scam.  taking the time to write this incase anyone has hope that it isn't, so I can save you some money.

here's the tell...  It's all marketing blah blah blah, with one slight nod to technical.  He copied some ideas/text from how real FPGAs are done.  But look how it is artfully danced around:

To do this, the device creates a clone of the miner and both pools inside itself. This must be done because network traffic sniffing/decryption in real time is too difficult and we have to create endpoints.

So, I have some questions:

• "clone of the miner and both pools".  What?  Why?  You're saying you are creating a hardware protocol in programmable logic?  OK, this is because you need microsecond response times?  Why?
• "real time is too difficult".  No it's not.  The task is done in real time.  A malicious task could also be done in real time.  Delays on the order of miliseconds don't matter.  Your task is single purpose; the reason smart firewall traffic sniffing & decryption is done with FPGA assist is to greatly expand the capabilities of the device.  This is not necessary or useful in your case.  You are not selling a general purpose device that must have broad and expandable capabilities.  You are selling a toaster, it does one thing only.  It's far easier to implement in software, but with software you can't make the argument for pre-orders because you have to fund physical device development, can you?
• "we have to create endpoints".  Right, I get it, you have to end the sentence with something that sounds like it matches.  Try harder next time, this makes no sense.  Your premise is not very complex.  By "endpoint" you mean "miner".  But "endpoint" sounds more technical, because that's hardware lingo (PCIe CEM spec among others).  All you're claiming to do is mine on two pools at once.  It's not rocket science.  It's not even particularly hard.

Can we just get this loser banned already, or does he have to successfully scam people first?

you must disconnect for much, much longer than 15 seconds.  i've noticed these devices have tremendous capacitance, on the order of minutes.

if you didn't see magic smoke, it's not likely claymore's latest actually broke anything, but this tip of disconnecting is a good one because AMD cards seem to like to crash HARD and not easily reset.

FYI - latest eXtremal kernel and params utterly crash my AMD systems over a period of time.  About half drop to 0 Sol/sec and the other half become unresponsive and require hard AC cycle.

It also won't function at all on gtx970.

something about this version doesn't mine on gtx970.
looks like the solvers crash once launched.  Any way to produce debug logs for you?

ok, I'm going to say this - and I'm probably wasting my time - but please, all, if you're not a programmer, and you don't understand what is being talked about, please just be patient and wait for the developers to release stuff.  These posts asking for these things are just not going to be answered.  All they do is clutter the discussion.

There is not a public miner available that deliver 165 sol/s on a single rx470.  And if you think there is, go bother people in that thread.  It's obvious that what is meant is "a rig of 6 or 7 RX470 cards".  Seriously, use your brain here; have you seen an explosion in hash rate at a point in time and a corresponding drop in price?  No?  Then probably there is no 6x improvement via a public miner. 

Tested on 750ti - for me I see a few % reduction in performance.

You can do that, but I think your odds of success are questionable.  The road is littered with corpses of guys who gave their boss the middle finger, said "I'm smarter than all of you", and left to chase bitcoin.  Every once in a while they come on our radar and soon are gone.  Yes, there are angles for profit but the industry you would be in is extremely irrational and erratic.  Nobody in their right mind would loan you money, and your competition is people with free power / zero opex.

EEs cost money, SW guys cost money, Fabs cost a LOT of money, and each and every one of these things you bring in runs you the risk of IP theft.  It's not going to be very long before your key SW guy or EE guy realizes how underpaid they are, and walks away.  So you either destroy your profit margin or watch your IP and principals walk.

There isn't enough steady margin in this business to find a workable angle.  And even if there is, what you've described is a really hard way to make money.  Better off putting that energy into mowing lawns or building a restaurant or real estate.  Greater returns, simpler, more scalable. 

I do mine, but to me this is beer money.

Bug report with linux v1.1.

Found this across several computers, exactly identically.  The miner was completely broken and idle.  I'm not sure if it thought there was tampering or what?  But I had to disable the miner and switch back to mining ETH.


ZEC: Stratum - connecting to 'equihash.usa.nicehash.com' <198.11.195.136> port 3357
ZEC: Stratum - Connected (equihash.usa.nicehash.com:3357)
DevFee: ZEC: Job timeout, disconnect, retry in 20 sec...
GPU0 t=32C fan=20%, GPU1 t=31C fan=20%, GPU2 t=30C fan=20%, GPU3 t=31C fan=20%, GPU4 t=32C fan=20%
DevFee: ZEC: Stratum - connecting to 'equihash.usa.nicehash.com' <198.11.195.136> port 3357
ZEC: Job timeout, disconnect, retry in 20 sec...
GPU0 t=32C fan=20%, GPU1 t=31C fan=20%, GPU2 t=30C fan=20%, GPU3 t=31C fan=20%, GPU4 t=32C fan=20%
ZEC: Stratum - connecting to 'equihash.usa.nicehash.com' <198.11.195.136> port 3357
ZEC: Stratum - Connected (equihash.usa.nicehash.com:3357)
DevFee: ZEC: Job timeout, disconnect, retry in 20 sec...
GPU0 t=32C fan=20%, GPU1 t=30C fan=20%, GPU2 t=30C fan=20%, GPU3 t=31C fan=20%, GPU4 t=32C fan=20%
DevFee: ZEC: Stratum - connecting to 'equihash.usa.nicehash.com' <198.11.195.136> port 3357
ZEC: Job timeout, disconnect, retry in 20 sec...
ZEC: Stratum - connecting to 'equihash.usa.nicehash.com' <198.11.195.136> port 3357
ZEC: Stratum - Connected (equihash.usa.nicehash.com:3357)
GPU0 t=32C fan=20%, GPU1 t=31C fan=20%, GPU2 t=30C fan=20%, GPU3 t=31C fan=20%, GPU4 t=32C fan=20%
DevFee: ZEC: Job timeout, disconnect, retry in 20 sec...
DevFee: ZEC: Stratum - connecting to 'equihash.usa.nicehash.com' <198.11.195.136> port 3357
ZEC: Job timeout, disconnect, retry in 20 sec...
GPU0 t=32C fan=20%, GPU1 t=30C fan=20%, GPU2 t=30C fan=20%, GPU3 t=31C fan=20%, GPU4 t=32C fan=20%
ZEC: Stratum - connecting to 'equihash.usa.nicehash.com' <198.11.195.136> port 3357
ZEC: Stratum - Connected (equihash.usa.nicehash.com:3357)

Have you even confirmed that constant consumption of 30% CPU is actually a problem for you?  Very few things constantly consume CPU; the only one that comes to mind is 1080p (and up) decode of youtube that can be taxing.

This particular miner seems more memory-hungry than CPU.  I've had to upgrade my rigs to deal with it.

They're welcome to their unbridled optimism, but I'm going to keep selling my LBC to them at a discount.  And I'll bet any one of them 1 BTC that 6 months from now, I'll have netted more fungible money.

I've tracked down the problem to, I'm guessing, unsafe malloc in the lbry kernel.  Noticed that my mining rig, with only 2mb ram sees 25% memory use from ccminer -- pretty high.  When i drop caches the problem instantly goes away.  It doesn't, however, go away when I embed that drop_caches in the miner launch mechanism, so it's me fighting against some probabilistic processes.

Does that help narrow it down?  At a glance the only malloc I saw was cudaMalloc - does that malloc system memory?

While mining lbry, I'm seeing this error:

GPU #0: an illegal memory access was encountered

It isn't every GPU, and it varies if I reboot.  Some GPUs do mine; at first 3, then only 1.

Any idea offhand what's causing this? 

No luck with an initial attempt on a 750ti + CUDA 7.0.

Can anyone confirm that this can work on 750ti?

awesome--will do.  thanks!

completely useless bug report follows, but I feel obligated to offer it.

4.1, dual mining in linux.  after several days,

read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable
read(): Resource temporarily unavailable

attempts to restart mining just hang.
soft (AHCI) reboot fixes it (implies not a hard hang at GPU/driver).

Took first crack at using this (linux).

Works fine for a bit, then crashes one of the GPUs.  Interestingly enough, the watchdog doesn't.  :-)  YOU HAD ONE JOB, WATCHDOG!

...excerpt from debug log...
WATCHDOG: GPU 2 hangs in OpenCL call, exit



can't fall back & try older version (no linux build for pre-4.1).  If you can fix this, I'd prefer to deploy this, even with a fee.  The problem here is that it's constantly crashing at variable rates.

I used the pump to sell out my positions.

Anyone intrigued by this should first consider that the wallet crashes and often prohibits you from moving large amounts of money.  It took me approximately 30 transactions to simply move my BBR from wallet to exchange.  Annoying, slow, BROKEN.

This coin is sad.  You should either spend the energy to fix it or give up and let it die.  

This is baloney.  He's using industry terms but saying nothing.

Ok, you have a design problem that can be fixed with a BOM change.  What kind of problem?  If you know it can be fixed with a BOM change, why does it have to "go back for analysis"?  Does alpha have any engineers of their own, or are they just project managers hoping that this time the disinterested Indian contractors won't screw it up?  (and why should they, because the more they screw it up, the more billable hours they generate)

Why does it take weeks to months to make any progress, get any analysis?  Conventional practice is to prototype the BOM change yourself and test... THE SAME DAY.

I could go on and on.  I'm deliberately only scratching the surface.  This is not how the industry really works.  But, he's betting that nobody knows that.

Actually, I don't/didn't know that.  I read technical stuff, and I'm not really interested in behind the scenes formulas of a web forum.  I don't post when I have nothing of value to say.  If that makes me a "newbie" so be it.  I value quality over quantity.

It was just intended as a suggestion, because the documentation of the formula I found seemed to equally weight time on forum and posts.  Guess not.

In short, I've been on this forum for ~2 years as a registered user, and read silently long before that.  I mainly sit and read, read, read.  Often, logged out.  I have about 14 posts, sporadically across the entire time period.

So this all bit me yesterday when someone I reached out to said "ewww, no, you're a newbie, go away".  I figured I had 1-2 posts, because I rarely am inclined to speak up.  But to my surprise, I have 14... and I've been registered for 2 years.. WTF?  I don't care about badges and stupid labels, so this "newbie" thing was not something I ever looked at.  It was a surprise to me that the system itself would try to scare off anyone I talk to via private message.

So I spend time to dig and find the formula, and it's all based on lowest common denominator of browsing-while-logged-in and posts.  You might consider factoring in years and frequency of posts?  Or using cookies more aggressively the way advertising companies do?  

Frankly, most of the time I spent browsing while logged out was DELIBERATE as a means of protecting the site itself from XSS attacks.  I didn't realize this would be held against me.  If I were properly credited for all the time I've spent on this system, I'd be one of the top ranks, not the lowest.