New Official AMT Thread

[NotFuzzyWarm]

Random thoughts on A1 board packaging & cooling. Can anyone give me outside dims on the current blades? I assume thickness is 1/2" (the caps) and a bit more for the pcb. Say total of 5/8". Don't care about any heatsink thicknesses.

Been kicking this around: ya know with liquid cooling done right using cold plates I can see putting 14 or more of the current 8 chip boards in a 4U rack case with a 1U high PSU bank on top...

Thermals not an issue for any component and if can fit more chips on same size board - cool. Should help OC as well until com or other issues arise. Do this with other chips like BFL' CoinTerra or HasFasts new chips and whooo daddy! 10TH or more in a 4U rack...

Hmm. if the A1 chips get cheap enough might play a bit with some help from here...

[1713535532]

1713535532

[FrictionlessCoin]

Random thoughts on A1 board packaging & cooling. Can anyone give me outside dims on the current blades? I assume thickness is 1/2" (the caps) and a bit more for the pcb. Say total of 5/8". Don't care about any heatsink thicknesses.

Been kicking this around: ya know with liquid cooling done right using cold plates I can see putting 14 or more of the current 8 chip boards in a 4U rack case with a 1U high PSU bank on top...

Thermals not an issue for any component and if can fit more chips on same size board - cool. Should help OC as well until com or other issues arise.

Hmm. if the A1 chips get cheap enough might play a bit with some help from here...

A1 going rate is about $40+ 

The Bitmine and AMT prices are complete bogus.

You can tell what the price should really be looking at technobit price .

[NotFuzzyWarm]

Mmm that's still $1,600 for chips alone per-TH (40 chips) with no discounts. They need to be maybe 1/3 cheaper to make the hashing density attractive as the selling point vs the A1's power usage.

[ISAWHIM]

Well, Woke to find the miners mining at 480, instead of the 580 where they were last night.

Turned it off, to see if that helped... It did. The middle card which was hashing at 98GHs, causing the 480, went back to hashing at 190GHs...

Came in this afternoon, it was back to 480GHs... Rebooted it, and now the middle module does not report being detected. Not to mention the chip seems to think the first module is in slot 2 and the third module seems to be in slot 4, and it does not check the fifth slot. (Checked from slot 1-4)

The cards are in physical slot 1-2-3. (2 is not responding.)

Other oddity... it came online several boots, with the same IP address. Could have been just crappy random luck. However, I suspect it is more than that. As I assume once an IP is obtained, it would try to keep it after rebooting. Thus why the page stays up and keeps refreshing.

However, PuTTYing into the miner and killing the detected cars, did nothing to make the non-replying card reply.

So, now I am hashing away at a sad 380GHs until I can remove this "non-replying" middle card, and add-in my last "possibly good card" which I have not tested yet.

It is not the cards that die, it is the controller that replies to the backplane/ras-pi that is failing. The card was running fine just as all the others were, before they stopped communicating. Rebooting is the possible source of the issue. It is either ignoring the communications, or skipping channels, or the communicating chip that reports back is failing.

Time to break out the thermal imager, to make sure this one isn't burning up behind the failed communications. It is drawing power and doing something. Just not talking to the Ras-Pi or doing actual mining.

Once again, I have to take the whole thing apart, to replace the faulting module.

Will be using this heatsink for one of the other modules that came without a heatsink. Though now, the most I will get is 990Ghs, if I can get five running cards. (The most they do is about 205GHs nominal. With all cores running and no errors, with a target temp of 60c.)

I need a 6th heat-sink to get close to my 1.2THs, and at the moment, the 1.5THs is a pipe-dream for me. (Still not sure why I was sent this, like this. This is not what I asked for. So, I still assume I am supposed to be "fixing" or "troubleshooting" these. However, I can't do anything without the proper test equipment or schematics for these boards.)

Not sure if I should mark mine as "delivered"... lol (Since this is not what I asked for, and not one of the options they submitted.)

[rik_khaos]

Well, Woke to find the miners mining at 480, instead of the 580 where they were last night.

Turned it off, to see if that helped... It did. The middle card which was hashing at 98GHs, causing the 480, went back to hashing at 190GHs...

Came in this afternoon, it was back to 480GHs... Rebooted it, and now the middle module does not report being detected. Not to mention the chip seems to think the first module is in slot 2 and the third module seems to be in slot 4, and it does not check the fifth slot. (Checked from slot 1-4)

The cards are in physical slot 1-2-3. (2 is not responding.)

Other oddity... it came online several boots, with the same IP address. Could have been just crappy random luck. However, I suspect it is more than that. As I assume once an IP is obtained, it would try to keep it after rebooting. Thus why the page stays up and keeps refreshing.

However, PuTTYing into the miner and killing the detected cars, did nothing to make the non-replying card reply.

So, now I am hashing away at a sad 380GHs until I can remove this "non-replying" middle card, and add-in my last "possibly good card" which I have not tested yet.

It is not the cards that die, it is the controller that replies to the backplane/ras-pi that is failing. The card was running fine just as all the others were, before they stopped communicating. Rebooting is the possible source of the issue. It is either ignoring the communications, or skipping channels, or the communicating chip that reports back is failing.

Time to break out the thermal imager, to make sure this one isn't burning up behind the failed communications. It is drawing power and doing something. Just not talking to the Ras-Pi or doing actual mining.

Seems similar to one of my cards. It would pull power, especially to one chip that was starting to overheat while the other sinks on that card were warm but cooler. This was a non-hashing board that was still drawing power. Two other boards wouldn't detect but didn't heat up at all. They were properly plug in for power they just didn't seem to be drawing any (or any significant amount)  I really should invest in a multimeter.

[rik_khaos]

As far as I'm concerned they have delivered one unit. Anything that doesn't work as advertised, or is faulty (no mare where in the process the fault lies) is not a delivered product, just junk I was sent.

[ISAWHIM]

Yea, Rik... I had that same issue... check that cards chip-mounting... I assume it was lifted on one corner, like mine...

Guessing it is the robot setting them down improperly, or obstructions, or impurities in the solder, or could even be just to thick of a silkscreen pattern there.

For whatever reason, my other card is not initiating. That one has no warm spots, no burnouts... just fails to initiate. (Thus, I blame the primary chip that handles the communications to the backplane, or one of the serial chip communications circuits. One fails, apparently they all fail to reply.)

Got the last living card installed, but now I am afraid to turn it off... Might not turn back on. lol. That seems to be a trend with the cards.

However, later, I will setup the other Ras-Pi, and see if I can detect a single point of failure on that card.

Then, begin the process of swapping heatsinks with the cards that came without heat-sinks, to see if those even work.

I did notice that this v3.0.A card only has one of the four "COUT" connections, per chip. Two, apparently not on the correct side. (Also note, there are screw-thread markings on the solderpoints where one of them should have been. (Apparently the fan-screws were drilled right into it. Possibly shorting it out? Not from me, I have fans mounted with plastic push-in mounts.) Not sure why the fan screws are that long in the first place. Might be why some fans didn't even have screws mounting them.

Not sure why they don't have all four junctions of the "COUT" filled on this board. Really not sure what damage the fan-screw being drilled across both terminals would have done to the board, but I can't imagine it being good.

[rik_khaos]

It seems like what ever wasn't fucked up by the manufacturer was fucked up by whoever was fixing their case issues or their assembly staff in general.


Do you even see a solution that could be easily fixed given their situation? Or are the current run of boards a complete loss. I'm betting IMET has put a lien on AMT property if they are claiming non-payment which would further complicate matters. Though they must have some cash reserves if they are funding a run of the tecnobit boards.

I'm just looking not to get fucked because of their mismanagement of the company.

[FrictionlessCoin]

It seems like what ever wasn't fucked up by the manufacturer was fucked up by whoever was fixing their case issues or their assembly staff in general.


Do you even see a solution that could be easily fixed given their situation? Or are the current run of boards a complete loss. I'm betting IMET has put a lien on AMT property if they are claiming non-payment which would further complicate matters. Though they must have some cash reserves if they are funding a run of the tecnobit boards.

I'm just looking not to get fucked because of their mismanagement of the company.

The Bitmine boards clearly have a problem as evidence by the difficulty of shipping by Bitmine.  However,  Bitmine designs are at least know to be functional in the field.

AMT could ship technobit boards but if you look at pricing one 8 chip 240GHs board cost $380.  So $5500 worth should equate to around 14 boards.... about 3.6 THs worth.   The depreciation of these miners are utterly crazy and that is why even a minor slip in shipment times can have devastating consequences.  AMT is late by over 2 months.   That is why a full refund may be the only remedy to make us whole.

[opieum2]

Well, Woke to find the miners mining at 480, instead of the 580 where they were last night.

Turned it off, to see if that helped... It did. The middle card which was hashing at 98GHs, causing the 480, went back to hashing at 190GHs...

Came in this afternoon, it was back to 480GHs... Rebooted it, and now the middle module does not report being detected. Not to mention the chip seems to think the first module is in slot 2 and the third module seems to be in slot 4, and it does not check the fifth slot. (Checked from slot 1-4)

The cards are in physical slot 1-2-3. (2 is not responding.)

Other oddity... it came online several boots, with the same IP address. Could have been just crappy random luck. However, I suspect it is more than that. As I assume once an IP is obtained, it would try to keep it after rebooting. Thus why the page stays up and keeps refreshing.

However, PuTTYing into the miner and killing the detected cars, did nothing to make the non-replying card reply.

So, now I am hashing away at a sad 380GHs until I can remove this "non-replying" middle card, and add-in my last "possibly good card" which I have not tested yet.

It is not the cards that die, it is the controller that replies to the backplane/ras-pi that is failing. The card was running fine just as all the others were, before they stopped communicating. Rebooting is the possible source of the issue. It is either ignoring the communications, or skipping channels, or the communicating chip that reports back is failing.

Time to break out the thermal imager, to make sure this one isn't burning up behind the failed communications. It is drawing power and doing something. Just not talking to the Ras-Pi or doing actual mining.

Once again, I have to take the whole thing apart, to replace the faulting module.

Will be using this heatsink for one of the other modules that came without a heatsink. Though now, the most I will get is 990Ghs, if I can get five running cards. (The most they do is about 205GHs nominal. With all cores running and no errors, with a target temp of 60c.)

I need a 6th heat-sink to get close to my 1.2THs, and at the moment, the 1.5THs is a pipe-dream for me. (Still not sure why I was sent this, like this. This is not what I asked for. So, I still assume I am supposed to be "fixing" or "troubleshooting" these. However, I can't do anything without the proper test equipment or schematics for these boards.)

Not sure if I should mark mine as "delivered"... lol (Since this is not what I asked for, and not one of the options they submitted.)

Pretty much the case with all the units delivered. Not one is what was originally advertised. On the linux side I am working on getting the distro up and running. Ran into an issue with the SDcard and had to recreate the image.

[ISAWHIM]

As far as I'm concerned they have delivered one unit. Anything that doesn't work as advertised, or is faulty (no mare where in the process the fault lies) is not a delivered product, just junk I was sent.

By that thinking, it is safe to say I have half a unit delivered. 590 GHs (3 boards), which sadly shows as only 500GHs on "BTCguild". Great, after taking 3%, the pool tops it off by throwing away roughly another 17% of your total hashing power. Coincidence that it totals 20%. Now you know why I don't like pools. Doesn't matter, as long as they are throwing away 17% of everyone-else's total power also. lol. (This is why I wanted to solomine. Sadly, the custom software does not do solomining. Seems to only mine to a stratum setup, not the standard HTTP, which is all a wallet can do. Since stratum is a third-party hack/invention.) Solomining, you get 100% for your effort, which is usually 20% more reward, or more, in the end. On coins you can solomine.

opieum2: If you can make the standard HTTP mining work, I would be forever grateful!

[ISAWHIM]

I created a "TEAM" on BTCguild, "AMT Miners"... Feel free to join it.

Will be interesting to see what share of that pool we can consume.

[ladi4ever]

I did a review of the code on github for the A1 driver. It appears that improvements were made to the delays scanning. However I am having this information confirmed as driver coding is not my strongest area. I asked a few more questions on improvements of the driver since the first version was made available. This might be helpful for us in getting these cards working. Since there is not alot of documentation (well none at all) on the AMT variants of these cards, this is alot of trial and error and guess work. I dont know what the result will be, BUT I am hopeful we might get some results and answers. Even with a single fried chip, it should be possible to bring the cards up partially unless they are chained in a way that brings the whole card down.

This is one hell of a learning experience here for me FYI. I know linux and am getting deep into hardware areas I never did before like drivers. So wish me luck. If I get a working miner/card tonight I will post an early alpha for testing. Once the initial step is done (zombie cards) I will work on features like a web interface and such which I will add later or someone else can. My plan is just a simple working base setup. I am going to call this project zombie lol. Seems like an appropriate name to get these miners back from the dead.

Uh... I think project name should be called 'Lazarus'.   Zombies are living dead,  these AMT systems are more like vegetables than zombies.

Eh, not a fan of overt religious references. Zombies works. These things were half dead when we got them, and now they are being raised from the dead. And to be honest if there were burned out chips on them who knows what state they will be in starting up, if at all. Hence zombies. Could be missing a chip or two but still work just like a zombie could still move even if they are missing arms, or crawl if they are missing legs same idea here. Hopefully the cards are not in some kind of chain themselves that if a chip blows it fails the whole thing. I am still figuring all that out from a design standpoint.

Oh my god Opium i see you are a very bored man...You posted on this forum 80% out of all..

I can also see why AMT ignores you, Asshat. Unlike you I am trying to help the people out here who are having issues. I got enough time on my hands to do what I wish with my time. You spend your time posting up people's personal cell phone numbers and what commenting on my post count? Now who is bored if you have to look that up? What have you done to contribute?100 percent of your posts are bitching We all are in the same boat and are handling it in a way that does not involve constant bitching and helps each other get results.

 So yea I might have 80 percent of the post count but its also at least trying to help people. If you don't like that then go fuck yourself. I'm pretty tired of the comments I get on that both public and PMs on the matter if anything its going to make me do more just to piss the trolls and flamers off.. If you can do better than bitch about AMT not getting you miners then do it and stop talking shit. Unlike you I got miners (broken as they were) and got results because of my actions. You....bitch bitch bitch. And got nothing. At the same time I am making contacts here that are both helpful and in turn I have been able to help. A useful community vs useless retards who bitch.

When you can contribute better you can talk...until then STFU troll.

I would say, AMT is ignoring everyone, maybe not you, (looks like you are their friend), second, he posted his info him self because they want to fraud more people by selling more orders. For your info, is not bitching, but complaining this makes more sense, well maybe not for you but for everyone else...
Committing fraud is a serious crime, accepting more orders that's even worst, apparently they have not learn the lesson.
You seems to be a very angry man when someone gets your attention, did mom feed you sugar soda when you were a Kid?

[opieum2]

Easier now to hit ignore on morons. I have no tolerance for stupid.

[opieum2]

I created a "TEAM" on BTCguild, "AMT Miners"... Feel free to join it.

Will be interesting to see what share of that pool we can consume.

I'll give it a go later today.

EDIT Added myself to the pool group. This is with the existing running hardware. (3 cards @ around 490 GHs average). I was unable to power up the miners last night with the new distro, I will post some updates tonight if I make progress.

[opieum2]

As far as I'm concerned they have delivered one unit. Anything that doesn't work as advertised, or is faulty (no mare where in the process the fault lies) is not a delivered product, just junk I was sent.

By that thinking, it is safe to say I have half a unit delivered. 590 GHs (3 boards), which sadly shows as only 500GHs on "BTCguild". Great, after taking 3%, the pool tops it off by throwing away roughly another 17% of your total hashing power. Coincidence that it totals 20%. Now you know why I don't like pools. Doesn't matter, as long as they are throwing away 17% of everyone-else's total power also. lol. (This is why I wanted to solomine. Sadly, the custom software does not do solomining. Seems to only mine to a stratum setup, not the standard HTTP, which is all a wallet can do. Since stratum is a third-party hack/invention.) Solomining, you get 100% for your effort, which is usually 20% more reward, or more, in the end. On coins you can solomine.

opieum2: If you can make the standard HTTP mining work, I would be forever grateful!

I should be able to port that over. It will take a bit more work than my initial effort to just get cgminer working with the latest firmware. But its doable. Just a feature to come is all.

[FrictionlessCoin]

I wonder if anyone saw these ratings:  https://bitcointalk.org/index.php?topic=456691.0

Rating system
Position in the list signifies trustworthiness, the higher the better. Criteria includes the use of preorders, delayed shipping, immoral behaviour, withholding of refunds etc.
   A = Highly established, top tier firm
   B = Established, mid tier firm
   C = Mid tier firm with problems, or new firm yet to prove products
   D = Mid tier firm with problems, or new firm yet to prove products + ethical problems
   E = Firm with horrendous problems (including bankruptcy)
   Z = BFL.

Ratings

Company	Product Lines	Thread	Website	Delivered?	OnTime?	Rating	Comments
ASICMiner	Erupters, Blades, Cubes	Link	N/A	Yes	Yes	A+	Publicly floated company, only sells in hand hardware. Any minor problems resolved immediately. New generation soon.
Bitmain	Antminer S1, U1	Link	Link	Yes	Yes	A+	Only sells in hand hardware, continued success.
Bitfury	H-boards, chips	Link	Link	Yes	Yes	A	First company to deliver on time, in quantity and without any problems.
KNC Miner	Saturn, Jupiter, Neptune	Link	Link	Yes	Yes	B+	Slight delays on most recent generation, but has compensated users and honors refunds.
Avalon	Avalon, Avalon 2, chips	Link	Link	Mostly	No	B+	Community boycotted due to the handling and delay of B3 and chip orders. Trade in orders were never refunded. B1 and B2 suffered no problems.
SpondooliesTech	SP10, SP30	Link	Link	Yes	Yes	B	Best product release of any company to date. Sold preorders quite far into the future. No problems to date.
CoinTerra	Terraminer IV	Link	Link	Yes	Mostly	C+	Products did not achieve advertised hash rates. Compensation not sufficiently being provided.
Black Arrow	Prospero, X1, X3, Minion	Link	Link	Yes (prev)	No	C+	Historically reputable, suffering delays on current generation. Delaying refunds until first products are shipping.
BitMine	Coincraft A1, Rig, Desk	Link	Link	Yes	No	C+	Continuing delays, communication has improved.
VMC	Fast Hash One	Link	Link	No (yet)	No	C	Offered and honoured refunds from the start. Clones now appearing, original chips not yet ready.
HashFast	Sierra, Babyjet	Link	Link	Yes	Delayed	D+	Finally delivering, however high failure rates in initial batch. Batches continue to be delayed. MPP scheme soon to be shipping
BFL	Jalapeno, Single, Monarch	Link	Link	Yes	Delayed	Z-	Delayed products 6-12 months, refused refunds. Constant unethical and illegal practices. Tries to swindle newbies into buying obsolete hardware.
ChipIntegrators
Lightning ASIC	LA1THS, LA5U, LA3M	Link	Link	Yes	Yes	B+	Products being delivered consistently, no problems. Only sells in hand hardware.
Technobit	S-Hash, Hex8A1, Hex16A2	Link	Link	Yes	Mostly	B	Offers boards based off multiple chips, some delayed, some excellent.
OneStringMiner	OSM	Link	Link	Yes	Yes	B	Uses Bitfury chips, only sells in hand hardware.
Drillbit	Drillbit 8 board, Thumb	Link	Link	Yes	Yes	B	Uses Bitfury and Avalon chips.
Redhash	Redhash 105 Avalon Clone	Link	Link	Yes	Yes	B	Uses Avalon chips for clone.
ASIC Runner	Hexfury USB	Link	Link	Yes	Yes	C+	Uses Bitfury chips. No reported problems.
Big Picture Mining	Blue fury USBs	Link	Link	Yes	Yes	C+	Use bitfury chips. Problems with overheating, refunds and replacements still pending.
Starfire X	BTCGarden Blades	Link	Link	Yes	Yes	C+	Use BTCGarden chips. Small sales, no reported problems.
AMT	80-1200 GH Coin Miner	Link	Link	Yes	No	D+	Very delayed, no communication.

[opieum2]

Yea after the fact lol. Honestly that should be a sticky on this forum IMO. I commented there to vote for sticky status. I suggest others do to just to get it there. That page alone is huge in deterring people away from scams or incompetent vendors.

[NotFuzzyWarm]

<snip>
Committing fraud is a serious crime, accepting more orders that's even worst, apparently they have not learn the lesson.

If you are referring to AMT's website - it has been frozen for a month or more due to the suits & investigations so it cannot be changed to reflect their current non-sales mode. It is what it is.

However, AMT is also NOT accepting new orders. Even Phin proved that.

[FrictionlessCoin]

Yea after the fact lol. Honestly that should be a sticky on this forum IMO. I commented there to vote for sticky status. I suggest others do to just to get it there. That page alone is huge in deterring people away from scams or incompetent vendors.

AMT wants to make the argument that all vendors are having trouble like they are.

Well, yes,  most of them have problems, but most of them have been known to deliver something.