I think they have tried some way to band-aid everything, for some reason the new cgminer binary takes a good 10 seconds to even start now, plus you get a stratum disconnect instant reboot. I had been using a proxy to switch pools now it causes a reboot. I only have 2 boards upgraded hope the jig shows up soon but now having to run the new software makes a lot of work I did useless if it is going to reboot my machines all the time.
I don't know, getting about frustrated I hope they let the code out, even if to a few people I would like to know what is going on with this, why there are so many external support programs that they run and why on earth do we need to reboot when there is a slight issue, I might need to test a little more to make sure but it did look like that is what is going on.
Edit
Does look like any type of connection issue now reboots the Pi, internet goes down, reboot loop, pool goes down reboot loop. Why we decided to do this I have no idea.
I feel your pain, been suffering off/on with same kind of issues. Here is what I know: I know that Innosilicon is actively working on the problem(s) so good news there. I've detailed my issues to them as well as posted about them here in some other threads. Bad news is we are in "hurry up and wait mode" -- No sarcasm intended here, I call it the "reboot dance" when it gets into the mode where it doesn't want to cooperate. It's like trying to get rabbit ears on a TV to get good reception, sometimes you have to stand there with 1 leg and 1 arm in the air to get it to do something different... I know they are in process of shipping 2nd Jig to USA Yea I do agree they are working on it and they have been good about talking to me about my issues and listening , I don't understand the lets reboot on any sign of a problem but I do know they want to make it work and now that there is someone else in town it makes it more critical. Just kind of stinks I spent a lot of time writing APIs and monitoring stuff to basically to maximize which pool i was working on but now that is useless. I did verify that it runs /sbin/shutdown when there is a stratum connection problem, renamed the bin file to test but it shuts down the board and just quits working just tons of comm errors so no way to stop it. |
|
|
|
Stability on multipools is one of the first potentials in optimization, which the very first I would do is extranonce. More stability means more net hashrate. More possibilities in under and overclocking in both clockspeeds and possibly voltage settings might increase efficiency for the long run or increase profitability (at expense of electric costs) on the short run.
I have however not heard back from Innosilicon.... Might be they are going with the other guy, or they are going to do it themselves, or they still have yet to get back to me.
I have been bugging them about the software as well, they said they have to talk about internally. Something new I have noticed that I am not happy about is I use a proxy here at the house to switch pools, now when I switch pools the pi instantly reboots. I don't know what kind of thinking is going into this, there are about 4 programs they start on boot, none of them are documented as to what they do. With my reboot issue, from what I have noticed but not yet confirmed -- 1 Get Failure causes the Pi to reboot. Haven't had time to dive into where that is in the code yet. I think they have tried some way to band-aid everything, for some reason the new cgminer binary takes a good 10 seconds to even start now, plus you get a stratum disconnect instant reboot. I had been using a proxy to switch pools now it causes a reboot. I only have 2 boards upgraded hope the jig shows up soon but now having to run the new software makes a lot of work I did useless if it is going to reboot my machines all the time. I don't know, getting about frustrated I hope they let the code out, even if to a few people I would like to know what is going on with this, why there are so many external support programs that they run and why on earth do we need to reboot when there is a slight issue, I might need to test a little more to make sure but it did look like that is what is going on. Edit Does look like any type of connection issue now reboots the Pi, internet goes down, reboot loop, pool goes down reboot loop. Why we decided to do this I have no idea. |
|
|
|
Ok, the L3 showed up a day early. I pulled it out of the box and powered it up and put it on Prohashing. Right now it is hashing above it's rating on Prohashing at the moment at 270mhs. Miner says 251mhs. It is slow to boot like the rest of the Antminer miners. Takes a while for it to power cycle. During the POST testing it fires up the fan at what sounds like 100%, then turns off... Then it reset the controller and then came back online. The fan is pretty quiet in operational mode at the moment. But the garage is rather chilly at the moment so it can load along for the time being. I will try to get you guys some data and screen shots later. So Eastshore.xyz did deliver what I ordered. Now to monitor and make sure it is happy.  I'd be more curious about the power use so far, pretty cool though. I have been tweaking all day with mine (A4) different freqs and voltages , but on an ltc pool so far, I see that they are more stable (A4 B3) but I think there is still an issue that is going to be a problem that is not fixed. I am not sure if it can be handled with just new software or a new reflash again. What I see is every time there is a coin change or basically block change there looks to be a board reset which takes time. Right now after some tweaking I am seeing 260-280 on ltcpool, will see what it does on prohash, plus got some power figures during all this, plus some fan mod everything doesn't need to run wide open all the time. |
|
|
|
I can confirm they are busy contacting people who can optimize the code.
They have contacted me, and for what I know of one other person, and we are in the process of constructing a deal (I don't own an A4 at the moment for example).
Awesome news! Whats the potential in optimizing the code? Are we talking higher hashrates and/or lower power consumption? Any estimate as to how much? I'm excited to see the beginnings of what I perceive to be innosilicon making the software open source they said they would. For me the priority is Stabilization > Optimization > New features. Right now unfortunately the A4 is not as stable for me as my A2's with Anx software. However, they have been improving it since release and I'm excited to hear they are reaching out to developers in an effort to make it even better. Stability on multipools is one of the first potentials in optimization, which the very first I would do is extranonce. More stability means more net hashrate. More possibilities in under and overclocking in both clockspeeds and possibly voltage settings might increase efficiency for the long run or increase profitability (at expense of electric costs) on the short run. I have however not heard back from Innosilicon.... Might be they are going with the other guy, or they are going to do it themselves, or they still have yet to get back to me. I have been bugging them about the software as well, they said they have to talk about internally. Something new I have noticed that I am not happy about is I use a proxy here at the house to switch pools, now when I switch pools the pi instantly reboots. I don't know what kind of thinking is going into this, there are about 4 programs they start on boot, none of them are documented as to what they do. |
|
|
|
|
Maybe I messed it but was there much before hand press about this new miner or just poof here we go? Does make me sad, I held out on the A4s , then decided to go for it then had a lot of lost mining with troubles ( which I still have on jig yet) then poof a new magical beast appears. I got a lot of hopes for the A4, been pestering them over the driver have a few ideas, plus they are built pretty good, at least none have thermal pads with paste on them like all the A2 boards I had. Oh well as it goes ..
|
|
|
|
|
I can't wait for it to arrive here, whenever that will be, I sent two boards off to be reflashed ahead of time to see if I could reflash my own but it didn't work. I am not a super expert at the ST MCUs maybe someone could tell me the memory areas to clone. I did the main program flash area 0x08000000-0x0807FFFF and it is the same but I just get a flashing LED which tells me the program must be running to some extent. Anyway the 2 weeks I was on vacation makes me programming them less important if the jig is almost here.
|
|
|
|
|
I can tell you that you more than likely have a blown ASIC on the board, I have see this LED combination on the old blue boards when you have a failed ASIC. I have repaired a few of these.
|
|
|
|
|
Well it is interesting for sure, I am guessing you just put the card in and hit a button? how does it get power , does it even need any computer interaction other than maybe power? I wonder what is in the tape blob, at any rate I milled out a board to see if my test pins line up and will check that , spring pins are not to terrible to buy, only need 4 connections really don't know which ones they are hooked to under that big booger on top. But at least it exists and works so we have that.
|
|
|
|
The jig arrived very late this afternoon. It is even more flimsy in real life than in the pics. I was frankly surprised it made it from China in one piece given it's construction. It didn't come with instructions, but I was able to figure it out from previous posts. Inno just emailed me the instructions, but I already had the miners flashed by the time I noticed they had already replied to my question about instructions.
The process to flash the boards themselves is very fast. Maybe 45 seconds per board. Sometimes the board will give you an error rapid flashing from your board. Don't freak out just repeat the process until the board checks out. It took me more time to disconnect everything, find a place to work on them, download the latest v2.0.2, and disassemble the cubes than it did anything else. I think if we get a bunch of these in at one time you could setup an assembly line and flash them all in a day or two pretty easy.
Now for the bad news on the re-engineering this jig. i don't think you can do it. The bottom box that you see in the pictures actually contains an A4 hashing board. It appears they copying the image from another board they already upgraded. I took some pics that I hope will be close enough that you can see how it is connected. I have my doubts that a cheaper alternative can be built based on what I can see. But then again I am no electronics expert so I will let you guys be the judge.
So my A4's have been flashed and they are up and pointed at Prohashing. So far so good. I have been in touch with Inno and Chris at Prohashing and we are getting things lined up for the jig and flashing. We will keep you guys posted as more of this story develops.
Well at least it made it and it worked, I am guessing it is a stand a lone rig, turn it on and put the card in and go? if so must just clone the memory from one board to the other, still might be doable for a cheaper model but will have to have a v3 board to copy from. May have one of mine upgraded then try from there. I main problem has been out of the 6 modules I have about 1/3 of the hash boards don't want to work , power it up and get a red LED blip then there off , software detects the board but there is only like 4 chips that report back out of 20 so who knows on that. Good job looking forward to the pics at least. |
|
|
|
I know Chris is listed as being in State College PA. So you could probably use 16801 zipcode as a general idea of what shipping to Prohashing would cost.
I am trying to get some clarification from Inno about something right now regarding shipping. I will keep you posted on that.
Do we need to contact Chris regarding scheduling so that he's not dumped on with dozens of machines all at once? If so, can you PM me a contact for him? Thanks. My thoughts on the order of operations for this reflash timeline: - Longsnowsm receives Jig
- Longsnowsm takes high res pictures of requested areas made by Mjgraham
- Mjgraham determines if a DIY reflasher can built
- If Yes > Jig may need to go to MjGraham first > Then on to Chris@PH
- If No > Jig goes to Chris@PH
- Chris@PH sets up timeline and scheduling for members/customers needing reflashing
- Fees? Costs? - TBD
If Mjgraham can make a DIY reflasher, I'd opt to go that route before sending to Chris@PH just because it saves time and costs overall. I think it is OK to send it to Chris, they can at least start, maybe when the first wave of updates die down I could take a look at it. I would like to have all the associated software, does anyone have any idea what were getting? We know the jig but a laptop? software to install on something? I can make my jig pretty easily, I have already taken apart one module so I can get my measurements off that, plus I have access to a circuit board mill so I can make it pretty easily. Of course were talking far more crude but hopefully effective. I hope and I think I'll ask if they are going to let us have any lower level access to the drivers or firmware, I do a lot of command line use of things and this new cgminer has a lot of strange things associated with it that are undocumented, plus I think it was a bare bones get it going attempt.
Until we know the costs for either option it's hard to say.
However, if the turnaround time is 2 weeks (2-way shipping + flash time on site, lost mining) as well as multiple rigs (multiple shipping), then it could possibly be cheaper/faster with a DYI rig.
Additionally, if I had my own DYI rig, I would feel much more comfortable about buying extra rigs going forward as I know I would have a quick/easy way to flash them.
That said, if the Inno rig is so expensive, why do we think we could build a DYI rig for cheaper? Wouldn't they have done that already?
My guess is this is something they use, just set it in push the contacts in and press go, may have advanced diagnostics on the boards and be able to do more than just flash, I can't really imagine what would make it cost more than $1000 but who knows , I wonder what is in the bottom of the box. At the least getting one board upgraded to B3 with the jig should be able to pull the firmware off and use that. It would be nice to know if there are any , it didn't make the shipping, firmware update smoked it , this think doesn't work at all kind of replacement from someone, I don't have any reason not to trust anyone but sometimes things just don't upgrade happily or survive the trip. |
|
|
|
It has been suggested we should have a vote regarding the flashing. So let the voting begin.
Ballot issue:
Should we ship the A4 jig to Prohashing and have them flash and upgrade the A4 miners to address the Batch 1 and Batch 2 issues?
1. Yes
2. No
3. Other: Write in votes here
Vote early, vote often, you cannot vote if you are dead... Sorry Chicago... LOL
Yes Even though I wanted to take a look at the rig. I sure hope they have done enough testing this round and this is the only round we need. I have only had my miners a week and am already tired of trying to invent a way to keep them going. |
|
|
|
Are they still gonna 'open source' this stuff for you guys? Hell with these jigs and such..perhaps you'd have enough info someone could improve the firmware and the SD software too boot ..it's not like (imho) innsilicon is gonna put a lot of effort into this  Just me watching this drama from the bleachers...don't have a unit...and at current LTC prices and their 1900 price with shipping looking more unlikely Innsilicon seems to be following KNC's scrypt and then adding some 'dread' to that process. Perhaps it is a front? KNC China.... All so very, very familiar. (BFL,KNC,A4's) I hope they do, although they didn't for the A2, I would like to make a few changes for the cgminer for sure, I'll just be happy when we can get them updated and that actually fixes the problem and there is not another issue that requires everyone to update their stuff again. |
|
|
|
As soon as I get the jig I will tear down the rigs and get you the pictures.
I was wondering what state you are located.. I think its better to just have the Jig at one place and not ship it around and chance it getting damaged.. I am located in MO OK just checked your profile..I see your in CO.. Any time frame when you will get it? They told me they shipped it with DHL last night. So I don't know the ETA yet for the jig to arrive. If we can get these jigs cloned that would be the best of all worlds if we can do it at a reasonable cost. That would be great if it can be made cheap everyone could buy there own it maybe cheaper then shipping miners round trip? If you don't want to ship around the jig and worry about every owner learning how to use it, I would be willing to set up a service at Prohashing (Located in Pennsylvania) to flash miners for others in North America. If we can get the jig, then people can ship their miners to us and I can do the flashing. We wouldn't be looking to make a profit from the service; we would just want users to cover the shipping cost of the miners both ways and a small fee for consumables. If anyone's interested, then we can work this out in more detail. I don't really have an issue with this either, I was hoping to get a shot at using it. I still would like the files so I can flash my own but I know that is not for everyone. |
|
|
|
As soon as I get the jig I will tear down the rigs and get you the pictures.
I was wondering what state you are located.. I think its better to just have the Jig at one place and not ship it around and chance it getting damaged.. I am located in MO OK just checked your profile..I see your in CO.. Any time frame when you will get it? They told me they shipped it with DHL last night. So I don't know the ETA yet for the jig to arrive. If we can get these jigs cloned that would be the best of all worlds if we can do it at a reasonable cost. It will be interesting when you get it, I am curious about the software end if it is a you pick a bin file and send it or is it a self contained unit. The setup looks fragile to a certain degree with all those floating wires and rats nest looking stuff , plus what is going on under the thing. I don't really see why we cant make a clone of the flashing setup. Plus I don't know why you need to take the fan off, I just unhooked it and pulled the back plate off that way you can hold the cards when you loosen the screw and pull one out. I don't get the two USB devices seems interesting for sure. |
|
|
|
If someone can send me some high res photos of both sides of the board, and some pics of the jig and where it interfaces on the board, im pretty sure I can whip up a flasher for you guys that would require about $30 bucks worth of parts. I pretty much did the same thing for the Alcheminer guys when their boards had a major firmware issue that needed reflashing.
I was able to connect to it with 3 pins used , I don't have a photo to scale but I can take one, my idea was just to make another board with a lip and some pin headers that you could just press down on the other board, something pointy to poke through that sticker, it takes scraping to remove then something to clean. But I just hooked to it with the st-link v2 adapter from China I had, and just read the memory area. I certainly don't mind handing the jig off to someone who has the time to do this. I know I certainly don't have the time to do this for everyone.
I can tell you that Inno knows of 44 units in the US that were sold through them directly, and there are others that were bought through other channels. So I need to reach out to those customers and direct them where ever we decide as a group to send these.
Was there some software associated with this or just a self contained unit? |
|
|
|
Thanks, well that is a nice jig for sure, other than the alignment part I don't see what all the extra stuff does, but I was powering the board with the normal setup this may do something different, plus that sticker over the test points that takes scraping to remove not just a peel off. I don't know I would almost be willing if I could get the software to try one board, really only needed 3 connections to read the chip, although this may do a more in depth sanity check. For anyone that is interested I dismantled mine completely the other day and took photos, it survived the process. Pinned out all the cables to what they do and signals they are. One thing I found odd is I was curious what dc-dc regulators they were used to control the voltage but they lazered off the tops of them so you don't know what they are. http://photos.mag-productions.com/Hobbies/A4-miner/n-skNV3f/Anyone have a photo of the jig? I found the test points for the St Link interface to reflash the chip I just need the firmware unless there is some goofy procedure to doing this, really building a simple pin setup just to pop open the back push it on and hit send shouldn't be to far outta reach really. I was able to connect to one blade of mine and pull what firmware was on there off just to test. I live in Virginia not to central there either.
Oh you can pull the firmware?  Then they didn't set the write protect bit, *good*. I wonder if you could find someone to send you one blade from a 3.0 unit, then try pulling that firmware and flashing it to your 1.0/2.0 board. Or ask Inno to send you the firmware. I was thinking actually of building a "hat" that would go on the chip with a Pi behind it to just flash away but it sounds like Inno has the jig. C yea no protect bit, wasn't set on A2s either, already thought about getting a B3 card to get it from, as for asking Inno I was talking to someone there regular about this process and being able to help and I thought it was going well but they disappeared 3 or 4 days ago. |
|
|
|
|
Anyone have a photo of the jig? I found the test points for the St Link interface to reflash the chip I just need the firmware unless there is some goofy procedure to doing this, really building a simple pin setup just to pop open the back push it on and hit send shouldn't be to far outta reach really. I was able to connect to one blade of mine and pull what firmware was on there off just to test. I live in Virginia not to central there either.
|
|
|
|
@mjgraham , can you help me run my miner through command line via ssh? Pls. let me know. It making me crazy that i cant adjust the parameters and strategy's. It would be beautyful to have it run on awesome miner  I was able to manage it with zeusminer zoomhash but my last slip with the A4 scared me to fiddle more. Just to confirm, on multipool/H2C the miner runs flawless since 3 days with no mh drops or blackouts. It also works well with power reset and autostart on my end. So no worries on the current software v2.0.2. @usao, maybe a programmable timer with auto power recycle once every 3 hrs. help...i worked for me for my zeusminers. Well you can run a cron job to do a reboot but what I did was write a perl script for my A2s that read the number of cores on each board via the API and when it detected one that was at 0 for more then 20 seconds it restarted the miner, first problem it required connecting a reset to the hash boards that they had almost ran all the circuits for but didn't hook up the last bit for some reason that made them more reliable. I just got my A4s today so I havent gotten them going yet to start the debug, I do know they are batch 2 and the only fix for batch 1 and 2 is a MCU replace or reflash which I have contacted Innosilicon about and they did confirm that and we'll see how they want to precede, I took one apart to see how it is laid out and will be doing some analysis on it , some things I didnt like , like in the controller the raspberry pi has so much rust on all the metallic parts made me sad, has a revamped controller board that is nice. The reflash should be easy I have the tools just have to map out the connections, they didnt put an easy to access header on it just test pads to it will require some kind of jig to do it but it can be done, although for most it will require shipping somewhere. I hope to get at least one going tomorrow , I was a dummy and didnt calculate my PCIe power connector stash correctly. On my A2s I just soldered right to the header with screw terminals but these are new and I have to void all the warranties although unless they just catch on fire should be fixable. I dont like a non command line interface I run all my stuff that way, actually net boot the units from a server so the CF cards dont die and have scripts that monitor everything , pools/ rates all that crap ran trhough proxies and switch stuff for best rates, it is a hobbled up mess that has grown crazy really. I was lucky enought the be able to fix some issues with the driver for the A2 and compile it with curses support so it has the fimilar interface . if you wanted to you could do a crontab -e then on a line do 0 3 * * * /sbin/reboot then save it Ctrl-X if it is nano editor My issue is that a reboot doesn't fix the problem. It requires a power cycle to get the board(s) back... I figured that would be the case, on the A2 boards they had the MCU reset line brought right out to the connector but didnt fit a jumper so like you said you needed a power cycle, with the reset you didnt just toggle a GPIO pin and you were set. I'll look into the A4 board today and see what options there are, although most will require soldering stuff unless they ran this and just dont use it. |
|
|
|
@mjgraham , can you help me run my miner through command line via ssh? Pls. let me know. It making me crazy that i cant adjust the parameters and strategy's. It would be beautyful to have it run on awesome miner I was able to manage it with zeusminer zoomhash but my last slip with the A4 scared me to fiddle more. Just to confirm, on multipool/H2C the miner runs flawless since 3 days with no mh drops or blackouts. It also works well with power reset and autostart on my end. So no worries on the current software v2.0.2. @usao, maybe a programmable timer with auto power recycle once every 3 hrs. help...i worked for me for my zeusminers. Well you can run a cron job to do a reboot but what I did was write a perl script for my A2s that read the number of cores on each board via the API and when it detected one that was at 0 for more then 20 seconds it restarted the miner, first problem it required connecting a reset to the hash boards that they had almost ran all the circuits for but didn't hook up the last bit for some reason that made them more reliable. I just got my A4s today so I havent gotten them going yet to start the debug, I do know they are batch 2 and the only fix for batch 1 and 2 is a MCU replace or reflash which I have contacted Innosilicon about and they did confirm that and we'll see how they want to precede, I took one apart to see how it is laid out and will be doing some analysis on it , some things I didnt like , like in the controller the raspberry pi has so much rust on all the metallic parts made me sad, has a revamped controller board that is nice. The reflash should be easy I have the tools just have to map out the connections, they didnt put an easy to access header on it just test pads to it will require some kind of jig to do it but it can be done, although for most it will require shipping somewhere. I hope to get at least one going tomorrow , I was a dummy and didnt calculate my PCIe power connector stash correctly. On my A2s I just soldered right to the header with screw terminals but these are new and I have to void all the warranties although unless they just catch on fire should be fixable. I dont like a non command line interface I run all my stuff that way, actually net boot the units from a server so the CF cards dont die and have scripts that monitor everything , pools/ rates all that crap ran trhough proxies and switch stuff for best rates, it is a hobbled up mess that has grown crazy really. I was lucky enought the be able to fix some issues with the driver for the A2 and compile it with curses support so it has the fimilar interface . if you wanted to you could do a crontab -e then on a line do 0 3 * * * /sbin/reboot then save it Ctrl-X if it is nano editor |
|
|
|
It will be interesting to see, if all it requires is an MCU reflash that shouldn't be to bad, granted not everyone has the tools to do that , now swapping the MCU makes me wonder a bit I pulled the MCU code off the A2 boards and they really hadn't used much of the memory space. I was looking into getting an A4 but I guess I will hold off a little longer although the power savings in my case is about to make my A2s non usable. I have about modified them as much as I can wanted to use them to heat the house this winter so I built a PWM fan controller so they didn't sound like airplanes in the house and I have about repaired all the boards I have parts for.
It sounds like with a custom jig the MCU can be flashed. Inno is presently trying to setup an arrangement in the US to have someone do the flash updates for us. I assume they will also do the same for the EU part of the world. We are trying to sync up Inno with a member here to do the work. Hopefully something can be worked out for the B1 and B2 repairs. B3 miners look like they may be working better. We should be getting more reports here over the weekend how things are progressing on multipool mining with the B3 miners. If it is just an MCU reflash I have the tools to do that already, I like to tinker a lot I did try the .a4 cgminer on and A2 and it detected the boards , surprisingly they have not changed any of the nomenclature since the A1. Still lists everything as an A1 but has the voltage mod parameter. Would be interesting to put a logic analyzer and see how different the protocol is, or if much at all, to bad the driver code is not floating around like the A2, I really never used the web interface and ran everything command line remotely . |
|
|
|
|
Show Posts
