Topic: GPUMAX | The Bitcoin Mining Marketplace - page 51. (Read 215554 times)

Good to know, thanks!

Hey guys,

While we started to update the systems, we noticed a little bug that seemed to get bigger and bigger with each update.  So we halted the update process and attacked the bug but its clever one. So we took the whole system down overnight to find out what's going on.

WE WILL BE BACK!!!

That sucks....

Is it just me, or does anyone else get a bit scared when Pirate is so 'silent'....
(or does that silence/worry only apply to the Bitcoin Bank thread ?....lol)

HA, great time to start testing on here. Got my invite a while ago and never did anything with it. Tried to setup a worker tonight (blank account, never used it before) and its spits out an error. Could be toast, who knows.....

MAX is out of the race...

Yeah I just noticed.  I'm bypassing the MAX for now.

A little after midnight UTC (~5 hours ago as of this post) even passthrough mining on the private pool isn't working for me, and I wasn't have any of the problems anyone else has mentioned today. The site is still accessible, though.

I'd be glad to set up a worker on my account for testing if you so desire.

Perhaps Pirate can once and for all put some of the recent posts/questions to rest with a couple of quick answers and a small update as to what exactly is happening currently with GPUMAX ?

As far as the IRC Chan goes, I didn't sign up for my account on IRC (as most people didn't) so I hope that I don't have to go there for news.

It would be nice to see this thread get used for more than arguing and complaints.

I would have offered a while ago, if he didn't initially approach it as if it were fatally flawed from the start and badmouth it just because of a lack of invites. #gpumax on Freenode has the developers, and I'm sure if asked, 'most anyone would have offered some access to their account.

You don't know anyone who could set up a worker on their account for you just to do your testing?

Some of the behavior that I've already accounted for is the semantics that GPUMAX implies for "200 OK" responses with zero Content-Length to long poll requests. This isn't mentioned in the standard at all, and several bitcoin-related software developers confirmed that this smells very very weird.
According to the GPUMAX guys, this behavior was actually reverse engineered from some (buggy) miner software. In the meantime I've patched MPBM to behave the same way, it's an unspecified corner case => undefined behavior after all, even though I'd much rather like to handle that case more cleanly (as an error). However, as reported by some MPBM users, there are some more issues. I can't figure out the details without access to GPUMAX though.


What kind of bounty? So far I have not been involved with any cgminer bounties at all. I just offered to give them protocol specifications if need be. I can't see anything wrong with that. In contrast to some other FPGA mining board vendors we support open standards and open source software.
Actually I've received some bountys to add MPBM support for some boards (Icarus and X6500), and I'm wondering if cgminer will ever support the X6500, even if they send a board to ckolivas. This is going to be way more complex than anticipated, and it would be even worse if I hadn't spent several days recently to streamline that code within MPBM.


Sure, most people are still mining on GPUs these days. But there are already several FPGA boards that cgminer doesn't support (yet). Ztex and X6500 just to name two. Actually several people have asked for MPBM GPU support as well. No idea how many MPBM installations there are, but I know that there are several hundred FPGA boards in the wild that don't work with cgminer.
I have no idea how cgminer handles these particular protocol oddities, it probably just silently ignores any errors and is therefore compatible with what GPUMAX expects, but I don't consider that good practice at all. And I just don't really consider making a testing account available to miner software developers "bowing down to them". It's one click him, and has no downside at all. Refusing to do that means just refusing to cooperate at all.

However, if you read my above post again, you'll notice that the comment about MPBM incompatibilities was just a little side note. This surely won't kill GPUMAX. (it isn't called FPGAMAX after all...) The recent developments that I've observed in this thread seem rather dangerous to that project though, as far as I can tell from the outside.

I've had invites waiting both to supply and buy hashrate - not happening

Wouldn't it be interesting, if, hypothetically, someone set up some sort of high-interest investment scam using bitcoin, offering investors several percent a week on the proviso that they invest very large amounts (thousands of BTC), then had a scheme which coins invested (over and above those required to keep paying interest to new investors) could be funneled into to pay miners to mine on their behalf at major pools, allowing the entirely hypothetical person who ran the scheme to disappear once the initial high-interest ponzi scheme failed with thousands or tens of thousands of freshly minted coins which are very difficult to trace.

Of course, nobody would ever do that, right? 

You really need to clarify that statement, as the way it's written, it's essentially an attack, albeit vague, but enough to give onlookers the wrong idea if taken out of context.
Not cool.

Also, as I recall, you have been given the opportunity to streamline YOUR operation by jumping onboard the CGMiner bandgwagon, by paying part of the bounty to have your hardware supported. Considering CGMiner makes up a large portion of the miner's software, I would start there rather than ranting about your 1-off software having problems with GPUMAX's service. Open up that fat wallet and take care of business.

What makes more sense ? YOU helping pay a bounty to get your hardware supported by CGMiner ? or, Pirate bowing down to the users of your software, which pale in comparison on a purely numbers-only basis.

Think about it.

I obviously can't really evaluate this without having an account, but given the recent pages of this thread, I get the impression that this seemed to work nicely while everything was on the old servers, with people being happy. Now that they switched servers (and pirate said they'd get everyone aboard within one week of that, which is like a month ago now) nothing seems to be working right any more and they seem to be fighting some kind of issues all the time, offering basically no leased work at all, making the project rather pointless for the miners, and probably also scaring away possible customers.
All in all, this looks like a very negative trend to me which will sooner or later kill this project. But, as I said, that's only the perspective I got from reading posts here.




Well, I don't really care any more, because it obviously isn't working right anyway.
And I also don't really care whether my miner software works well with GPUMAX in this case, given that I can't test this easily to work around some of their violations of the protocol standard (which, I must agree, isn't really documented well).

OUCH !
I think it's fair to say that the system is and always has been in BETA testing (up to this point and for a bit longer it seems).
Pirate has never promised any more or any less than that, nor has he set a 'Production-Phase' Start Date, so no one should be let down....

Posting negative comments about the service is no way to speed up the processing of your invite to participate...lol

Same here. No reaction from the pirate since like a months regarding this topic.
This seems to be all-broken these days anyway, so there's probably no point in even waiting for an invite. Potential customers will probably have run away by the time they get this working right.

Forget the invites, get the work rolling otherwise the invites are pointless

Are new invites still being sent out?  I signed up 2 months ago and haven't heard anything.

OCZ PSU's sucked if they were anything but seasonic made =P Cuz all their other OEM's suck.