I Bought 150+ GPUs for Ethereum mining. Think I'm Insane?

[evenotto]

Very powerful farm, really quite insane. The price you turned up is very good....
I have a question - for mining, do you rent a room or do you keep it in your basement \ house?
Just interesting about amount of costs.

[pieppiep]

I only could say wow, you are ready to get big rewards With all of your hardware, you can get a big hashrate, and if you are mining small coins which can increase later, I am sure that you can double or even triple your profit when the next bull market comes. But I wonder why you don't buy RX 570 or RX 580? Is still too expensive right now? Because I heard that some mid-end GPU hardware was down because of the miners selling their hardware and that makes the price of each series get down too for the price.

[sxemini]

you are making 9$ per month per Card after electricity (at best), and that's considering you don't have to pay taxes or you evade them.

I guess i will never understand why people do this kind of frikin investments.

Because I can spend say 15G and invest in Ethereum at the current market price of $260(June 19') and get 57 Ethereum. Or I can put 15K into mining GPUs and earn 16 ETH a month or about 200 in a year. What makes more sense? yes there more involvement with mining, but it takes risks to make money. With what I'm putting out for investment is minimal compared to the returns for cryptomining. Plus I'm just an IT/Tech enthusiast also work as IT for a living. It all goes hand in hand.

Right now I'm trouble shooting the RX 470's. I got one rig of P104-100s up and running hashing away fairly easy.

I cannot get the newest Claymore 14.6 STRAP function to work without crashing. Unsure what's causing it. Now there is a bit of a caveat Since these were used some cards were BIOs modded others left stock. I put whatever was BIOs modded back to stock from other GPUs that were stock, however, I noticed in windows that quite a few GPUs have the 'ERROR 43' in windows which can only be fixed by applying the ATIpixel patcher.

Setup:

13x Rx470 4GB (Stock BIOs)(Hynix, Samsung, Elpida)

Asrock H110 Pro BTC+

G3930/4GB RAM/EVGA 450w PSU (for motherboard)

Delta 2400w Server PSU (For GPUs)

Windows 10 (1903)

AMD 19.5.1 Drivers



Now when I tested the STRAP function with a single card, works fine. Will launch and I'm seeing good hash rates of 31mh/s~. However, when I plug in all 13x 470's try the STRAP feature, it will get to 2nd GPU applying STRAPs then Windows will crash and restart itself. This happens every time without fail. Now if I only apply it GPU 0 the STRAP then works fine again, however applying it to GPU 0 and GPU 1 it will hard crash and restart windows. Trying GPU 0 and GPU 2 results in the same thing. Also, note that RXBOOST works perfectly fine. The cards will hash fine with stock BIOs or even Modded BIOs using RXBOOST.



So Far I've tried

Raising voltages for Core/Memory or just leaving it stock. Doesn't matter.

Updating Microsoft .net framework and C++
Flashing various Stock BIOs(found online) and even Modded BIOs.

Different Windows versions 1809 vs 1903.



Furtherest I've gotten was 4-GPUs using the STRAP function. Once I started adding the 5th and 6th or more it would crash and restart windows. My Assumption is that it has to do something with the BIOs and the ATIpixel patcher? However, I'm unsure how to remedy this since only 1 or 2 GPUs seem to not have the error 43 in device manager. I'm unsure of this and wondering if anyone has encountered the same problem.

LOL you bought over 150 cards and can not handle it? If a AMD card has a Modified bios, windows give you the error 43. Apply Pixelpatcher and reboot. If you put another card in this rig, then you must apply the patcher again.
 I Think you have bought old mining cards and some has modified bios and other has original bios.
Sorry bro i found your thread interesting, but now. These are basics and all know that for long time. And forget the strap function from claymore, flash bios is the best way.

Windows version does not matter, all work fine

[Rewqpro]

Claymore straps are fine, 32 mh/s for my Micron babies with Claymore straps is the best I could ever get from them.

I wish all the luck to the topic starter, 10c per kWh doesn't sound attractive though

[oldtminer]

mining on windows is nothing but headache and way too much work, with the number of riggs you are dealing with is a full time job and than some. the down time alone will eat up your profit quickly if i were you i would look into some linux distros like ethos smos.

i am a hobby miner IT guy myself and been testing mmpOS and i really like it it's been super stable for my AMD cards

Best of luck!

about your power cost, is it 10 cents or 1 cent?
most power companies in US charge extra premium after 1000KW or so, is yours a fixed amount?

[babos8383]

now it is very profitable to take AMD rx570 SERIES graphics cards, they come out at $ 145-150 which is very cheap for such a hash rate. concerning nvidia, they are cheaper, too. and it turns out not a bad farm

[Michail9090]

I had a farm of RX570 and RX580 cards, I did not know any problems at all, as soon as I managed to sell them at a better price. but now I look even at this course, they are not yet relevant. better I think to wait

[Xazax310]

you are making 9$ per month per Card after electricity (at best), and that's considering you don't have to pay taxes or you evade them.

I guess i will never understand why people do this kind of frikin investments.

Because I can spend say 15G and invest in Ethereum at the current market price of $260(June 19') and get 57 Ethereum. Or I can put 15K into mining GPUs and earn 16 ETH a month or about 200 in a year. What makes more sense? yes there more involvement with mining, but it takes risks to make money. With what I'm putting out for investment is minimal compared to the returns for cryptomining. Plus I'm just an IT/Tech enthusiast also work as IT for a living. It all goes hand in hand.

Right now I'm trouble shooting the RX 470's. I got one rig of P104-100s up and running hashing away fairly easy.

I cannot get the newest Claymore 14.6 STRAP function to work without crashing. Unsure what's causing it. Now there is a bit of a caveat Since these were used some cards were BIOs modded others left stock. I put whatever was BIOs modded back to stock from other GPUs that were stock, however, I noticed in windows that quite a few GPUs have the 'ERROR 43' in windows which can only be fixed by applying the ATIpixel patcher.

Setup:

13x Rx470 4GB (Stock BIOs)(Hynix, Samsung, Elpida)

Asrock H110 Pro BTC+

G3930/4GB RAM/EVGA 450w PSU (for motherboard)

Delta 2400w Server PSU (For GPUs)

Windows 10 (1903)

AMD 19.5.1 Drivers



Now when I tested the STRAP function with a single card, works fine. Will launch and I'm seeing good hash rates of 31mh/s~. However, when I plug in all 13x 470's try the STRAP feature, it will get to 2nd GPU applying STRAPs then Windows will crash and restart itself. This happens every time without fail. Now if I only apply it GPU 0 the STRAP then works fine again, however applying it to GPU 0 and GPU 1 it will hard crash and restart windows. Trying GPU 0 and GPU 2 results in the same thing. Also, note that RXBOOST works perfectly fine. The cards will hash fine with stock BIOs or even Modded BIOs using RXBOOST.



So Far I've tried

Raising voltages for Core/Memory or just leaving it stock. Doesn't matter.

Updating Microsoft .net framework and C++
Flashing various Stock BIOs(found online) and even Modded BIOs.

Different Windows versions 1809 vs 1903.



Furtherest I've gotten was 4-GPUs using the STRAP function. Once I started adding the 5th and 6th or more it would crash and restart windows. My Assumption is that it has to do something with the BIOs and the ATIpixel patcher? However, I'm unsure how to remedy this since only 1 or 2 GPUs seem to not have the error 43 in device manager. I'm unsure of this and wondering if anyone has encountered the same problem.

LOL you bought over 150 cards and can not handle it? If a AMD card has a Modified bios, windows give you the error 43. Apply Pixelpatcher and reboot. If you put another card in this rig, then you must apply the patcher again.
 I Think you have bought old mining cards and some has modified bios and other has original bios.
Sorry bro i found your thread interesting, but now. These are basics and all know that for long time. And forget the strap function from claymore, flash bios is the best way.

Windows version does not matter, all work fine

I know what to do, I've owned RX580s previously and BIOs modded them myself. Spent hours tweaking. With Claymore's v14.6 I was REALLY hopeful that it would simplify this entire process without the pain of BIOs flashing each card, then tweaking each card for Memory voltage/overclock based on whether they can handle it or not. Looks like I was too hopeful.

I found I can reach 5-GPUs and Claymore STRAPS work fine. Anything after the 6th GPU crashes regardless of anything. I've tried different drivers 18.6.1,19.5.1 and 19.6.1 all same the problem. Tried various C++ and .Net framework same crash.

Looks like I'll just have to BIOs flash them myself.

[Xazax310]

mining on windows is nothing but headache and way too much work, with the number of riggs you are dealing with is a full time job and than some. the down time alone will eat up your profit quickly if i were you i would look into some linux distros like ethos smos.

i am a hobby miner IT guy myself and been testing mmpOS and i really like it it's been super stable for my AMD cards

Best of luck!

about your power cost, is it 10 cents or 1 cent?
most power companies in US charge extra premium after 1000KW or so, is yours a fixed amount?

It's a fixed amount but PA is weird They charge a 'supplier rate' and 'Distribution rate' So while my Supplier rate is .04599c/Kw hr the distribution is what it costs to get that power to my house. So my total is .10C/kw. There is no Seasonal rate or anything.

Claymore straps are fine, 32 mh/s for my Micron babies with Claymore straps is the best I could ever get from them.

I wish all the luck to the topic starter, 10c per kWh doesn't sound attractive though

How many GPUs do you have per rig using Claymore straps? I can't get more than 5 to work. No idea why.

[Rewqpro]

mining on windows is nothing but headache and way too much work, with the number of riggs you are dealing with is a full time job and than some. the down time alone will eat up your profit quickly if i were you i would look into some linux distros like ethos smos.

i am a hobby miner IT guy myself and been testing mmpOS and i really like it it's been super stable for my AMD cards

Best of luck!

about your power cost, is it 10 cents or 1 cent?
most power companies in US charge extra premium after 1000KW or so, is yours a fixed amount?

It's a fixed amount but PA is weird They charge a 'supplier rate' and 'Distribution rate' So while my Supplier rate is .04599c/Kw hr the distribution is what it costs to get that power to my house. So my total is .10C/kw. There is no Seasonal rate or anything.

Claymore straps are fine, 32 mh/s for my Micron babies with Claymore straps is the best I could ever get from them.

I wish all the luck to the topic starter, 10c per kWh doesn't sound attractive though

How many GPUs do you have per rig using Claymore straps? I can't get more than 5 to work. No idea why.

I am really sorry to disappoint you now, but I am running 12 AMD GPU rig almost with no problems. Uptime is around 6 days. Using Windows (as it's free and straps are working; once strap function is available in HiveOS I might switch to it).

My setup is a little different though; MOBO is Gigabyte's

upd: I re-checked your first post and haven't found anything about your MOBO lol. I had a feeling you use different one or mentioned it in some other post.. w/e

my piece of advice is to distinguish each GPU's bios and put GPUs with the same bios in the same rig.

This would help you to flash them much easier.

[Xazax310]

mining on windows is nothing but headache and way too much work, with the number of riggs you are dealing with is a full time job and than some. the down time alone will eat up your profit quickly if i were you i would look into some linux distros like ethos smos.

i am a hobby miner IT guy myself and been testing mmpOS and i really like it it's been super stable for my AMD cards

Best of luck!

about your power cost, is it 10 cents or 1 cent?
most power companies in US charge extra premium after 1000KW or so, is yours a fixed amount?

It's a fixed amount but PA is weird They charge a 'supplier rate' and 'Distribution rate' So while my Supplier rate is .04599c/Kw hr the distribution is what it costs to get that power to my house. So my total is .10C/kw. There is no Seasonal rate or anything.

Claymore straps are fine, 32 mh/s for my Micron babies with Claymore straps is the best I could ever get from them.

I wish all the luck to the topic starter, 10c per kWh doesn't sound attractive though

How many GPUs do you have per rig using Claymore straps? I can't get more than 5 to work. No idea why.

I am really sorry to disappoint you now, but I am running 12 AMD GPU rig almost with no problems. Uptime is around 6 days. Using Windows (as it's free and straps are working; once strap function is available in HiveOS I might switch to it).

My setup is a little different though; MOBO is Gigabyte's

upd: I re-checked your first post and haven't found anything about your MOBO lol. I had a feeling you use different one or mentioned it in some other post.. w/e

my piece of advice is to distinguish each GPU's bios and put GPUs with the same bios in the same rig.

This would help you to flash them much easier.

The motherboard I have is an Asrock H110 PRO BTC+
Of course, I've done this. Each rig is separated by memory type. Samsung/Hynix/Elpida. Strange that you don't have this issue. Wonder if it's something to do with the motherboard then?

[Marvell2]

Should bout gpus for grin/beam lo

[Rewqpro]

mining on windows is nothing but headache and way too much work, with the number of riggs you are dealing with is a full time job and than some. the down time alone will eat up your profit quickly if i were you i would look into some linux distros like ethos smos.

i am a hobby miner IT guy myself and been testing mmpOS and i really like it it's been super stable for my AMD cards

Best of luck!

about your power cost, is it 10 cents or 1 cent?
most power companies in US charge extra premium after 1000KW or so, is yours a fixed amount?

It's a fixed amount but PA is weird They charge a 'supplier rate' and 'Distribution rate' So while my Supplier rate is .04599c/Kw hr the distribution is what it costs to get that power to my house. So my total is .10C/kw. There is no Seasonal rate or anything.

Claymore straps are fine, 32 mh/s for my Micron babies with Claymore straps is the best I could ever get from them.

I wish all the luck to the topic starter, 10c per kWh doesn't sound attractive though

How many GPUs do you have per rig using Claymore straps? I can't get more than 5 to work. No idea why.

I am really sorry to disappoint you now, but I am running 12 AMD GPU rig almost with no problems. Uptime is around 6 days. Using Windows (as it's free and straps are working; once strap function is available in HiveOS I might switch to it).

My setup is a little different though; MOBO is Gigabyte's

upd: I re-checked your first post and haven't found anything about your MOBO lol. I had a feeling you use different one or mentioned it in some other post.. w/e

my piece of advice is to distinguish each GPU's bios and put GPUs with the same bios in the same rig.

This would help you to flash them much easier.

The motherboard I have is an Asrock H110 PRO BTC+
Of course, I've done this. Each rig is separated by memory type. Samsung/Hynix/Elpida. Strange that you don't have this issue. Wonder if it's something to do with the motherboard then?

Gosh I hate Asrock stuff

I don't question your flashing skills, but could you share what exactly have you done with AtiFlash?

I faced the problem that the bios of my GPUs was 256kb instead of needed 512kb in order to flash correctly.

So I believe I haven't flashed them at all, and just used Claymore in-built settings.

[Xazax310]

Well, I've played around a lot. Trying to get Claymore STRAPs to work with multiple cards

So I used whatever stock BIOs were on the cards that were shipped. It crashed.
Then I tried stock BIOs from Techpowerup. No difference still crashed.
Right now I'm running modified BIOs using PBE and RXBOOST just to get the rigs running.


I can get up to 5 GPUs working with STRAPs. Soon as I add the 6th and launch claymore with STRAP, bam hard crash right at the "Detecting best straps", and windows reboots. it doesn't matter the 6th GPU either I went one by one and it does the same thing every time. I can get -STRAPs to work if I enable it for only GPU 0 out of 13 then it works normally. If I then apply it to GPU 0,1 same thing. Crashes at "detecting best straps"

After work today I plan on testing some more trying various configurations. It seems to be a Claymore, Windows, AMD Driver problem, unfortunately, I just don't know.

As of right now I have 4 13x ELPIDA RX470's BIOs modded, setup, and running...or crashing. I hate BIOs mod, I'm on fairly conservative settings yet 2 are crashing. 1950mem/1125 900/900 clocks. I used the BIOs mod from here,
https://1stminingrig.com/sapphire-rx-470-nitro-oc-4gb-bios-flash-mod-tutorial-hynix-elpida-samsung-micron/
I noticed 2 rig's won't Hash right with that BIOs from FirstMiningRig they get 24/26~ so that may be on what you're talking about different BIOs sizes. So I used different BIOs for those 2.

Current Hashrate
https://ethermine.org/miners/531E893595875098c10994dF91edF6D886aF6B63/dashboard
 

[dragonmike]

How have you set up your motherboard's bios? Are you using PCIe Gen1/2/Auto?
What AMD driver are you using? Make sure you do anything but install Wattman, that piece of crap has nothing to do on your system.
How many AMD cards per mobo? How good/old are your risers? How strong is your PSU?

[Xazax310]

How have you set up your motherboard's bios? Are you using PCIe Gen1/2/Auto?

The motherboard is setup like so: PCI-E GEN 2 all, Use onboard IGP, POWER-ON, Security settings Disabled,

What AMD driver are you using? Make sure you do anything but install Wattman, that piece of crap has nothing to do on your system.

AMD 19.5.1 also tried 19.6.1. Correct I only installed AMD Driver, I don't use Wattman. I use Claymore's miner to overclock the GPUs.

How many AMD cards per mobo? How good/old are your risers? How strong is your PSU?

13x GPU per motherboard, this is done for Density and space purposes. I've ran 13x MSI RX580's 8GBs previously without any issues. Risers are risers? I've previously used them before in my mining setup which ran for a year +. I run a PSU/Server combo. EVGA 450w PSU to run the motherboard/SSDs etc and a Delta 2400w PSU that runs all the GPUs.

I'm not saying I'm having issues there, everything mines fine using BIOs mod. Just fine tuning on what memory speeds/voltage to get them from crashing takes a little time since I've never had 4GB cards before. I just noticed in my testing(using 1 GPU) that Claymore's straps are far more effective than any BIOs mod for Elpida/Hynix 4GB's. When I BIOs modded one I noticed tons of memory errors around 2000mem clock, using claymore's STRAP feature better hashrate and same clock, but no memory errors. Obviously got me excited.

[abhiseshakana]

Maybe you should try to specify the value for each GPU

eg: 3x GPU used parameter -strap 1@2100,2@2000,3@1900

it means 1st GPU used strap #1 with a memory clock 2100 Mhz, 2nd GPU used strap #2 with MC 2000 Mhz and 3rd GPU used strap #3 with MC 1900 MHz. You can play with strap index and memory clock values for getting the best hashrate and maybe avoid crashes. The higher strap index (strap 1) supports higher MC than strap index 2 (You can learn it from readme.txt in your claymore folder)

[philipma1957]

I think people are offloading their ethereum mining gear before they go pos. Mining gear will flood the market around that time so you'll have trouble selling them afterwards

Pos = death for eth.

Do I think op was smart. No as he has 10 cent power.


But if he pulls it off good for him.

[Xazax310]

I think people are offloading their ethereum mining gear before they go pos. Mining gear will flood the market around that time so you'll have trouble selling them afterwards

Pos = death for eth.

Do I think op was smart. No as he has 10 cent power.


But if he pulls it off good for him.

Thanks. Still hoping to pull it off lol! When Ethereum goes PoS that will be the BIG question considering how many GPUs are mining it. Are those farms going to shut-down? how much GPU-mining power is going to mirgrate. This is things I'm thinking of. But we don't know till it happens. So I'll keep my farm running until then and play a 'wait and see'

Maybe you should try to specify the value for each GPU

eg: 3x GPU used parameter -strap 1@2100,2@2000,3@1900

it means 1st GPU used strap #1 with a memory clock 2100 Mhz, 2nd GPU used strap #2 with MC 2000 Mhz and 3rd GPU used strap #3 with MC 1900 MHz. You can play with strap index and memory clock values for getting the best hashrate and maybe avoid crashes. The higher strap index (strap 1) supports higher MC than strap index 2 (You can learn it from readme.txt in your claymore folder)

Yeah, I've tried everything in the Readme, what happens is as soon as reads the memory type "Applying best Straps for POLE4"(for example) and once it hits the 2nd GPU GPU 0, GPU 1, then bam Computer crashes with a black-screen and reboots itself. This happens when the 6th GPU(or more) is added. 5 GPUs work normally. So far still at a loss of what causes it. ill take a video and post it later today.

[Sebahl]

I actually did the same thing a few weeks ago. Still had all the components for two rigs here - only thing missing were the GPUs. Now I bought 16 more RX470 for around $50-$55 each. But I have free electricity (included in rent), so I am pretty sure I will ROI within a few months.