Building Cheap Miners : My "Secret"

[sundownz]

I have seen similar issues with systems that are powered 24/7 that really aren't meant to be. If the capacitors for the on-board regulators are not of high quality then over time they swell up and sometimes even leak their guts.

I have a Windows 7 desktop computer that runs 24/7 as my home DVR. It is running the Free Windows Media Center. I have had to replace the motherboard once already because of bulging capacitors. Then the replacement motherboard began to have issues. Since I now knew the problem was capacitors I replaced the ones I found bulging which fixed the problem.

Going back to my failing Z400 the reason it failed was when it got powered down the capacitors had pressure reduced because of no power. The constant pressure then no pressure is what caused it to open up and thus lose it's value causing incorrect power on the motherboard.

Because of this issue you may need more than a few Z400 spare motherboards. If a time comes when you have to power off your Z400 miners a number of them may not power up again without issues.

I will keep the post updated -- as of today I have 70 of the Z400 workstations running (8 more will be coming online to finish my current mining plans).

My oldest one is from May 2017 -- so far just the one has this type of issue.

But, I will be sure to update if any more issues pop up in the interest of accuracy / objectivity.

[1713858044]

1713858044

[Neurolicious]

I can get those Dell machines for free. I brought 2 old server cases from my work and 2 intact dell machines.
Modified server cases a bit to have plenty of space for 5 cards. Works like charm and is stack-able.

[MinersRus]

I can get those Dell machines for free.

What model number are those Dell machines?

I brought 2 old server cases from my work and 2 intact dell machines.
Modified server cases a bit to have plenty of space for 5 cards. Works like charm and is stack-able.

[Mohondoa]

Good project for cheaper mining and high hashrate also. I hope to build like yours after i finished with a bounty project ico and i hope to get some the capital from it to start build a rig like that. Thank you for information, and very inspire project.

[sundownz]

I have seen similar issues with systems that are powered 24/7 that really aren't meant to be. If the capacitors for the on-board regulators are not of high quality then over time they swell up and sometimes even leak their guts.

I have a Windows 7 desktop computer that runs 24/7 as my home DVR. It is running the Free Windows Media Center. I have had to replace the motherboard once already because of bulging capacitors. Then the replacement motherboard began to have issues. Since I now knew the problem was capacitors I replaced the ones I found bulging which fixed the problem.

Going back to my failing Z400 the reason it failed was when it got powered down the capacitors had pressure reduced because of no power. The constant pressure then no pressure is what caused it to open up and thus lose it's value causing incorrect power on the motherboard.

Because of this issue you may need more than a few Z400 spare motherboards. If a time comes when you have to power off your Z400 miners a number of them may not power up again without issues.

I will keep the post updated -- as of today I have 70 of the Z400 workstations running (8 more will be coming online to finish my current mining plans).

My oldest one is from May 2017 -- so far just the one has this type of issue.

But, I will be sure to update if any more issues pop up in the interest of accuracy / objectivity.

So... the plot thickens.

I changed the motherboard out -- seemed okay for ~2 days.

BAM -- same exact 4-beep code.

I was a bit frustrated but noted the hard drive was a 13-year old Maxtor... while no errors specifically showed hard drive... HP documents say ANY bad hardware can cause the 4-beep failure.

Changed hard drive to brand-new unit, re-installed all software, and now it's been on since last Thursday.

So... maybe just that old hard drive the whole time!

This is the 3rd refurb that I have replaced the hard drive on... but this first one that caused this type of issue.

[sundownz]

Good project for cheaper mining and high hashrate also. I hope to build like yours after i finished with a bounty project ico and i hope to get some the capital from it to start build a rig like that. Thank you for information, and very inspire project.

Glad that you enjoyed the thread!

[MinersRus]

What are the opinions on Intel's new NUC's with 128MB of eDRAM for mining Cryptonight_V7.

Intel Coffee Lake NUCs will have eDRAM
https://www.fudzilla.com/news/46662-intel-coffee-lake-nucs-will-have-edram

I tried the Gigabyte BRIX before but with the constant throttling because of poor cooling it was not worth it.

The Intel® Core™ i7-8559U Processor is priced at $431: https://ark.intel.com/products/137979/Intel-Core-i7-8559U-Processor-8M-Cache-up-to-4_50-GHz

I believe that price along with only 4 cores (8 with HT) will limit it to only 40 threads (5x mining threads for each of the 8 cores (real and HT)) which only uses 80MB of the 128MB eDRAM will not make it a good product at all for mining.

[sergey1301]

Hello! Looks good!  Thank you for sharing it! But will it work? And will this idea save costs? 5 can I use video cards this way?  But anyway, thank you!!

[sundownz]

Hello! Looks good!  Thank you for sharing it! But will it work? And will this idea save costs? 5 can I use video cards this way?  But anyway, thank you!!

I've been running Z400s since May 2017 -- so, yes, it works well and reliability has been great. Out of 70 machines total I've had 3 hard drive failures as they are refurbs... otherwise no issues with the Z400s themselves.

I have used up to 6 cards with PCI-E splitters in these machines as well -- a few of them running for several months so far.

[sundownz]

I have seen similar issues with systems that are powered 24/7 that really aren't meant to be. If the capacitors for the on-board regulators are not of high quality then over time they swell up and sometimes even leak their guts.

I have a Windows 7 desktop computer that runs 24/7 as my home DVR. It is running the Free Windows Media Center. I have had to replace the motherboard once already because of bulging capacitors. Then the replacement motherboard began to have issues. Since I now knew the problem was capacitors I replaced the ones I found bulging which fixed the problem.

Going back to my failing Z400 the reason it failed was when it got powered down the capacitors had pressure reduced because of no power. The constant pressure then no pressure is what caused it to open up and thus lose it's value causing incorrect power on the motherboard.

Because of this issue you may need more than a few Z400 spare motherboards. If a time comes when you have to power off your Z400 miners a number of them may not power up again without issues.

I will keep the post updated -- as of today I have 70 of the Z400 workstations running (8 more will be coming online to finish my current mining plans).

My oldest one is from May 2017 -- so far just the one has this type of issue.

But, I will be sure to update if any more issues pop up in the interest of accuracy / objectivity.

So... the plot thickens.

I changed the motherboard out -- seemed okay for ~2 days.

BAM -- same exact 4-beep code.

I was a bit frustrated but noted the hard drive was a 13-year old Maxtor... while no errors specifically showed hard drive... HP documents say ANY bad hardware can cause the 4-beep failure.

Changed hard drive to brand-new unit, re-installed all software, and now it's been on since last Thursday.

So... maybe just that old hard drive the whole time!

This is the 3rd refurb that I have replaced the hard drive on... but this first one that caused this type of issue.

It's been basically a full week now after replacing that hard drive -- the system hasn't shut off or had any errors since.

So, I suppose, the motherboard wasn't even bad at all so I will hang onto it as another spare.

[kwchmining]

I have seen similar issues with systems that are powered 24/7 that really aren't meant to be. If the capacitors for the on-board regulators are not of high quality then over time they swell up and sometimes even leak their guts.

I have a Windows 7 desktop computer that runs 24/7 as my home DVR. It is running the Free Windows Media Center. I have had to replace the motherboard once already because of bulging capacitors. Then the replacement motherboard began to have issues. Since I now knew the problem was capacitors I replaced the ones I found bulging which fixed the problem.

Going back to my failing Z400 the reason it failed was when it got powered down the capacitors had pressure reduced because of no power. The constant pressure then no pressure is what caused it to open up and thus lose it's value causing incorrect power on the motherboard.

Because of this issue you may need more than a few Z400 spare motherboards. If a time comes when you have to power off your Z400 miners a number of them may not power up again without issues.

I will keep the post updated -- as of today I have 70 of the Z400 workstations running (8 more will be coming online to finish my current mining plans).

My oldest one is from May 2017 -- so far just the one has this type of issue.

But, I will be sure to update if any more issues pop up in the interest of accuracy / objectivity.

So... the plot thickens.

I changed the motherboard out -- seemed okay for ~2 days.

BAM -- same exact 4-beep code.

I was a bit frustrated but noted the hard drive was a 13-year old Maxtor... while no errors specifically showed hard drive... HP documents say ANY bad hardware can cause the 4-beep failure.

Changed hard drive to brand-new unit, re-installed all software, and now it's been on since last Thursday.

So... maybe just that old hard drive the whole time!

This is the 3rd refurb that I have replaced the hard drive on... but this first one that caused this type of issue.

It's been basically a full week now after replacing that hard drive -- the system hasn't shut off or had any errors since.

So, I suppose, the motherboard wasn't even bad at all so I will hang onto it as another spare.

Typically most of the issues I have run into with PC Errors after a rig has been running stable for some time are due to hard drive's failing. Typically we will just rebuild the rig fresh when this happens from time to time, then it's good for quite awhile (haven't rebuilt any a second time yet that failed from this).

[melpheos]

*** WARNING for all Dell R815 owners ***

DO NOT flash to the latest 3.4.1 BIOS as it will probably brick your system.

I found this out the hard way.

The new 3.4.1 BIOS and other system firmware updates ISO was released 06-12-2018 and available on the DELL site so I downloaded it and burned the ISO onto a DVD and then booted from it. Updates were going smoothly when all of a sudden the monitor screen went blank. I waited for hours but all I had was a blank screen. I forcibly had to power off. When I then powered-on the fans spun up and down but no BIOS screen was ever seen either on the built in VGA port or on the GTX 750 discrete card. The front display status (when holding right button for 5 seconds) states "No Post Code".

The R815 in question had quad Opteron 6238 Engineering Samples (ZS262445TCG45) in it. It was working great on the 3.2.2 BIOS but had older other firmware in it which was my main reason for updating it. The 3.4.1 Bios Update is a Microcode Update (probably for Meltdown/Spector) and I am not sure if having the Engineering Samples in it was why it got bricked.

Needless to say I am not updating any of my other R815's.

FYI: The R815's with quad Opteron 6238 Engineering Samples hash at 1700 H/s. I have three R815 systems with them and two other R815's with quad Opteron 6234's that produce 1570 H/s. In each of these systems I have added a single GTX 750 which adds 230-240 H/s.

If it ain' broke don't try to fix it !

[MinersRus]

*** WARNING for all Dell R815 owners ***

DO NOT flash to the latest 3.4.1 BIOS as it will probably brick your system.

I found this out the hard way.

The new 3.4.1 BIOS and other system firmware updates ISO was released 06-12-2018 and available on the DELL site so I downloaded it and burned the ISO onto a DVD and then booted from it. Updates were going smoothly when all of a sudden the monitor screen went blank. I waited for hours but all I had was a blank screen. I forcibly had to power off. When I then powered-on the fans spun up and down but no BIOS screen was ever seen either on the built in VGA port or on the GTX 750 discrete card. The front display status (when holding right button for 5 seconds) states "No Post Code".

The R815 in question had quad Opteron 6238 Engineering Samples (ZS262445TCG45) in it. It was working great on the 3.2.2 BIOS but had older other firmware in it which was my main reason for updating it. The 3.4.1 Bios Update is a Microcode Update (probably for Meltdown/Spector) and I am not sure if having the Engineering Samples in it was why it got bricked.

Needless to say I am not updating any of my other R815's.

FYI: The R815's with quad Opteron 6238 Engineering Samples hash at 1700 H/s. I have three R815 systems with them and two other R815's with quad Opteron 6234's that produce 1570 H/s. In each of these systems I have added a single GTX 750 which adds 230-240 H/s.

If it ain' broke don't try to fix it !

Having old iDRAC firmware does cause me some issues which was why the update was tried.

Dell released a system wide update that fixes the iDRAC but also does the 3.4.1 BIOS update. That BIOS update is what bricked the system with the ES Opterons.

[grendel25]

Sure, ansible seems to be the way to go and there is a lot of info widely available for that.  It even sounds like ansible can be used to manage windows servers.  

I was more interested in first-hand experience from anyone in this thread.  Like I said above... I am mostly interested in "FLEXIBILITY" to mine whatever I want to mine.  

Sometimes that means mining under Windows and sometimes that means mining under Linux.  Given the equipment I have, it looks like I would be looking to build 4 dual-boot (Linux/windows) systems and then managing them all through ansible.  

If anyone has specific links to info on similar setups, it would be great.  I've seen some configuration guides so far but I'm looking more to see what the comprehensive setup and final product looks like for managing something like this.

For anyone interested in my server-mining virgin perspective... I finally have experience.  I didn't end up using ansible... at all.  Maybe I'm missing out but what I'm doing seems to have worked just fine.  It's easier than I expected.

So my experience went like this:

1. ordered the server (4-node, 128 core Opteron 6376) and set it up on a test bench (my bedroom/office/computer desk table extension).  

2.  From there I installed Win 10 Pro followed by Linux for a dual boot configuration.  I can easily remote into linux and send commands to all four nodes using Xshell.  For windows, any remote access is fine... teamviewer, google remote desktop, Microsoft RDP, etc...  Dual boot configuration is also very easy.  Default bootup goes to the Linux boot grub.  From Linux I can tell it to boot to Windows.  When I reboot out of windows it just goes back to Linux.  All so easy.

3. Tested various algo mining to see power consumption.  Maxed it out and was just over 1300 watts.  For my current infrastructure... this is about as much as I can handle unless I upgrade power.  I didn't get to measure algorithms that used around 50% of CPU resources but I estimate consumption would be around 900 watts for that.  I'm not really happy about the extra power but I am glad to have the extra CPU power as well as learn more about servers.

4. ordered a TP-link 8-port managed switch and power-line Ethernet adapters.  Tested these on various circuits and they all worked.  Home was built in 2004 so fairly new-ish construction/wiring which I've heard helps.  I've also heard folks having problems on linking across circuits but I didn't have this problem.

5.  From there I just moved everything into my mining cave which is now nice and warm.  Really looking forward to the winter!  lol..  Oh and the server CPUs stay nice and cool (as far as my limited experience tells me)   That is, CPU temps around 52C ... But boy does it blow out some heat!  it's right under where my dog sleeps so I bet my dog is enjoying it at night hehe

So yeah, got me some first hand experience and greatly enhanced my CPU-only algo flexibility which was my main goal in all this.

Thank you for this thread.  This thread was the genesis of motivation to arrive at this implementation.  

[MinersRus]

Just a heads up: Ebay has a 10% off coupon good only for today.

Today Only! 10%
Coupon On All Tech

Save with code PRIMOTECH. Valid 7/10
from 5am to 11pm PT. Minimum $50 purchase.

https://pages.ebay.com/promo/2018/0710/67034.html?_trkparms=%26clkid%3D5206987769407708000

I just picked up a Vega 56 using it and saved $43 off the original $430 price.

[lok8nusa]

Just a heads up: Ebay has a 10% off coupon good only for today.

Today Only! 10%
Coupon On All Tech

Save with code PRIMOTECH. Valid 7/10
from 5am to 11pm PT. Minimum $50 purchase.

https://pages.ebay.com/promo/2018/0710/67034.html?_trkparms=%26clkid%3D5206987769407708000

I just picked up a Vega 56 using it and saved $43 off the original $430 price.

Thanx for the heads-up!!!

I just picked up an ASUS Prime Z270 AR mobo for under $100. PC and eBay Bucks saved me >$27. Add a couple of Acorns and some gpus (or FPGAs) and it's time to Mine on!

[MinersRus]

I am garage mining and have the garage door 4" opened at the bottom. My miners are on a workbench at the back of the garage and I have  a fan to push the air from above the miners to a ceiling "whole house" fan that pushes the heat into the attic where it exists from the roof ridge vents.

Here in the Dallas area we are having a heat wave where the temperatures are getting to 111 degrees around 5 PM.

I have four HP DL580 G7's all running on 240V AC. These four DL580 also have 23 GTX 750's running on top or in them.

These four systems produce 11,692 H/s mining Monero.

The DL580 will shutdown at 107 degrees.

One thing I noticed is that with the heat comes higher energy usage. At the 106 degrees the four processor fans in each of the DL580's are running at 100% and they take quite a bit of power to do that as seen by these total system numbers:


106 degrees: 16.7 Amps, 4008 watts

96 degrees: 15.1 Amps, 3624 watts

86 degrees: 14.4 Amps, 3456 watts


The increase in power is mostly from the DL580 fans with some coming from the GTX 750's.

I am only somewhat profitable at 106 degrees:

https://www.cryptocompare.com/mining/calculator/xmr?HashingPower=11692&HashingUnit=H%2Fs&PowerConsumption=4008&CostPerkWh=0.071&MiningPoolFee=1.0

Only $88 in profit while using $205 in electricity isn't much so at this time I have stopped mining in the garage until this heat wave breaks which is forecast around July 24th.

My "whole house" fan has also just decided to stop working and only hum when powering on so I will take this time off and see if it has a bad run capacitor as it is 15 years old. I will also be cleaning all the systems of dust.

[sundownz]

So. My PNY 1080s ran out of warranty last month -- and I had two die on me in the heat.

To my surprise I discovered broken GPUs have some fairly decent value on E-Bay.

I paid $499 each in June 2017 for the cards... sold for $200 each on E-Bay as broken / for parts.

[Kahooli]

If I were you (in texas) I'd set them to all turn down to 50% usage for the 8 peak heat hours, or 0% for the peak 4 hours. Electronics do not want to run that hot. Cap life is inverse to operating temp. Silicon transistor MMTF is log inverse to operating temp

[PharmEcis]

Since this thread is what got me started scaling up I'd figure I'd update as well.  I'm also in TX near Dallas and the heat has been atrocious.  I also suffer from another problem.  I had a fixed rate of 7.9c/kw which I could live with.  My energy provider abruptly exited the market 2 months ago leaving me hanging and with providers of last resort.  After much scrambling I ended up signing up with GoGriddy.  They charge you a $10 monthly fee, direct pass through transmission charges and wholesale electricity cost updated every 5 minutes.  I'm actually saving money... Well I was until this last heat wave hit.  I went from $25/day in juice to $60, consuming about 400kw per day.  I've seen spot rates as high as $3/kw at 4PM when it's 109F out.  So because of this I power down at 1PM and power back up after 6pm.  It's the only choice I have or I'm way upside down on the electricity cost.  The bonus is that outside of these crazy hours, I'm actually paying LESS for electricity than I was because night time hours are under 2c/kw and I pull the same amount of juice at lowest demand as I do at peak.

I've picked up some of the FPGA hardware and am sending it to Mineority for hosting with them.  Their fee is reasonable when you factor in all the additional expenses, not to mention effort, for running rigs.  With a point and click interface I don't know what I'm going to do with myself.  Oh yeah, maybe do something other than constantly babysit machines. 

I've also got some Acorns on the way to test with.  I'm seriously looking forward to cooler weather.  Ambient temp forced cooling sucks when it's hot.  (It's actually nice right now at 86F and a nice breeze inside the cointainer lol)  I will continue to mine and at some point probably expand my farm.  I will not expand until I've built and tested an immersion setup.  If I am to host at my current location going forward, I have to implement that as a method to eliminate a host of issues I have faced by converting a shipping container in TX into a location to house electronics running full tilt.