Topic: It is 2020 time for a new diff thread. - page 9. (Read 14862 times)

Ja. The problem lies with the built-in ones.
Old way was to feed in 12vdc and then go through a Vcore regulator on each board to feed the strings. Problem is that final regulator gives ~2-5% efficiency hit per-board. Solution is to use the PSU itself to do the regulation by making it programmable. That in it itself was responsible for a large part of better miner efficiency these days.

That also let mfgrs make longer strings of chips because a custom PSU can put out whatever voltage is spec'd. Gee - shades of Bitfury who used to use boards with strings long enough to take either 24 or 48v feeds... As I recall the Bitmains range from 15 to over 20vdc depending on the model. As always, using higher voltage (to feed longer strings) while pulling the same current per-chip is a win-win.

Only main glitch to that is that all the different PSU's respond differently to the signal the control board provides to set the output voltage. Should be a piece of cake for a mfgr to standardize that across their models but noooooo....

yeah no work around. this is the outcome for the gear now.

they learned with the s9 and truly have made you commit to one or two models and order spare psus.

er..I know folk with PSU's ...not sure if that is what you mean..lots and lots....

brad

edit: unless you mean no 'workaround' for the built in psu's and single bus?

psu's used to be interchangeable.

they are not.

well they are less interchangeable.

My 68th m20s psu died.  I sourced a direct replacement oem from hmtech.co. it came as model p21 not p20 it is a bit more powerful. it has an id chip and direct talks to the control board.  So In order to make the gear work I had to upgrade the firmware so the unit would recognize the psu.

Now the control board only will use psu p21.

https://pangolinminer.com/product/power-supply-unit-m10s/

it is in stock and works on 3 different miners. I have 2 m20s and 1 m21s. So I have to think is it worth it to buy  a spare or hope that they are still in stock when the psu dies.

it is easy for me to have purchased a s17 psu I have 17 s17 pro and 1 s17. I have 1 spare. and once I use it  I have no access to any more.

https://shop.bitmain.com/product/detail?pid=00020200403105249382Ow2fHTq80690

the 1 spare I have will work on my
2x   t17
1x   s17
17x s17 pro.   20 units only 1 spare  more then 1050th

my
1x innosilisicon. t2 30th no spare
1x innosilisicon t3  39th no spare no source I lost it or 39th
1x innosilisicon t3 50th no spare but I think I can source it at hmtec.co

my whatsminer

2 m10
2 m20s
1 m21s I think I can source from hmtech.co

my Avalon
2x 1047
2x 1066.  no spares no source   I lost a a1066 or 50th

So for me if I buy asics for btc it should be t17 or s17 or s17pro as I have a spare and may be able to source another

but those Avalons and those innos = no psu's available.

so eventually I will lose them.

Fans? Yes, mostly.

PSUs? Hell no. They vary by every producer and even model by model. Just look at 17th generation of Antminer series: APW9 (S17, S17 Pro and T17) and APW9+ (S17e, T17e, S17+, T17+). Then there is S19, S19 Pro, T19 with APW12.  They all can vary by Wattage, dimensions, connectors and many more things. S9s were easy because you could literally use any PSU with enough PCIe connectors. But after S9, everything changed.

Innosilicon also has different PSUs for every model or almost every model afaik.

control boards i get, but arent PSUs and fans more or less bog standard across the mining space? like 8 pin power and 120/140 mm screamer fans, maybe different fan connectors but easy fix no? 3/4 pin is the same even if different order for pwr/gnd/pwm/tach or whatever?

What I like about this topic is that we go into very deep details, I am sure most miners miss out on these critical points, sadly I haven't been thought this lesson by anyway and I had to learn it the hard way, one thing I learned is that you must always try to own the same gears, by "same gears" I don't mean the exact same model but at least models that use the same control board, PSU and fans.

Eventually, you will either lose a hashboard or a PSU, if you have 10 different mining gears, then you must buy 10 different spare PSUs, 10 fans, 10 control boards, etc, this does not make any sense economically, compare that to someone who has 50 gears of the same kind, in the worst-case scenario, they don't even need to get any spares, eventually when 3 boards are dead you end up with 2/4 fans, 1 PSU and 1 control board, so by limiting the number of different miners you can always be prepared for when the spare parts go out of stock.

Most asic builders are now selling one of psu's

The t15 and s15  psu sold out. I had a spare  So I saved 28th

The t17 t17e t17+ s17 s17pro s17e s17+  psu's are sold out.   I have a spare  just 1

you can get s19 s19 pro t19 psu

My 39th innosilcon psu is sold out I have a dead 39th

My 50th avalon a1066 psu is sold out I have a dead 50th

both my whatsminer m20s psu died I got spares.

so I lost 89 th to dead psu's this summer and can not replace them.

I saved 136 in whatsminer m20s
I saved 28th in s15
I did replace 164th with spare psu's

I do have 1 s17 psu spare.

Point is this all serves to keep diff lower.

And you need to decide how many spare psu's to buy when they are still in stock.

All of the above has pushed me to get some gpu's vs asics.

I won't complain if that happens, although I prefer lower price and more BTC rather than higher price and lower BTC income, I think bitcoin price has much more room to grow than the hashrate does, I am almost sure that a few years later even very old gears of 100w/th and above will be profitable to mine with, by simply looking at the hashrate vs price, you can see how possible that is.

Or BTC moons like a mofo to 13k then 17k then 20k then ath of 25K.  driving the diff way up past  20t.

I am glad to be the first person to report the massive -1.2% drop in bitcoin mining difficulty. 


We had 20% drop in bitcoin price, and only 1.2% drop in diffiuclty, this has to do with the fact that difficulty doesn't adjust fast enouh, if we could keep the current price levels then I expect to us a 3-4% drop in difficulty for the next adjusment.

Current critical suppurt for BTC price is the 21weekly MA which is right at $9933, in the past couple of years everytime we closed below that we dipped another 20-30% which means the 7k levels, but let's wait and see, overall, things look pretty bearish for the short-term for makets mainly due to the U.S elections, post election IMO will be bullish for all markets.

This should keep difficulty at check, small drops of 1-2-3-4-5% all the way to after the elections is what I expect.

Very close to -1%


over last 30 days
btc price at $10,179.00 a big drop from 12,305

over last 30 days
dow was      29,100  dropped to 28,133
nasdaq was  12,056 dropped  to 11,313
S&P.     was   3,543  dropped. to   3,426

Gold was 2060 dropped to 1934
silver was 29.14   dropped to 24.76 but is now 26.94

So the Dollar rallied last few days.

This is why I stick to diff calc's and not price calc's

I hope so too. I sold most of my ETH and BTC near ATH of both, now want to go back in .

I would love to see 9400 stay around for a few weeks. it will drop the diff a bit.

Bitcoin touched $9980 which is a -17% from the high of this week and just about 14.5% from the opening, we might see some recovery from here, but overall this is bad for miners who are mining at the edge of being profitable if we don't make a quick bounce I expect to see some of the hashrate leave tomorrow.

Current base is as follow


Not so many blocks left for this epoch, so I wouldn't expect a huge turnover in these numbers, maybe we wipe out the 3% we gained from the last difficulty adjustment although with only 508 blocks left to be mined that might be a bit hard to achieve.

yeah

t17 t19
s17 s19
s17 pro. s19 pro

all need 1 pair of power cords and an eth cable a minute or 2  from the truck to the shelf to powered up.

if it's a tenth of a Yuan then it's 0.26 Yuan which is 3.6 cents, it also says 2 to 3 so could be as low as 0.2 Yuan or 2.9 cents per Kwh, which then would justify why to move back from Sichuan.

Time                          Place                Cost

May to September      Sichuan             1-2       cents  USD
October to April          I Mongolia         2.9 -3.6 cents USD

Depending on how many gears you have, and how does it cost to move them around twice a hear, it might not be worth it for everyone but certainly is for some.


They could be doing both, but without a doubt most of them "shlep" those gears, if all wires are ready then it takes a few seconds to connect the miner, a few people can unplug and re-plug a thousand miner in a few hours, you just need to drop it, plug in the power cord and the LAN cable and off it goes.

eh 2 kiloclicks, 1 day's continuous travel https://goo.gl/maps/hcYXrKG6wFoxxxoYA
they do just bung 'em on trucks, there were pictures of the swap-over from ~a year back
labour, transport, rent and loss of time all cheap compared to savings on elec.
the containers were marketed by bitfury longer ago, no idea if they really sold
now you can get from here https://www.upstreamdata.ca/datacenters kinda, probably elsewhere

shipping hash around the globe.. are those shipping containers fitted out for miners still a thing (saw some but havent paid attention of late)? or do they shlep thousands of individual miners around and disconnect/reconnect each one? just curious although i suppose manpower is cheap and if its only done a couple times a year..

most mining facilities ive seen (admittedly not many) are in buildings specifically built or retrofitted for mining. but if they are cheap enough to leave empty part of the year i suppose it doesnt matter.

sorry sorry it says 'mao' which is a tenth of a yuan afaiu. highlighted:


not a great sino-linguist here. can anyone confirm? would explain why they transport the rigs back to Southern Mongolia after the rainy season in Sichuan. still even with the extra 1/3rd price hike, very cheap