GTX1080ti's: Better Options?

[chrysophylax]

Which one for an 8 rig machine?

MSI GeForce GTX 1080 Ti FE DirectX 12 GTX 1080 Ti Founders Edition 11GB 352-Bit GDDR5X PCI Express 3.0 x16

Or

GIGABYTE GeForce GTX 1080 Ti DirectX 12 GV-N108TGAMING OC-11GD 11GB 352-Bit GDDR5X PCI

8 gpu? ...

we build 5gpu max - aorus 1080ti extreme ...

even though it has the capacity for 6 - we have always come across problems with it ... though havent tries lately with the newest drivers ... as for 8gpu - you need a server borad that allows for memory allocation for the 8gpu in the bios from what we have researched ...

#crysx

[VyprBTC]

Which one for an 8 rig machine?

MSI GeForce GTX 1080 Ti FE DirectX 12 GTX 1080 Ti Founders Edition 11GB 352-Bit GDDR5X PCI Express 3.0 x16

Or

GIGABYTE GeForce GTX 1080 Ti DirectX 12 GV-N108TGAMING OC-11GD 11GB 352-Bit GDDR5X PCI

8 gpu? ...

we build 5gpu max - aorus 1080ti extreme ...

even though it has the capacity for 6 - we have always come across problems with it ... though havent tries lately with the newest drivers ... as for 8gpu - you need a server borad that allows for memory allocation for the 8gpu in the bios from what we have researched ...

#crysx

u have pix of 5 gpu aurus builds? definitely intersted! sounds awesome! I'm up to >40 1080Ti's now, have about 10 different cards but not a single aurus - i've been staying away because i'm scared of teh 3 slots it takes up -

I'm pretty much maxed out now but I am hoping to get another few locations to add a "few" more cards. maybe trying to expand too fast but IMO no such thing. go hard or go home

[EvilMonkey]

Which one for an 8 rig machine?

MSI GeForce GTX 1080 Ti FE DirectX 12 GTX 1080 Ti Founders Edition 11GB 352-Bit GDDR5X PCI Express 3.0 x16

Or

GIGABYTE GeForce GTX 1080 Ti DirectX 12 GV-N108TGAMING OC-11GD 11GB 352-Bit GDDR5X PCI

8 gpu? ...

we build 5gpu max - aorus 1080ti extreme ...

even though it has the capacity for 6 - we have always come across problems with it ... though havent tries lately with the newest drivers ... as for 8gpu - you need a server borad that allows for memory allocation for the 8gpu in the bios from what we have researched ...

#crysx

u have pix of 5 gpu aurus builds? definitely intersted! sounds awesome! I'm up to >40 1080Ti's now, have about 10 different cards but not a single aurus - i've been staying away because i'm scared of teh 3 slots it takes up -

I'm pretty much maxed out now but I am hoping to get another few locations to add a "few" more cards. maybe trying to expand too fast but IMO no such thing. go hard or go home

I just got back into mining after ages away and must say it is pretty fun when first starting out and I was initiatlly disapointed in the 1080ti aorus until I found threads like this one and I am definately thinking about getting some more... but 40 .. how do you have all that setup?  That is like almost 40k worth of gear.

[Vaccinus]

Which one for an 8 rig machine?

MSI GeForce GTX 1080 Ti FE DirectX 12 GTX 1080 Ti Founders Edition 11GB 352-Bit GDDR5X PCI Express 3.0 x16

Or

GIGABYTE GeForce GTX 1080 Ti DirectX 12 GV-N108TGAMING OC-11GD 11GB 352-Bit GDDR5X PCI

8 gpu? ...

we build 5gpu max - aorus 1080ti extreme ...

even though it has the capacity for 6 - we have always come across problems with it ... though havent tries lately with the newest drivers ... as for 8gpu - you need a server borad that allows for memory allocation for the 8gpu in the bios from what we have researched ...

#crysx

is the gigabyte 1080ti extreme better than the palit super jet stream? just to know if i did the correct choice by buying the palit, or maybe the giga is now better? how many slot the giga is?

[oup59]

I prefer to keep full system load around/under at most %80 of PSU capacity.

May rig 4 x1080Ti or 3x1080Ti and 1 GTX1080/1070?

Does a 1000 Platinum PSU such as Corsair HX1000i enough for such a setup?

It sounds like this setup will/can exceed 800Watts on the wall.

But if I optimize hash rate and power consumption, would it be OK?

[KaydenC]

I prefer to keep full system load around/under at most %80 of PSU capacity.

May rig 4 x1080Ti or 3x1080Ti and 1 GTX1080/1070?

Does a 1000 Platinum PSU such as Corsair HX1000i enough for such a setup?

It sounds like this setup will/can exceed 800Watts on the wall.

But if I optimize hash rate and power consumption, would it be OK?

Its fine to go to 95% of capacity, just buy a reputable brand like corsair, evga and such.

[philipma1957]

I prefer to keep full system load around/under at most %80 of PSU capacity.

May rig 4 x1080Ti or 3x1080Ti and 1 GTX1080/1070?

Does a 1000 Platinum PSU such as Corsair HX1000i enough for such a setup?

It sounds like this setup will/can exceed 800Watts on the wall.

But if I optimize hash rate and power consumption, would it be OK?

if you run 4 1080ti's at 100& tdp it could be about 1060-1100 watts for whole system
if you run 4 1080ti's at 80% tdp  it could be about 850-900 watts for whole system
if you run 4 1080 ti's at 70% tdp it could be about 750 -800 watts

now some 1080ti's are jacked up to start the aorus is based on 300 watts so:
 4 at 100 % = say 1250-1300 for whole system
 4 at 80%   =  about 1010-1110 for whole system
 4 at 70%  = about    890-940  for whole system
 

[lexele]

I have 1070 aorus and the're very cool under load (first time I started it I thought the ventirad was broken :-D),
 what about the Aorus 1080ti?

[oup59]

if you run 4 1080ti's at 100& tdp it could be about 1060-1100 watts for whole system
if you run 4 1080ti's at 80% tdp  it could be about 850-900 watts for whole system
if you run 4 1080 ti's at 70% tdp it could be about 750 -800 watts

now some 1080ti's are jacked up to start the aorus is based on 300 watts so:
 4 at 100 % = say 1250-1300 for whole system
 4 at 80%   =  about 1010-1110 for whole system
 4 at 70%  = about    890-940  for whole system
 

Thank you very much for the breakdown.

Is the decrease in power consumption necessarily proportional to losing hash rate as well?

If yes I may prefer 1080 rather than 1080Ti.

Although I plan to get 3 1080Ti initially, I may add another one when these cards mine and gain some money.

So far 1080Ti looks really good on SKEIN. Also Nvidia cards are good on ZEC.


I was planning to utilize my own gaming PC and buy 2x1080Ti but changed my mind and plan to set up a dedicated Nvidia Rig. For such a rig:

Is a cheap AMD mobo and cpu fine such as Gigabyte GA-970A-UD3 or even maybe a Gigabyte GA-790XTA-UD4 with enough PCIE slots for 3-4 1080Tis?

For sooner ROI I prefer to get cheaper components that will do the job for mobo, cpu and RAM.

[EvilMonkey]

if you run 4 1080ti's at 100& tdp it could be about 1060-1100 watts for whole system
if you run 4 1080ti's at 80% tdp  it could be about 850-900 watts for whole system
if you run 4 1080 ti's at 70% tdp it could be about 750 -800 watts

now some 1080ti's are jacked up to start the aorus is based on 300 watts so:
 4 at 100 % = say 1250-1300 for whole system
 4 at 80%   =  about 1010-1110 for whole system
 4 at 70%  = about    890-940  for whole system
 

Thank you very much for the breakdown.

Is the decrease in power consumption necessarily proportional to losing hash rate as well?

If yes I may prefer 1080 rather than 1080Ti.

Although I plan to get 3 1080Ti initially, I may add another one when these cards mine and gain some money.

So far 1080Ti looks really good on SKEIN. Also Nvidia cards are good on ZEC.


I was planning to utilize my own gaming PC and buy 2x1080Ti but changed my mind and plan to set up a dedicated Nvidia Rig. For such a rig:

Is a cheap AMD mobo and cpu fine such as Gigabyte GA-970A-UD3 or even maybe a Gigabyte GA-790XTA-UD4 with enough PCIE slots for 3-4 1080Tis?

For sooner ROI I prefer to get cheaper components that will do the job for mobo, cpu and RAM.

The 1080 Ti is very effiecient as far as power per Hashing power in Skein.  For example at 55% TDP my aorus is most effienceint at 55% but only like very very slightly.  All the way upto 75% it was similar effieciency. It started to drop off at around 80% and dropped off even more at 100% stock.

At 55% I was getting 821 MH/s,  At 80% I am getting 947 MH/s and at 100% I am getting 1009.

The thing is even with the Efficiency better at a lower rate it was ALWAYS better to run it at the highest Hash rate because the return is higher because of the price of power and the return you are getting.  So you need to do some testing and put your numbers into a calcuater.  I run a 90% to keep the heat down.

[oup59]

The 1080 Ti is very effiecient as far as power per Hashing power in Skein.  For example at 55% TDP my aorus is most effienceint at 55% but only like very very slightly.  All the way upto 75% it was similar effieciency. It started to drop off at around 80% and dropped off even more at 100% stock.

At 55% I was getting 821 MH/s,  At 80% I am getting 947 MH/s and at 100% I am getting 1009.

The thing is even with the Efficiency better at a lower rate it was ALWAYS better to run it at the highest Hash rate because the return is higher because of the price of power and the return you are getting.  So you need to do some testing and put your numbers into a calcuater.  I run a 90% to keep the heat down.

Appreciate the feedback.

What is wall reading on %55TDP of 821Mh/s and on %80 for 947 Mh/s?

Does it directly/approximately correlate as with default TDP (%100 and 250W) to reduced TDP by X percent so wall is 250*X Watts.

Another thing that got my  attention is that, maybe a coincidence, Auros 1080Ti is preferred by many for 1080Ti mining as far I seen in the threads and post.
I may give it a shot as well.

[Storx]

The 1080 Ti is very effiecient as far as power per Hashing power in Skein.  For example at 55% TDP my aorus is most effienceint at 55% but only like very very slightly.  All the way upto 75% it was similar effieciency. It started to drop off at around 80% and dropped off even more at 100% stock.

At 55% I was getting 821 MH/s,  At 80% I am getting 947 MH/s and at 100% I am getting 1009.

The thing is even with the Efficiency better at a lower rate it was ALWAYS better to run it at the highest Hash rate because the return is higher because of the price of power and the return you are getting.  So you need to do some testing and put your numbers into a calcuater.  I run a 90% to keep the heat down.

Appreciate the feedback.

What is wall reading on %55TDP of 821Mh/s and on %80 for 947 Mh/s?

Does it directly/approximately correlate as with default TDP (%100 and 250W) to reduced TDP by X percent so wall is 250*X Watts.

Another thing that got my  attention is that, maybe a coincidence, Auros 1080Ti is preferred by many for 1080Ti mining as far I seen in the threads and post.
I may give it a shot as well.

Actually its a multitude of things that go into, especially when you have multi gpu setups.. for example
My 2 x 1080ti rig is running on the stock oem fans.. with settings at 70% tdp, when you look at the multiple cards each one will be running at different rates more than likely.. for example...

GPU0 TDP70%, actual usage TDP63%, fanspeed 68%(Auto) to maintain 69c temp limit. card mines at 69c
GPU1 TDP70%, actual usage TDP70%, fanspeed 51%(Auto) to maintain 69c temp limit. card mines at 58c

If you realise the program is not forcing the GPU to 70%, what it is doing is checking all of the settings you changed, it goes down the list if ANY of the numbers = (Equal) or are greater than the numbers you set, it performs internal throttling. So in the example above my GPU0, because temp limit is hit, it pulls voltage automatic tell temp limit is reached along with raising fan speeds.. it will go back and forth and finely increase the both tell the limit is hit again and then maintains at that point it seems..

So if i go under GPU0 and manual change ONLY that card to 79c limit. because the gpu is capable to maintain temperature of 77/78c at TDP70%, it will gradually raise the actual wattage used and stop when TDP is reached.. in my case the card temp sits at 77/78c, which i really dont care for.. so i force it to stay below 69c by changing the limit to 69c... which brings the usage back down to 63% TDP... the reason the first card is suffering is because i currently have them sitting next to each other in a mid atx case, which is causing the exhaust air of GPU1 to mostly be intaked on GPU0, so GPU0 suffers because its being cooled by hotter air...

One thing to note is once you set fan speed to manual you remove stage 1 throttling, which is what i explained to you above. The cards i have, have a built in throttle which is the manufactures max card temp, which throttles the core, mem, and voltage back based on a sliding scale... so if stage 2 throttling starts, it automatically overrides anything stage 1 throttle is performing to attempt to prevent damage to the card.

[oup59]

The 1080 Ti is very effiecient as far as power per Hashing power in Skein.  For example at 55% TDP my aorus is most effienceint at 55% but only like very very slightly.  All the way upto 75% it was similar effieciency. It started to drop off at around 80% and dropped off even more at 100% stock.

At 55% I was getting 821 MH/s,  At 80% I am getting 947 MH/s and at 100% I am getting 1009.

The thing is even with the Efficiency better at a lower rate it was ALWAYS better to run it at the highest Hash rate because the return is higher because of the price of power and the return you are getting.  So you need to do some testing and put your numbers into a calcuater.  I run a 90% to keep the heat down.

Appreciate the feedback.

What is wall reading on %55TDP of 821Mh/s and on %80 for 947 Mh/s?

Does it directly/approximately correlate as with default TDP (%100 and 250W) to reduced TDP by X percent so wall is 250*X Watts.

Another thing that got my  attention is that, maybe a coincidence, Auros 1080Ti is preferred by many for 1080Ti mining as far I seen in the threads and post.
I may give it a shot as well.

Actually its a multitude of things that go into, especially when you have multi gpu setups.. for example
My 2 x 1080ti rig is running on the stock oem fans.. with settings at 70% tdp, when you look at the multiple cards each one will be running at different rates more than likely.. for example...

GPU0 TDP70%, actual usage TDP63%, fanspeed 68%(Auto) to maintain 69c temp limit. card mines at 69c
GPU1 TDP70%, actual usage TDP70%, fanspeed 51%(Auto) to maintain 69c temp limit. card mines at 58c

If you realise the program is not forcing the GPU to 70%, what it is doing is checking all of the settings you changed, it goes down the list if ANY of the numbers = (Equal) or are greater than the numbers you set, it performs internal throttling. So in the example above my GPU0, because temp limit is hit, it pulls voltage automatic tell temp limit is reached along with raising fan speeds.. it will go back and forth and finely increase the both tell the limit is hit again and then maintains at that point it seems..

So if i go under GPU0 and manual change ONLY that card to 79c limit. because the gpu is capable to maintain temperature of 77/78c at TDP70%, it will gradually raise the actual wattage used and stop when TDP is reached.. in my case the card temp sits at 77/78c, which i really dont care for.. so i force it to stay below 69c by changing the limit to 69c... which brings the usage back down to 63% TDP... the reason the first card is suffering is because i currently have them sitting next to each other in a mid atx case, which is causing the exhaust air of GPU1 to mostly be intaked on GPU0, so GPU0 suffers because its being cooled by hotter air...

One thing to note is once you set fan speed to manual you remove stage 1 throttling, which is what i explained to you above. The cards i have, have a built in throttle which is the manufactures max card temp, which throttles the core, mem, and voltage back based on a sliding scale... so if stage 2 throttling starts, it automatically overrides anything stage 1 throttle is performing to attempt to prevent damage to the card.

Thank you very much. This is really informative.

[EvilMonkey]

The 1080 Ti is very effiecient as far as power per Hashing power in Skein.  For example at 55% TDP my aorus is most effienceint at 55% but only like very very slightly.  All the way upto 75% it was similar effieciency. It started to drop off at around 80% and dropped off even more at 100% stock.

At 55% I was getting 821 MH/s,  At 80% I am getting 947 MH/s and at 100% I am getting 1009.

The thing is even with the Efficiency better at a lower rate it was ALWAYS better to run it at the highest Hash rate because the return is higher because of the price of power and the return you are getting.  So you need to do some testing and put your numbers into a calcuater.  I run a 90% to keep the heat down.

Appreciate the feedback.

What is wall reading on %55TDP of 821Mh/s and on %80 for 947 Mh/s?

Does it directly/approximately correlate as with default TDP (%100 and 250W) to reduced TDP by X percent so wall is 250*X Watts.

Another thing that got my  attention is that, maybe a coincidence, Auros 1080Ti is preferred by many for 1080Ti mining as far I seen in the threads and post.
I may give it a shot as well.

I have an auros and I am just using one card.  I find that is better to crank the fan speed than lower the TPD but some people will have to do both depending on their setup.  I do have the numbers for you but it includes say maybe 100 Watts for 2xLCD and cpu hdds etc. I could work out the actual power for just the card but I don't really have a use for that figure.

So at 100% stock but fan at 90% It is using 350 Watt.  At 50% 35% fan speed it is using 200W.  At 70 with 65% fan it is using 260 W.  So it is all relative.  I set my fan speed to keep the GPU temp under 60C and I set the limit at 65C so if it goes over that it will throttle it down a bit more. But so far all those settings it stays under 60.

[EvilMonkey]

Ok to everyone using zpool.  My hash rate is super stable @ 980-1030.  So I set it going for 12 or 24 hours and calculate based off a conservative 950MH/s.  Most of the time it comes back pretty spot on.  Other times it is like 2 or three hours just disappeared randomly.

Is  this some kind of lag or something?

Sometimes it seems fine other times not.   WHen its not there is usually big chunks in the graph like this.



I Have literally watched it mining for hours on end while I game or TV or whatever and never seen it drop below 1000 if I am not touching it.  Is this just a bug on zpool and it will catch up? because sometimes it seems like there is more in there than I could have possibly mined.  I know some of the reasons why this may happen but there is a real shortage of info on how it all works.   Anyone have some good links?

I am going to re-read this thread as I know someone  posted some good info on here but I got a bit confused by it really.

[oup59]

I have an auros and I am just using one card.  I find that is better to crank the fan speed than lower the TPD but some people will have to do both depending on their setup.  I do have the numbers for you but it includes say maybe 100 Watts for 2xLCD and cpu hdds etc. I could work out the actual power for just the card but I don't really have a use for that figure.

So at 100% stock but fan at 90% It is using 350 Watt.  At 50% 35% fan speed it is using 200W.  At 70 with 65% fan it is using 260 W.  So it is all relative.  I set my fan speed to keep the GPU temp under 60C and I set the limit at 65C so if it goes over that it will throttle it down a bit more. But so far all those settings it stays under 60.

Thank you ver much for the numbers? Can you give hashrates for these wattage values as well?

[KaydenC]

Ok to everyone using zpool.  My hash rate is super stable @ 980-1030.  So I set it going for 12 or 24 hours and calculate based off a conservative 950MH/s.  Most of the time it comes back pretty spot on.  Other times it is like 2 or three hours just disappeared randomly.

Is  this some kind of lag or something?

Sometimes it seems fine other times not.   WHen its not there is usually big chunks in the graph like this.



I Have literally watched it mining for hours on end while I game or TV or whatever and never seen it drop below 1000 if I am not touching it.  Is this just a bug on zpool and it will catch up? because sometimes it seems like there is more in there than I could have possibly mined.  I know some of the reasons why this may happen but there is a real shortage of info on how it all works.   Anyone have some good links?

I am going to re-read this thread as I know someone  posted some good info on here but I got a bit confused by it really.

Seems like something was down (the pool, your internet/rig, etc) for that period. I have the issue too, sometimes it just shows stratum connection errors and keeps retrying. I have since switched to miningpoolhub from zpool.

[daemonfox]

Just for reference, I have a 4U 3X EVGA 1080 TI FE rig, OCd and tuned, pumping over 2200 sols @ 885 watts on the kill a watt meter.

Stable and hashing away for a week now. Evan dumped it on ZEN for a few days and snagged some coin.

My temps arent ideal... my garage is NOT a decent place to host in the summer... but I have a COLO rack opening in a few days and that will stabilize my temps.

885 with overhead isn't too bad IMO... its on an i7 6700K with a water cooler and extra drive space so it will also be acting as a node... so there are a few extra watts being pulled that could be trimmed by low end CPU, cooler and an SSD.

Otherwise, the above rig pulls in at least $25 a day at current rates and I never cash out at calced rates. On ZEN it made $60 a day easy till yesterday.

[EvilMonkey]

Ok to everyone using zpool.  My hash rate is super stable @ 980-1030.  So I set it going for 12 or 24 hours and calculate based off a conservative 950MH/s.  Most of the time it comes back pretty spot on.  Other times it is like 2 or three hours just disappeared randomly.

Is  this some kind of lag or something?

Sometimes it seems fine other times not.   WHen its not there is usually big chunks in the graph like this.



I Have literally watched it mining for hours on end while I game or TV or whatever and never seen it drop below 1000 if I am not touching it.  Is this just a bug on zpool and it will catch up? because sometimes it seems like there is more in there than I could have possibly mined.  I know some of the reasons why this may happen but there is a real shortage of info on how it all works.   Anyone have some good links?

I am going to re-read this thread as I know someone  posted some good info on here but I got a bit confused by it really.

Seems like something was down (the pool, your internet/rig, etc) for that period. I have the issue too, sometimes it just shows stratum connection errors and keeps retrying. I have since switched to miningpoolhub from zpool.

Ok thanks.

Nothing on my end has gone down I have things monitoring that and other setups that havn't missed a beat.  I couldn't go back far enough in the buffer to see if there was errors showing I will leave it going and monitor it and try and see if errors come up.  I lost another 1 hour last night on the graph.  It stops for an hour then keeps going.

[Storx]

Ok to everyone using zpool.  My hash rate is super stable @ 980-1030.  So I set it going for 12 or 24 hours and calculate based off a conservative 950MH/s.  Most of the time it comes back pretty spot on.  Other times it is like 2 or three hours just disappeared randomly.

Is  this some kind of lag or something?

Sometimes it seems fine other times not.   WHen its not there is usually big chunks in the graph like this.



I Have literally watched it mining for hours on end while I game or TV or whatever and never seen it drop below 1000 if I am not touching it.  Is this just a bug on zpool and it will catch up? because sometimes it seems like there is more in there than I could have possibly mined.  I know some of the reasons why this may happen but there is a real shortage of info on how it all works.   Anyone have some good links?

I am going to re-read this thread as I know someone  posted some good info on here but I got a bit confused by it really.

Seems like something was down (the pool, your internet/rig, etc) for that period. I have the issue too, sometimes it just shows stratum connection errors and keeps retrying. I have since switched to miningpoolhub from zpool.

Ok thanks.

Nothing on my end has gone down I have things monitoring that and other setups that havn't missed a beat.  I couldn't go back far enough in the buffer to see if there was errors showing I will leave it going and monitor it and try and see if errors come up.  I lost another 1 hour last night on the graph.  It stops for an hour then keeps going.

From the twitter messages, it seems Zpool was targeted by DDOS yesterday for a period.. thus the report servers were down.. but the hashrates mined were still maintained via the Cloud server setup they have in place as backup.. it just doesnt have the reporting side on the cloud servers....