Topic: Buysolar and I will getting the A1041 to review and demo. (Read 736 times)

no more  samples are available

I did order 2x 48 port switches  from newegg they are cisco and should work well. they are the ones I found via google.

The new network will be

verizon modem to
router to
3x  48 port switches
to miners in 3 groups

the route will also handle the warehouse owners net traffic  it has 1 gb wan in

8 ports out
and 3 bands of wireless.

he will have under 7 pieces of gear on the network.

we should have about 85 pieces of gear setup with more to come.

even if we max out available power we will be near 170 pieces of gear.

So we should be good to go.

very soon on our end.  but we will only have 2 a1041's in the short term.

1gpbs ethernet mandates the use of 5e. Testing 5 is pointless at this point in time, no one should be buying those anymore. If the miner only does 100mbps 5 vs 5e the 5e will let less errors in (translates in slightly faster speeds). Similar to...

Using cat 6 for 1gbps. Its not really needed, but its better. For 10gbps it can work in short lengths only (50m vs 100m). 10gbps should be done by cat 6a or cat 7. I don't think you are going to find any miners with 10gbps lan ports, but it might be available for cascading switches and such.

And since you enjoy the fun, there are also cat 8 patch cords...

In short:

cat5 100mhz
cat6 250mhz
cat6a 500mhz
cat7 1000mhz
cat8 2000mhz

If you ask me, 5e or 6 at most is what any mining operation (and most LANs) would ever need.

I did not want to confuse you, but you can actually use anything from 1 bit to 32, not necessarily just 8, 16 or 24. You could use, 12, or 20 bits. Just remember there is always a network address and a broadcast address, any remaining IPs in the middle are the IPs you can use.

Use this calculator and see the endless possibilities. Private nets are up to 24 bit (10.0.0.0/8)

Most switches are layer2, they don't care about this except the very smart ones that can to layer3, and those, of course, can do the whole 32 bits. Limiting yourself to 8 bits is nonsense. Maybe that was more critical when people used 10mbps hubs instead of switches, but that's long gone in the past. Unless you have real cheapo gear... (TP-Link? )

Computer with pfsense is good too! There are also many embedded devices with multiple ports, such as pc engines you can install any decent os yourself for routing, firewall, etc.

Yep I agree some smart switches can handle a loop with zero issues.

I will have to look for one soon.

Since I want to do extensive testing of the a1041.

A ring setup with 50 units may be really good.

Buysolar and I want to cut a deal for a goodly amount of these units.

While I am here do you have a smart switch in mind that can handle a ring setup?

---

Edit
Thanks for pointing this out I did some googling

I think Cisco me-c3750-24te-m 3750. Can do what you say.

I was on Cisco website and they say they can do it.

If so then the tech offered by Avalon is even better then I had hoped.
Huge networks can be done far easier then any other miner I know of.

As I wrote in the other topic, you need a good switch or you will get a "Switching-Loop", that has nothing to do with the Avalon

Ring idea makes a loop and crashes the net completely.

This has happened to me when I do a lot of eth switch work and make a mistake causing a loop on a switch.

So for now the loop idea is out.

I want 10 of these as soon as I can get the ten.

If I get that many I will do extensive tests of eth cables.

Chain of all cat 5
Chain of all cat 5e
Chain of all cat 6
Chain of all cat 7

I will measure hashrate on all 10 units with each set of ten cables

As we all know most of the time cat 5 or cat 5e are fine.

But with ten cables linking cat 6 or cat 7 may be better.

It will be interesting to see if

A. Gear is stable in a ten unit setup.
B. Gear hashes lower with cat 5
C. Gear  hash descends as link grows.
D. Gear hash is improved with cat 7.

Ten unit test should be enough to help people decide if they want the gear.

And how long to try to make the chains.

I am just getting bigger now. I know about subnets but in a classroom 10-15 years ago.
Real world 150 miners is most I have done in one spot.

the modem is verizon the router is verizon.  belong to the warehouse owner.

so a 5 port switch to my 2 routers was cheap enough and my 2 routers were also cheap. my third router will also be cheap.

so being isolated from the owners gear still gives me an easy 750 miners which is not happening.

I do have to say have a different ip for the avalons in my case is good and not an issue.

It is pretty much good and allows easy access teach one. up to 250 or 254 then things get more complex.  So is bad the right word maybe not .

That ring idea is good I will check it later.

255 is typically reserved for broadcast and 1 is usually for a gateway depending on your setup. The range is 1-254. To your question on the daisy chain I believe this answers it


@Phil instead of getting a router and setting it like that what about a pfsense box, operates as a router.

While I wouldn't suggest it, on your current one you should be able to change the subnet from 24 to 16 and increase your available IP's. It depends on  your setup but if you can change it to 16 you would be able to run machines on 192.168.x.x instead of 192.168.10.x.

Most big mines can run tens of thousands of machines off one firewall, they don't typically go to a /16 but a /24 with a change in octet 3 for groups of units.

Honestly though Phil, saying that the 254 address limit is bad is pretty nit picking, it’s a limitation with IP addressing and not the miner. The IP range limits are present in most miners outside of canaans due to their independent controllers . I think daisy chains are a brilliant idea that would help reduce clutter and cost, it would be even better if it had ring type topology to help with redundancy, can you test this Phil? Meaning the second miner goes back to the router and if any one cable is unplugged the miners still are connected.

doesn’t IP address ranges end at 255?

Found out how to hook up daisy chain

Ran it over night  does about 11 amps or 11 x 230 = 2530-2600 watts  and about 37.05 th  so 37.05/2530 = 68.29 watts to 70.17 watts on high speed.

Second unit was picked up from buysolar.

I have both running at normal speed

they do around 30-31 th at 2016-2015 watts

so 61 th at 4031 watts on normal speed.

both have loose style psu.

and basic demo/beta gui.

eth daisy chain does not seem to work, but I will fuck with it some more.

so I have 2 ports 2 ip's at the moment.

Even with these its getting harder.

Blokforge has upped the price of the August batch to $2300, September sold out, so October is the earliest at reasonable prices.

Canaan has some September left, but August all sold out.

As I'm not shipping to the US I'd probably choose to get them from Canaan direct.

Lets say

once fully tweaked it is 60 watts a th for the 31th setting not 56
and it is 68 watts a th for the 37.2 setting not  65th.

at current prices it is viable gear.
it does not need many switches if it allows 100 or more for a daisy chain.
It Steve said to me he could send me ten this weekend I would take them with  the loose sloppy psu setup and wait on the buss bar psu's

Avalon is reliable gear for me so far.
eth daisy chaining seems like a good idea.

I will drive to buysolars house and bring his to my house and setup a 2 unit daisy chain. Should video it on sat.

Thanks for keeping us up to date Phil, as you said there isn't much hardware around at the moment.

maybe they are maybe not as the psu I have could simply be a 88% psu.

I am posting a video in a minute you can see this is clearly in beta status.


Your video will be live at: https://youtu.be/btdHHbfSJ84

the above will take time maybe 10 more minutes to load.

this demo does not have the same psu setup as passthepopcorn's unit had.

I have multiple meters and will clock this gear  very closely.  the rooms temp is 30c or 86f

the rating of 1736 watts is for 20c or 68f

I thought they were supposed to be 1736W @ 31TH?

volts at 230.6



watts at 2061



we will not be fully solar here this site has 5/6 cent power deal.

as of now we plan 90kwatts of solar 85kwatts of paid power.

cost say 3-3.5 cents net for 175 kwatts this is 126000 monthly kwatts at 3.5 = 4410 USD a month bill

we could add 300kwatts more of 5/6 cent power but that is for 2020 or dec 2019

she seemed to settle in at 31.5 th
and 2055 watts

or 65.8 watts a th.

so if we were all avalon a1041 we would have 83 of them doing 31.5 th or 2.61 ph

earning 2.5 coins a month   which is why gear is  selling like water in a desert.

we have gear and do not have the room for 83 of these but depending on the next set of tests, which will be the ethernet cord daisy chaining. We are looking at getting 10 of them.

My assumption is the we all go to built in psu with buss bars. and not the loose style I am testing today. I also assume more pools and more features in the GUI. As mine is exactly the same as passthepopcorn.

next test is high speed so far it pulls 2650 vs 2050 watts it hashes at 36.9 vs 31.0 th

so 71.8 watts a th I will let it settle more.

normal 2050 watts 30.8-31.0 th  this was about 1 hour 20 minute test. 66.12 watts a th

turbo   2640 watts 34.9-37.1 th  about 1 hour and change   71.16 watts a th

So beta psu
and beta software   it looks good  but not great

Still with todays prices this gear is a moneymaker.

biggest quirk was it did not run well on mmpool.org a small btc mining pool.

What is it worth? good question

Hard to tell with the beta psu  I would not rush to buy it.

I do assume Steve of canaan will post on the psu as I prefer the attached style over this loose beta model.

It has arrived. Unlike the other review I have been given a stand alone psu with cables.  What does this mean it means  I had to take off the controller case to see the buss bars as they are not marked pos or neg.

The psu did show pos and neg so I was able to correctly wire the gear. A relief for sure.

It is not that loud it can be sound proofed with the psu I was given from them.
I have it set to normal  and I am around 29th and 2060 watts.  That is better then the m10 not as good as the m20s. from whatsminer.  It is  pretty well made I will need to give in my case examples vs pass the popcorns


note the clean built in psu with buss barr attached on passthepopcorn's gear.

and the loose psu hook up



worse  then that is the miner buss barr screws are not marked



I removed the controller lid and could see the - + markings.
thus hook up was okay

close to 29th at the pool.  and 2050 watts  that is  2050/29th on low speed 70.6 watts a th

my m10s would do 2250/31th on high speed 72.5 watts a th

Okay the miner arrived but not the psu. Disappointed mostly seems to be fault of DHL gear was due thur.
it is fri.

2 boxes
one for miner came today
box with psu is floating about

buysolar is getting a set of 2 boxes like I am in his case neither box has arrived

here is youtube video of opening of miner box

Your video will be live at: https://youtu.be/vyQWSx8XL40
Published  you can see it

full size image
https://i.imgur.com/nAOQnDA.png

I was always able to use one 24 port switch

But the clifton build changed my world of mining.

we are using 70 units
two 24 port switches
one 48 port switch
three routers
modem to one with 192.168.1.1 for all non-mining gear to two mining routers

one with 192.168.0.1 for all sha gear uses 24 port and 48 port about 40 pieces
one with 192.168.10.1 for the L3+ uses 24 port about 22 pieces

the setup above could do  say 240 + 250 + 250 = 740 miners  you would many many many switches I guess 30+ 24 port or 15 plus 48 port.

but if we push avalon we could do 1000's of them with no extra switches.

realistically add the 87kwatts use 35-40 avalon

then if we do 400 more use 150 avalon  not a single extra switch.

Avalons are a beast when it comes to reliability, I've never had a real issue with one. It's been a shame they haven't been able to come out ahead of everyone else on an efficiency launch. I think it's been a huge step getting rid of the external controller then following up with the Eth daisy chaining keeps them unique and easy to network. All the benefits without the potential points of failure.  I wonder if these will still be able to operate as 1 worker for all linked chains or if they will now be individual workers. It didn't even dawn on me the switch savings, I'm lucky enough that I'm still small and can get away with a couple home routers networked together to meet my needs and boost the home wifi network.