which you agree on this.

33 pcs , as underclocked will hash approximately at 5 Th/s ( 33 x 150 = 4,950 Gh/s).

33 pcs normaly cost 33x0.938 = 31 BTC.  So, 31BTC / 5 Th/s = 6.2BTC per Th/s.
Which is lower than any preorder S2 (6.37BTC PREORDER CLOSED)

You have 4 potential mining setups being discussed:

1) 24 OC units running 4.9TH/s @ ~9.9kW (200GH/s + 410W each)
2) 45 OC units running 9.0TH/s @ ~18.5kW (200GH/s + 410W each)
3) 57 UV units running 8.0TH/s @ ~9.5kW ( 140Gh/s + 166W each)
4) 24 OC + 33UV gets 9.4TH/s @ ~15.3kW (200GH/s + 410W for the OC, and 140GH/s + 166W for the UV)

I think he's talking about moving from setup 1) to setup 3), which would increase your hashrate, keep you at your same power, and cost you 31BTC.

You're talking above moving from setup 1) to setup 4). Yes, 33 UV Ants would get you ~5TH/s, but you're not looking at the loss of hashrate that you'd be getting from undervolting the other 24 units. Moving the original 24 Ants from 200 to 140 loses you ~1440GH/s, while you'd be gaining ~4950GH/s, netting an increase of ~3.5TH/s.

Think of it this way: You have 5 Ants that are running 200GH/s @ 410W. You're getting exactly 1TH/s @ 2050W. If you were to add 2 Ants, and undervolt all 7 of them to 140GH/s @ 166W, you're looking at a combined 980GH/s @ 1160W. You paid 1.88BTC (not including coupons/discounts) to stay at the same hashrate, and drop your power consumption.

At current Coinbase, that 1.88BTC = ~$1050 USD. Moving from 2050W to 1160 saves you 890W, which at $0.14/kW, is ~$90 USD/month. You're paying $1050 up front to save $90/month. Might not be worth it.

This same calculation is the same whether you're buying all your Ants now, or buying some ants to compensate for undervolting your existing farm. If you're looking at buying now, and you have to choose between 5 and 7, you're still ordering 2 extra units to save power.

The only scenario this is beneficial is either A) You're running into massive temp issues, and cant seem to get these things cool, and they're constantly being cooked and destroying themselves (aka you live in a desert), or B) you're up against a power limit, and this is the only way to increase your hashrate. I think Tzortz fits into this category, so it might not be an entirely bad idea.

Exactly this!


 

[klondike_bar]

In order to grow from 4.9Th to 8.5Th you would have to go from 24 antminers to 45 units if all run at ~200GH overclocked.
If underclocking to ~150GH, you are looking at 57 miners. 

that means paying >32 BTC in order to add 3.6TH  (or 12BTC/TH)

I need 57 miners to reach the 8.5 Th/s , as underclocked.

I already hold 24 miners, so I need 33 pcs more. Which you agree on this.

33 pcs , as underclocked will hash approximately at 5 Th/s ( 33 x 150 = 4,950 Gh/s).

33 pcs normaly cost 33x0.938 = 31 BTC.  So, 31BTC / 5 Th/s = 6.2BTC per Th/s.
Which is lower than any preorder S2 (6.37BTC PREORDER CLOSED)

You have 4 potential mining setups being discussed:

1) 24 OC units running 4.9TH/s @ ~9.9kW (200GH/s + 410W each)
2) 45 OC units running 9.0TH/s @ ~18.5kW (200GH/s + 410W each)
3) 57 UV units running 8.0TH/s @ ~9.5kW ( 140Gh/s + 166W each)
4) 24 OC + 33UV gets 9.4TH/s @ ~15.3kW (200GH/s + 410W for the OC, and 140GH/s + 166W for the UV)

I think he's talking about moving from setup 1) to setup 3), which would increase your hashrate, keep you at your same power, and cost you 31BTC.

You're talking above moving from setup 1) to setup 4). Yes, 33 UV Ants would get you ~5TH/s, but you're not looking at the loss of hashrate that you'd be getting from undervolting the other 24 units. Moving the original 24 Ants from 200 to 140 loses you ~1440GH/s, while you'd be gaining ~4950GH/s, netting an increase of ~3.5TH/s.

Think of it this way: You have 5 Ants that are running 200GH/s @ 410W. You're getting exactly 1TH/s @ 2050W. If you were to add 2 Ants, and undervolt all 7 of them to 140GH/s @ 166W, you're looking at a combined 980GH/s @ 1160W. You paid 1.88BTC (not including coupons/discounts) to stay at the same hashrate, and drop your power consumption.

At current Coinbase, that 1.88BTC = ~$1050 USD. Moving from 2050W to 1160 saves you 890W, which at $0.14/kW, is ~$90 USD/month. You're paying $1050 up front to save $90/month. Might not be worth it.

This same calculation is the same whether you're buying all your Ants now, or buying some ants to compensate for undervolting your existing farm. If you're looking at buying now, and you have to choose between 5 and 7, you're still ordering 2 extra units to save power.

The only scenario this is beneficial is either A) You're running into massive temp issues, and cant seem to get these things cool, and they're constantly being cooked and destroying themselves (aka you live in a desert), or B) you're up against a power limit, and this is the only way to increase your hashrate. I think Tzortz fits into this category, so it might not be an entirely bad idea.

thanks for going into depth - i guess my brief post originally was not clear enough. As mentioned, you are looking at a lot more actual miners to setup and maintain in order to recoup whatever power savings over a period of about 10 months.

IMO, you are better off running what you can right now, and trying to sell off the antminers for top dollar (you can probably fetch 1-1.5BTC each) and re-investing into dedicated 1TH/<1kW devices.

as it stands
7 miners undervolted = 980GH/1160W for ~6.6BTC
1TH/1kW miner = $3500-4000  (about 6.6BTC at current prices - I assume the cash prices will drop soon to accomodate the lower BTC rate)

you would still be drawing less power, and fewer units overall. However, if power is your limiting factor you should look into opportunities to operate miners in other locations. underclocking is throwing away profits right now. (In a month or two it would make more sense of course)

[crazyates]

you would still be drawing less power, and fewer units overall. However, if power is your limiting factor you should look into opportunities to operate miners in other locations. underclocking is throwing away profits right now. (In a month or two it would make more sense of course)

We're getting ready to UC/UV, not so that we can convert our entire farm over now, but so that we can prep for the future.

My math says that with the current price, it's worth it to OC until the diff reaches somewhere around 150PH/s. After that, it's more profitable to UV.

When that time comes, Ant S1s won't be worth 0.938BTC. They'll be worth a fraction of that. If I bring up that 5->7 Ant scenario again, but this time it only costs me 0.25BTC/Ant to upgrade, it'll only cost me 0.5BTC ($280 USD) to save me $90 USD/month in power. That's worth it to me!

[klondike_bar]

^in a month or two time is a different story. my point was just that switching today would be a bad decision with regards to profitability.

[crazyates]

^in a month or two time is a different story. my point was just that switching today would be a bad decision with regards to profitability.

Sometimes, I forgot that I don't have to argue with people who agree with my standpoint.

[klondike_bar]

^in a month or two time is a different story. my point was just that switching today would be a bad decision with regards to profitability.

Sometimes, I forgot that I don't have to argue with people who agree with my standpoint.

+1.  I thought you were crazy enough to underclock your farm today - doing it in a month from now would be much more reasonable (though you might have better luck selling the hardware to buy 28nm gear at that point in time)

[lucazane]

for information, the mod is stable for 4 days !

[tzortz]

for information, the mod is stable for 4 days !

Well done! Good news!

[claudesdad]

I've got to say, I'd be really interested in knowing what kind of math people are using to try and make the mining thing work.

Looking at the prices of the 1 and 2 TH units that are coming out - and running that thru one of the mining calculators - I'm really at a loss to see where anybody is ever going to make the money back on those units.

Even looking at another example:  the cost of an S1 right now - which is ~ $446.50  ( ~$500 per BTC X .893BTC on Bitmain) - and then plug that into one of the calculators - and what I get at the current difficulty level - is 54 days to pay the machine off.

If I plug in the estimated next difficulty level of 6,085,479,004 - that goes up to 69 days, and even that is not bringing into account subsequent difficulty level increases.

Just throw in 8,085,479,004 (because I don't feel like trying to calculate where the higher difficulty levels will actually come out) - and you're talking 100 days to pay the machine off.

I gotta say I'm beginning to see the viewpoint of all the people who are saying that the mining thing is just a losing game.  The numbers simply don't work.

Granted - all the above numbers are based on the current BTC price.

But even if I bump the BTC price up to $700 - that last calc of 100 days only goes back down to 65 days (all other factors being equal)

[crazyates]

I've got to say, I'd be really interested in knowing what kind of math people are using to try and make the mining thing work.

Looking at the prices of the 1 and 2 TH units that are coming out - and running that thru one of the mining calculators - I'm really at a loss to see where anybody is ever going to make the money back on those units.

Even looking at another example:  the cost of an S1 right now - which is ~ $446.50  ( ~$500 per BTC X .893BTC on Bitmain) - and then plug that into one of the calculators - and what I get at the current difficulty level - is 54 days to pay the machine off.

If I plug in the estimated next difficulty level of 6,085,479,004 - that goes up to 69 days, and even that is not bringing into account subsequent difficulty level increases.

Just throw in 8,085,479,004 (because I don't feel like trying to calculate where the higher difficulty levels will actually come out) - and you're talking 100 days to pay the machine off.

I gotta say I'm beginning to see the viewpoint of all the people who are saying that the mining thing is just a losing game.  The numbers simply don't work.

Granted - all the above numbers are based on the current BTC price.

But even if I bump the BTC price up to $700 - that last calc of 100 days only goes back down to 65 days (all other factors being equal)

https://tradeblock.com/mining/a/e2b6bd9a70

Try playing with different numbers on there.

[GenTarkin]

meanwhile, I did the opposite mod and am pushing near 220GH out of some of my ants ... =) hope they dont die! ... been a week or so ... still goin strong.

[tzortz]

meanwhile, I did the opposite mod and am pushing near 220GH out of some of my ants ... =) hope they dont die! ... been a week or so ... still goin strong.

Holly!

We need some explanation here!

[klondike_bar]

meanwhile, I did the opposite mod and am pushing near 220GH out of some of my ants ... =) hope they dont die! ... been a week or so ... still goin strong.

thats very good - i did not expect them to be capable of much more than 205GH even with pencil mod. did you have to add a second fan or shrouding at all?

[lucazane]

meanwhile, I did the opposite mod and am pushing near 220GH out of some of my ants ... =) hope they dont die! ... been a week or so ... still goin strong.

Holly!

We need some explanation here!

The same as I did but with the resistor R12 (the one at the left of R3).

It will increase the voltage at the chip, so be carefull

[GenTarkin]

meanwhile, I did the opposite mod and am pushing near 220GH out of some of my ants ... =) hope they dont die! ... been a week or so ... still goin strong.

thats very good - i did not expect them to be capable of much more than 205GH even with pencil mod. did you have to add a second fan or shrouding at all?

Yeah, I just vmodded one of my '14 units to around 1.15-1.16v per 'module'
Its doing 431mhz @ .55% HW @ 219.5GH for the last 4hrs ... =)
(previously the same unit has been at 425mhz @ .18% HW @ 216GH for a few days...)

I may try it on my other 2 '14 units tonight.

My '13 units dont vmod as well =/

NOTE: no, didnt modify cooling and this particular ant is running at an avg of 44C

[jjiimm_64]

awsome mod.

I applied it and set the clock at 325.   

165 Gh pulling 277watts at the wall

normally 190gh pulls 380watts

[jelin1984]

We need only the pencil?

[Raptor2213]

Good to know.  This mod makes the S1s theoretically cheaper per Ghash than the S2 without sacrificing much power consumption.