[10 capacitors]

For years I have been using 24 x ver008c risers without issues - never had one go bad.  But there's been numerous new versions come out since then and the latest seems to be the ver016 which has two LED displays (voltage and temperature).  It also has 10 capacitors!  I've ordered some to try out but I'm a bit skeptical that 10 capacitors will make any difference whatsoever compared to 4 capacitors.  If anything, there's more chance of failure considering a single shorted capacitor can kill the board.  Sounds like a marketing gimmick like those 10 bladed razors.  More isn't necessarily better.

I also wonder if having a metal shielded slot makes a difference.  Has anyone found improved stability with the newer risers?  Like ver010 and up?

After 5 years of stirling service and a lot of modifications I'm retiring my rig of 8 x 1070's.  My remaining 2 rigs, each with 8 x 1070ti's are running at around break even so will keep them spinning a few more months, at least until the Australian winter is over.

The question now is what to do with them?  The 1070's are old cards now and I can't imagine there is much of a market for them especially considering their mining history.  (Though one could argue they are thoroughly tested!).  I doubt it would be worth the hassle of listing them on ebay.  But I can't yet bring myself to throw away perfectly good cards.  So I will probably store them for 10 years and then throw them away!

What do you do with old mining cards that are no longer profitable and unlikely to ever be profitable again?

...was a classic 70's scifi flick (which most of you probably haven't seen) but that's not what this thread is about 

A few months ago, I decided to replace all the fans in my rigs with noctuas with the goal of reducing the noise level.  72 fans later, I am impressed with the result. 



My cards are Gigabyte 1070s and 1070tis, 3 rigs, 8 per rig.

They now run so quiet to be virtually inaudible unless it's a very quiet room.  In fact the main remaining noise was from the Silverstone 1500ti power supply so I went a step further and replaced that fan with a noctua NF14 industrial, connected to a motherboard header for control and monitoring.  My rigs are now so quiet that I can use them in a bedroom (which is ideal during winter.)  Another advantage is noctuas are rated for 150,000 hours which is like 20 years and laugh at dusty hot environments.  The stock fans began to seize up after 6 months and over the years I have replaced all the original sleeve bearings with ball bearings which stopped them seizing up but didn't help with the noise.

Cooling seems to be slightly improved too as temperatures with the fans running at maximum speed are slightly less than the originals running at their maximum speed.  The only downsides are losing the pretty lights (meh) and the fan calibration monitoring is completely wrong now as the bios thinks 4200rpm is 100% (original fan), so when the noctuas are running at maximum (2200 rpm) monitoring indicates 54%  This makes absolutely no difference to the cooling; it's just I need to do some mental gymnastics to determine the real percent.

And finally, the noctuas use less power.  Around 5 to 7 watts per card which adds up when you multiply it by 24 cards running 24/7.  Around 3kw less per day.  Which adds up over time.

So in summary:

PROS

Very quiet!
Way less prone to failure.
Slightly better cooling.
Slightly less power used.

CONS

Expensive x 24
Displayed fan percent incorrect.
Losing the pretty lights.


The result while costly, means that even when mining barely covers the cost of the electricity, I can happily run my rigs during during winter, using them as heaters.  Each rig draws around 1100 watts which provides the same heat as an 1100 watt heater.  BUT they are actually quieter than an equivalently powered heater!

One of my cards was running hot so I took it apart to see what was going on and discovered a crater maybe 2mm wide and maybe 50-100 microns deep in both the heatsink and the GPU.  It's like some tiny people let off a tiny stick of dynamite and made this not-so-tiny crater.

Card works totally fine except for getting hot.  How to fix?  Sandpaper?

I find it perplexing that so many (especially newbies) don't seem to be concerned about the pending Eth POS.  I believe the following to be facts but if anyone can rationally argue any of these points, I would be very interested to hear it.  It's quite possible I'm the stupid one but so far I haven't heard any reasonable rationale to counter the following logical chain.

1. Eth goes POS
2. Immediately, all the gpus currently mining Eth, switch to mining much smaller coins.
3. The enormous amount of hash power causes an enormous increase in difficulty for these smaller coins, resulting in:
4. Smaller coins mining profitability crashing through the floor, way below the electric cost for most miners.

At the moment, other coins are profitable.  Not as profitable as Eth but nonetheless still can generate more revenue than the electricity costs.  But consider a 100x increase in hash power that will result when erstwhile Eth miners switch over.  100 x hash power = 1/100 revenue per hash  (This is a very rough approximation).  Therefore if a card is currently earning $1 a day in revenue, it will drop to 1 cent a day.  Obviously no profit at all here.

Of course very quickly people will start turning off their rigs when they see they can no longer make money.  After a short period of time (weeks maybe), difficulty will drop and an equilibrium will be reached whereby the few miners with the lowest electric cost and the most efficient cards will be left mining with some degree of profitability.  But the majority of current miners will be out of the game.

This is not my opinion, it is factual.  Unless someone can fault my logic. (Please, please do!)

The Eth POS is not like a bear market.  It's a game changer for gpu mining.  There is no other POW coin that is anywhere near Eth's dominance.  The best you can hope for is that EIP1559 continues to be delayed because when it happens, it will be the end of gpu mining for most people.

Probably the worst time ever to be buying overpriced cards and starting mining.

In an effort to save watts and have fun, I would like to try to make an 8 gpu rig using one of those cpu/motherboard combos that typically have a tdp less than 20 watts.  To run 8 cards, I just need 2 slots feeding 4 way pcie splitter cards.  But I also need the "Above 4g decoding" bios option.  Is anyone aware of any combos that include such?

I've been running 24 Gigabyte GTX 1070/1070ti's for the last 4 years and they're all still running fine.  But one card has always had rather odd characteristics.  It runs around 10 degrees (C) less than the others.  (both gpu and hotspot).  All settings the same but it just runs a lot cooler with the same hashrate.  Sometimes the "fan stop" light comes on as the fans slow down and stop cos it's just not getting hot enough.  And yet it hashes just as good as the rest of them.

I can only assume it's due to some random serendipitous manufacturing "defect" that has made it unusually efficient at dissipating excess heat. Incidentally, makes no difference what position in the rig the card is.



You get some weird shit in electronics sometimes.  Years ago, I was working on a faulty circuit in a module that I finally traced to a copper track that had a micro crack causing the track to exhibit diode like properties.  i.e. It would only pass current in one direction.

Many of the BTC-37 boards dont have an ATX 24 pin connector, instead using the 12 power from the slot power sockets to power the board.  Which is ideal if you have a server power supply.  The question I have is does the T version's existent ATX connector require something plugged in?  Or is it just there so you can easily choose either type of power supply?

Reason being I am considering trying one of these boards in conjunction with a server power supply.  I'm hoping having something plugged in to the ATX connector is optional.  But no amount of googling has found me an answer.

Anyone know?

One of my rigs has 8 1070's that have been running nicely for almost 4 years.  But something I have never understood is that overclocking the memory makes no difference to the hashrate. But increasing the core clock does.  It's precisely the opposite of what is supposed to happen.  My other rigs which have 1070ti's don't exhibit this behavior.

My 1070's are tuned to use around 110 watts hashing at around 28mh/s which seems pretty optimum but I only achieve this by leaving the mem clock alone and bumping up the core clock.  All my rigs hardware are identical excepting the 1070ti's on the other rigs, which I have running at around 110 watts and 32.3mh/s by increasing the mem clock.

I haven't bothered asking this before because I doubt very much there is an answer.  But has anyone experienced this and know what causes it?

Its getting close to 4 years I have been mining with 3 rigs of 8 cards (8 x 1070, 8 x 1070ti, 8 x 1070ti).  They run 24/7 in a smallish room (2.4 x 3 metres), sitting on old carpet in an old flat. In an Australian summer, ambient can go north of 40 degrees Celsius.  Basically, hell on earth for mining rigs.  Fans began to fail after 6 months due to the dust, solved by replacing the sleeve bearings with double ball bearings.  But the biggest problem is heat.  Remarkably, the only issues I have had besides the fans is the pcie splitters melting.

Each rig has a silverstone 1500ti power supply rated to handle 150 watts from each of the 8 pcie outputs.  Each output supplies one card through a pcie splitter, (one for the riser and one for the card)  Each card is limited to around 120 watts so there should be no problem with this arrangement.

However over time I discovered some of the pcie splitter connectors began to overheat to the point of wires melting/breaking.  Some were fine but many were too hot to touch even if they were still working.  The wires were cool, it was just the splitter connector that was failing.  Clearly the dusty hot environment was causing issues.  So I implemented a final solution of cutting off the connectors and soldering them on.  That's 24 cables I had to butcher so it took a while.





That has entirely solved the issue with the bonus of saving around 40 watts of power per rig.  That presumably was how much heat was being wasted in the connector.

Connectors are always going to be a weak spot, the fewer the better and those you have, you should check occasionally.  If they're getting hot, do something about it!

Had an old Gigabyte GTX1070 that worked but had a circuit fault in that it wouldn't run fans.  (Tested fans and wiring, definitely circuit fault).  I could have jury rigged some fans to run off the motherboard but never got around to it.

When I stumbled across this ID-COOLING IceFlow 240 AIO VGA Liquid Cooler, I thought I would give it a shot and I'm pleased to report it works very well indeed.  It is whisper quiet compared to the standard fans and maintains an even temperature roughly 16 degrees (celsius), less than the aircooled standard.  It's designed to fit a wide range of different GPUs, (includes a lot of different brackets).

Uses around 10 watts in total but the biggest negative is the size.  I have 8 GPUs in my rigs and now I have one radiator hanging off it, I can't imagine trying to fit 8 of them.  Pity though, as the absence of noise is blissful.  Anyway, thought I would post this in case anyone else is curious about watercooling their GPU.

I'm thinking given I power limit my cards to around 60 or 70% and the TDP is 225 watts, I don't really need to use both power connectors?  The power connector on the riser supplies 75 watts and one 8 pin connector can supply up to 150 watts so having 2 power connectors might not be strictly necessary especially if I limit power to say 70%

Question is, Will it work?  I mean speaking of Gigabyte cards, are they fixed such they will refuse to run without power supplied to both connectors?  I need to know the answer before I buy this card.  I know there is a version that has just one connector but that's so out of stock and the only one in stock has 2 connectors. 

Has anyone tried running the card with just one power connector, leaving the other one empty?

Ok so my gigabyte 1070's, less than 6 months old are starting to fail.  They are gummed up with dust and one of them seized up altogether.  Checked out http://cryptomining-blog.com/8138-how-to-maintain-and-repair-dual-x-and-other-non-serviceable-gpu-fans  and https://youtu.be/Tf0FxUe30rU and decided to have a go disassembling and replacing the sleeve bearings with ball bearings.

Got the 2x5x2.5mm bearings and shims off ebay and set to work.  Took all up about 5 hours!  Which included going out for some superglue.  The fan mounting brackets are extremely fragile to the point where a stern look is almost enough to make them crack.  Also on one fan I managed to break off the metal spindle attached to the fan so superglue definitely needed!

Apart from being extremely careful the only really hard bit is separating the fan blade section from the stator.  These fans are NOT meant to be disassembled!  You need to apply just enough force in just the right location and pray the bits you want to separate, come apart and nothing else.  There are very small bits of plastic that will break easily, not to mention the small built in circuit board which attaches to the stator coil by 3 current bearing pins.  When applying separating pressure, a lot of stress is put on these pins and you are relying on the quality of the solder joints to hold them.  You want the tiny plastic holding ring to give way before the solder joints do and I can tell you, they are pretty evenly matched!

Removing the fans requires separating the heatsink from the chip so the silver lining to this ordeal is that you hopefully apply better quality heatsink compound.  I discovered the existing compound had lots of voids and some shiny bits of chip were visible.  So hopefully my application was an improvement.

Anyway, put it back in the rig... and the middle fan not spinning.  Totally free to spin but apparently no current reaching it.  Ran it for 30 minutes with just the 2 adjacent fans doing the work and it seemed to run ok, without overheating.  Which at least shows the upgraded fans are working way better as previously when the middle fan was seized up, the temperature soared.

I wasn't very optimistic but decided to take out the middle fan and resolder the circuit board to coil pins.  Put it back and much to my surprise, all 3 fans are now working perfectly!

In theory, with the nice new ball bearings, it should run for years, long after the card ceases to be mining economic.  The bearings are sealed so no matter how much dust gets thrown at it, they shouldn't seize up.  In theory.

I have 23 more dusty cards running that will likely seize up in the next few months so I have 23 x 3 = 69 more fans to replace with bearings.  If anyone knows of any magic special way of separating the fan blades from the stator without risking breaking anything, I would so love to hear it!  Anyway worst that can happen is I break a fan that isn't working anyway so no big deal.

I built my first rig just before a long hot summer and now the summer is drawing to a close the fans are starting to die.  I've been looking around and I can't find any part numbers or sources of replacement fans.  Each card has 3 fans and I measure them as being about 75mm but I don't know what kind of connector is used. Nor can I find any instructions on how to remove the and replace the fans.  I'm hoping someone here can point me in the right direction.

I thought I would share my 8 GPU rig design.  The 8 GPUs are mounted vertically in a frame of total size 50cm x 50cm x 35cm.  In theory, there is room to mount up to 14 GPUs but I'm happy with just 8 in each rig.  The one on the left with it's guts hanging out is the prototype.  (One of these days I should shut it down and tidy up the cabling)

The overall size and dimension allows for easy pickup and carry and I suspect the extra space between each card allows them to run a little cooler.  Single 1500 watt Silverstone titanium power supply in each along with an ASUS Z270-A motherboard, 120GB M.2 hard drive, 8GB ram, running windows 10 pro.  GPUs are GIGABYTE 1070's in the first rig, 1070ti's in the 2nd.

I discovered recently quite by accident that if I install the normal nvidia driver for my MSI notebook (has 1070 gpu), hashrates are at least 10% higher than with the notebook nvidia driver.  Both drivers identical versions.  Tried reverting and installing the nvidia notebook driver, hashrate dropped.  Put the standard nvidia driver in and hashrates back up again.

Fan speeds, temperatures, power usage identical.  Only hashrates change.

Not what one would expect!

Woke up this morning to find I've lost $3000 worth of BTC that was in my nicehash wallet.  Ah well c'est la vie and all that...  So now I'm looking for an alternative and am confused.  I only started serious mining a couple of months ago and have lost the lot. 

ZPOOL seems to pay out in BTC irrespective of what coin is mined.  Is this true?  I'm also trying out "Awesome Miner",setting up a "Managed Profit Miner" asks for a BTC address so again whatever I'm mining I can get paid in BTC.  This is ideal for me but is it so?  I'm hoping someone can confirm.  Help a guy out who is trying to get past the instant loss of $3000 supposedly due to the site being hacked.

I want to run my rig without a monitor and just use remote desktop.  However if I don't have a monitor plugged in, the onboard video adaptor is automatically disabled.  This degrades performance quite noticeably.  Lower hashrates and very sluggish remote desktop response.  It seems like in the absence of enabled onboard video it uses one of the 7 gtx 1070's to show the desktop.

The motherboard has 27 million bios options but I can't seem to find any option to keep the onboard video enabled when no monitor is plugged in.

Would be super thankful if anyone can point me to a solution.

Prior to ver 2.0.1.2 beta, nicehash sets P0 on starting up.  However after upgrading to the newest version (2.0.1.2 beta), there is no longer an administrator request to set P0 and hash rates are noticeably reduced.  Uninstalled and back to 2.0.1.1 beta and it works fine.  Has anyone else encountered this?

I understand windows won't work with more than 8 gpus of the same brand.  By "Brand" I assume the meaning is either amd or nvidia i.e. you can't use an MSI nvidia card if you have 8 other gigabyte nvidia cards.

That being the case, what is the best AMD based card broadly as good as a GTX 1070 in roughly the same price range?  This is for nicehash.  RX580 seems to be a big step down for mining anything but ether.  I'm looking for something that can be underclocked to use around 120 watts.  Any suggestions?