Topic: Happy New Years! Seventh alt coin thread! - page 57. (Read 34194 times)

https://news.bitcoin.com/report-crypto-miners-bought-3-million-gpus-last-year/

"....More than 3 million graphics cards have been sold to cryptocurrency miners in 2017, with sales reaching $776 million, a new report revealed. According to a major manufacturer, prices of GPUs will continue to increase in 2018, despite expectations of decreasing demand in the mining sector.  ..."

1 fully good msi aero   ----------------------  this = 800
1 fully good asus turbo ---------------------- this  = 800
1 funky fan repair asus turbo --------------- this will be less as the fan repair voids  warranty

these are the 2 asus

Yep, I'm in Tampa, FL. I'll PM you my details and await your go ahead to send you the BTC.

You have a deal assuming you are USA based.

Give me a few hours to figure where I put it.

Phil how much for the TIs and how many?

I'll definitely take you up on this if it isn't already sold.

Okay  Waiting for the 16 gpu beast  to be shipped  I am  looking at my

list of gear sitting idol

I will need to make a large  sales thread.

I have  an asus turbo 1080ti in good shape.
I have an asus turbo  1080ti with a fan repair.
I have 2 msi aero       1080ti  in good shape.------- one is spoken for.

They are going.
I have many mobos  for smaller builds  they are going
I have some psus for smaller builds they are going.
Many fans  of shapes and sizes the are going.

I have a damaged b250 8 slot + 4 riser board.  I tore off 1 sata connection in an act of clumsiness but  someone with skills could fix it.



I would sell it to someone in  the usa for 35 usd in btc  that includes shipping.  It fits in a large flat rate box for gameboards

https://store.usps.com/store/product/shipping-supplies/priority-mail-large-flat-rate-board-game-box-P_GB_FRB

So, only one power cable for the psu in the middle ? or all? for what aim in the end ?

If I was going for density, I'd go with a used Supermicro 847.  They make a 36 bay version that will hold a motherboard (same layout as the Alibaba one but much better quality I would imagine) and a 45 bay version that requires an external host.  They can be found for around $500 shipped.  Filling the higher density one with 45 8TB drives @ $160, works out out to $21.39/tb.  So way cheaper than either the Backblaze or Alibaba solutions.

I have used Supermicro chassis' and motherboards for 15+ years and they are rock solid.  They share 3rd place (or 2nd place, depending on how you look at it) in the server market worldwide:

Space is not too much of an issue, but I'd like to avoid USB 3.0 cables/hubs to improve read times and remove failure points. Did some rough estimation on excel.

https://docs.google.com/spreadsheets/d/1ShvF8epPEdnPPgHUwL1vnFUdcTAjVr6jOnogX5scQhk/edit?usp=sharing

Backblaze 60 drive pod comes to $33.57/tb. The alibaba 36 bay solution is $28.65/tb. Backblaze cost 1.17x per tb, but has 1.67x the density.

There's also a weird 48 bay 2.5'' plus 12 bay 3.5'' that comes to $31.32/tb. This one will give best read times since many drives are just 5tb not 8. Details in the spreadsheet. Since they all have the same compute power, the alibaba 36 bay will be fastest to switch to another HDD coin if necessary.

If I can get the bare backblaze pod with just sata backplanes, wires and pcie to sata connectors for $1500 or lower I will be all over it.

I am looking at this design and awaiting it on tues next week.

I have 14 1080ti hybrids.

After I do a long review and test with enclosed 1060s

I may pull them and try it out with 14 hybrids 😀

this is awesome stuff.

Just FYI, if your using the HP common slot psu's like in the picture you can just daisy chain them together via the breakout boards now with the same black and yellow power cables...atleast this has worked flawless for me on all my HP model psu's, not sure about other versions....

Since i buy up used ebay server psu's for like $8-20 a pop, i try to run them at half or less the rate they are rated for, so my 1200watt units i run them around 600watts or less and such...

Every rig i have running on HP common slot PSU's with the breakout boards i have, i have them all daisy chained together.....so if one fails the extra one can take up the load, but you cant just throw them together and expect them to work well, you have to take the time to drill the hole in the case to access the voltage adjuster screw. When i setup a rig, i plug each server psu into a kill-a-watt meter, then i boot up the rig, adjust the voltage before loading up the rig so all 3 are the same according to volt meter (12.4v), then i start mining, i typically let it mine for 20 minutes to get everything nice and hot, then i come back and look at the kill-a-watt meters, typically from the start there will be 1 psu that is trying to do like 75-90% of the load of all the gpu's... while the others are just barely doing anywork, so i take my screw driver and adjust the voltage through the drilled hole i made carefully and adjust them all tell the watt meters on all of them are roughly equal load, thus they are all being used equally and let the rig mine typically overnight setup like this monitoring the load on each unit, typically i can spot a weak PSU this way due to the load drifting all over the place, so ill replace it with another used unit and if i can return it for my money back if the ebay sale has some sort of like 30 day warranty or something.. twice a weak PSU was only drifting because the fan on it was worn out and not working correctly, so i just bought replacement fan from china and those units are actually back in rigs doing work.

I can completely turn off any 1 of the 3 PSU's while the rig is mining and it will not skip a beat and continue mining shifting the load to the other psu's

If space is at a premium, a current-model Backblaze 4u rackmount case mounts 60 drives - abet it's a little longer than a "normal" 4u case, but it is specifically optimised for lots of drive throughput.
They use a Supermicro motherboard, should be able to upgrade the specified CPU to one with more cores *IF* needed to handle that many drives in a reasonable read timeframe.

try to lower every clock by 5%

or  try to switch to smos

I don't have much advice for you.  I was having similar issues at one point with a 6 card rig I had on ETH.

My solution was two fold so I don't know what might have resolved the issue.

Solution was since my windows boxes are low resources, they didn't run 6 cards well.  So i switched to four card rigs.  I also tuned down some of the overclocking settings.

I suspect my issue was resolved more with not trying to run 6 cards on a low end windows PC.  More than likely in your case you have more available resources, as I was running some lean boxes.

For those of you still mining ETH... I've got a weird issue on 3 of my 4 6x470 rigs that Im hoping to get some advice on.

Seems like ever since the Window10 Fall Creators Update (i put it off as long as i could), I've been getting this strange behavior using Claymore 10.2. I also tried 10.5 & 10.6 but made it worse, so went back to 10.2.

Here's what happens:

- miner starts fine but in some cases after restarting, GPU0 starts running slower than all other 5 GPUs. Usually between 20-23 vs the 25-26 of the others

- it can run like this for a while, but then many times will eventually end up with a "GPU 0 hangs in open CL - miner restart".  Sometimes it does this reset every 30-60 minutes but sometimes it ends up doing it every couple of minutes, which of course kills mining returns

- since this is occurring consistently across 3 rigs and is always GPU 0 being the culprit, it seems very likely that this is a software issue and not hardware.

- I had a somewhat similar thing happen with Cryptonite miner... GPU0 was always way slower on all 4 rigs.  This was solved by manually designating the "-h" value for GPU0 in the .bat file, but there does not appear to be an "-h" option in ETH dual miner.

Does anyone have a suggestion on a setting tweak within Claymore to try to and fix this issue? 

 

I tried numerous Burst pools back around mid-'17 due to all the DDOS attacks and finally got onto pool.burstmining.club and have been happy there ever since (about 6-7 months running).  The only reason i hesitated to go there in the first place was that many in the Burst community believed some of the owners of this pool were the ones launching at least some of those DDOS attacks. It was a bit strange that through that really rough period of time, this was one of the only pools up and running consistently, but finally I just gave in and joined where i could get the most return and stopped reading all of the in-fighting within the community.

5 x 1200 = 6000  watts   which would do an entire   30 amp  240 volt  circuit   no need for a pdu  just an l6-30r socket and a cable to the backplane


It could do 16 1080tis  set to 80%  tdp

16  x 200 = 3200  watts     

load on the psus = 3200/6000 or 53.33%

load on the circuit  = 3200/7200 or  44.44%   you could use 20 amp  240 volt citcuits    which would be 3200/4800  or 66.7%  load

What a great idea, I just love HP DPS 1200 PSU's. 
They're robust and reasonably priced.