ASIC testing on main net is wrong, PERIOD.

[Bogart]

We can type all day in this thread, but really we can't stop these companies from cheating us by mining BTC for themselves with OUR units, already paid for with OUR funds, if they choose to be so dishonorable as to do so.

We can only hope that the honest maker(s) ship(s) first, and vote with our monies when placing additional orders.  And/or try to cancel orders with any companies who refuse to pledge not to mine main net with customer units.

[1715163327]

1715163327

[1715163327]

1715163327

[1715163327]

1715163327

[1715163327]

1715163327

[Bogart]

Test net in a box seems like the best way to do ASIC testing... who is testing on main net?

BFL. People got pissed when their "burnin" account kept popping up on eclipsemc, now inaba has control of 500Gh/s of FPGAs  

...aren't there still a lot of people waiting on their BFL FPGAs to be shipped to them?  Are you saying there is 500Gh/s of them undergoing "burn in" while customers still sit empty handed?

[sgravina]

ELLIPSIS ...

[Shadow383]

Test net in a box seems like the best way to do ASIC testing... who is testing on main net?

BFL. People got pissed when their "burnin" account kept popping up on eclipsemc, now inaba has control of 500Gh/s of FPGAs  

...aren't there still a lot of people waiting on their BFL FPGAs to be shipped to them?  Are you saying there is 500Gh/s of them undergoing "burn in" while customers still sit empty handed?

I'm not saying anything   Maybe Inaba has a quarter of a million dollars worth of mini-rigs, who knows?
I'm just stating the facts, which are highly suspicious.

[Explodicle]

Did they ever say they wouldn't test on main net? If they didn't, then it's not fair for us to add new requirements after money has changed hands.

Furthermore, making the coming ASIC-fueled difficulty increase smoother is a good thing for the bitcoin economy.

Chalk it up as a lesson learned. When the next big thing is announced, demand either testnet-only testing or lower prices up front. I suspect that when otherwise identical manufacturers compete, the ones who do main net testing will out-compete those who don't.

[crazyates]

I get you, but here's the point, it's not just your ASIC and since difficulty only adjust every 2016 blocks, well the effect could be lasting.

No, that's my point. There are almost ZERO "lasting effects" after ~2 weeks, and any that actually are there are minuscule. I wait 1 Difficulty change after I get my ASIC, and whatever burn-in they did no longer affects my rewards. I'm getting the same amount of Bitcoins as if they had used a test-net.

As far as the whole "there are only 21 million coins, and their stealing the coins we should be mining" argument: Lets say they mine 1,000 BTC in their burn-in. All this does is push forward the reward half by a matter of about 3 hours. Any "lasting effect" of this reward half coming 3 hours sooner is, again, minuscule.

The Bitcoin network was designed from it's very conception(with increasing difficulty, reward halves, and TX fees) to eventually level out.

[Bogart]

Did they ever say they wouldn't test on main net?

Thank "bob" cablepair (bASIC) said it, and I think ngzhang (Avalon) said it too, though I will have to go back and look to be sure.

I suspect that when otherwise identical manufacturers compete, the ones who do main net testing will out-compete those who don't.

I hope to see the opposite happen, and will "vote with my monies" to help make it so.

[jjshabadoo]

@crazyates

bro it's not just your units or the ones shipped with your batch.

They will keep testing and testing can't you see that?

You'll be competing against the next batch of orders before they are even shipped.

It's unbelievable that people buy into this stuff.

And yes, those 1000 coins matter, ANY coins matter in a finite supply.

Well I suppose the apologists won't ever quit on this one.

[Shadow383]

I get you, but here's the point, it's not just your ASIC and since difficulty only adjust every 2016 blocks, well the effect could be lasting.

No, that's my point. There are almost ZERO "lasting effects" after ~2 weeks, and any that actually are there are minuscule. I wait 1 Difficulty change after I get my ASIC, and whatever burn-in they did no longer affects my rewards. I'm getting the same amount of Bitcoins as if they had used a test-net.

As far as the whole "there are only 21 million coins, and their stealing the coins we should be mining" argument: Lets say they mine 1,000 BTC in their burn-in. All this does is push forward the reward half by a matter of about 3 hours. Any "lasting effect" of this reward half coming 3 hours sooner is, again, minuscule.

The Bitcoin network was designed from it's very conception(with increasing difficulty, reward halves, and TX fees) to eventually level out.

You mean, if they ran all the ordered rigs for about 5 minutes total? 

[meebs]

I would say it is an issue of disclosure. If say you buy an ASIC being CLEARLY told before purchase that it would be used for 72 hours to main-net mine and the price you pay takes that into account that is 100% acceptable, as those are the conditions of the sale.

Other then that.. 100% wrong unless 95%+ of all BTC earned during testing was given to the buyer, in which case they are almost doing you a favor.

[crazyates]

@crazyates

bro it's not just your units or the ones shipped with your batch.

They will keep testing and testing can't you see that?

You'll be competing against the next batch of orders before they are even shipped.

It's unbelievable that people buy into this stuff.

And yes, those 1000 coins matter, ANY coins matter in a finite supply.

Well I suppose the apologists won't ever quit on this one.

I'm not "buying into" anything. I honestly just think you're making mountains out of molehills. I told you here:

Now they do this with 100 orders (or the equivalent of 6TH/s). They mine with 6% of the network for 1/14 of the period, for an increase of 0.42% increase in difficulty. This might actually be noticeable, but still, hardly worth freaking out about.

They could burn in 15TH/s of ASICs (with our numbers of a 100TH/s network, that's 15%) for 24 hours, and the increase would only be 1%, and even that's only for 1 difficulty period! After 2 weeks, when that 15% increase is hashing 24/7 in the hands of customers, it wouldn't even matter!

And this is at a 5X network size. I've seen estimates anywhere from 5X to 100X. The larger the network, the less their burn ins affect it.

So with the worst case scenario, the smallest projected network increase and a high amount of ASICs being sold per week, we're talking about a 1% difficulty change for one difficulty period (~14 days). That's going to cost miners a fraction of a BTC, over the entire grand scheme of things. Again, why is this such a big deal?

[Kaliecious]

I do think the asic suppliers should do a burn in on the main net. but what should be done is contact the purchaser of the unit that would be burned in and get a worker login for thier said pool, and burn in with the purchasers login information. inorder to keep burnins being done hush hush if a purchaser was to announce that burn ins are being done the go to last of shipping list.

[Graet]

testnet and testnet in a box were designed for testing before bringing to main net - that's what should be used, no questions asked.No discussion entered into.

If it must be mainnet (for whatever imagined reason)
well... if only I could supply login details for a worker - they could test my asic against my account on a pool I choose- this issue would dissolve and could even be seen as a gesture of good will towards people that have pre-ordered and waited.
Otherwise it will be profitable mining at lower difficulty with asic than anyone else is able to, will push difficulty up before asics start getting delivered (its not going down after they start getting delivered) and affect every miner not just the ones ordering asic.

As a bonus miners would know when their ASIC is getting close - they could see hashes on that worker

<sarcasm> radical idea!!! no way it could get adopted - too logical for Bitcoin </sarcasm>

any asic maunfs planning to test on mainnet want to balls up and do this?

goodwill is priceless
my thought
Graet

I typed too many words Kaliecious beat me

[Kaliecious]

but yours made better sense, my grammer sucks.

[Sitarow]

ASIC Pre-Shipment Testing Policy

Update:

We are often asked what our official policy is towards testing the ASIC equipment prior to shipping it to customers.

Our official policy is that we will burn in/test your equipment for a minimum of 24 hours on what is called "Testnet-in-a-box." This will allow us to verify that you equipment is working and able to sustain operations at normal operating conditions for a minimum of 24 hours.

https://forums.butterflylabs.com/showthread.php/52-ASIC-Pre-Shipment-Testing-Policy

[crazyates]

testnet and testnet in a box were designed for testing before bringing to main net - that's what should be used, no questions asked.No discussion entered into.

I thought that was more for software/protocol changes?

Disregard any previous debate going on in this thread: does the testing of any new mining hardware (GPU, FPGA, or ASIC) require the use of a test-net?

I'm going to assume the answer is yes. When support for a hardware is written into a miner program, be it an openCL kernel, or a FPGA bitstream, I'm assuming it gets tested on a test-net first.

I'm going to also assume that BFL or anyone else adding support for BFL hardware will also do this, while configuring the firmware/mining software. But once that's done, why does a test-net need to be used for every single piece of hardware?

I'm asking this seriously, because I greatly respect your opinion, Graet. Half of me agrees with you, but I just like a good argument.

[crazyates]

ASIC Pre-Shipment Testing Policy

Update:

We are often asked what our official policy is towards testing the ASIC equipment prior to shipping it to customers.

Our official policy is that we will burn in/test your equipment for a minimum of 24 hours on what is called "Testnet-in-a-box." This will allow us to verify that you equipment is working and able to sustain operations at normal operating conditions for a minimum of 24 hours.

https://forums.butterflylabs.com/showthread.php/52-ASIC-Pre-Shipment-Testing-Policy

So BFL will be using a test-net. Cablepair will be using a test-net. Are any manufacturers using the main-net?

[Graet]

well testnet is for um testing Bitcoin stuff
all aspects
with testnet in a box you can even choose difficulty and stuff - surely an advantage to high hashrate devices

um I like bad analogies - someone raised cars before
so if my car manufacturer is testing my new car I would prefer he tested it on a test track before it went onto public roads (I know no-one will get killed if an asics brakes fail - but hopefully you get my drift )

the difference between testnet and mainnet in the Bitcoin software is this flag
-testnet           Use the test network
cheers
Graet

[dentldir]

Dear BFL,

If you test on main net, please forward the earned BTC to my BTC address.

Sincerely,
Someone who is not very concerned about thus issue.

p.s.  Please test my order first and for as long as you can.

[kaerf]

i certainly hope there will be some testing on mainnet. i don't want to get a product from any company, if it hasn't been proven on real live pools.

how will pools respond to some much additional hashing power? will a sustained 10-20x increasing in hashing power have an adverse effect on pools? i don't want to see pools going down after ASICs come out.

will the product work with stratum? how about getblocktemplate?

i think it's unreasonable to think/ask that that no testing occurs on main net. bfl, btcfga, and avalon should all have engineers/developers performing functional and stress testing on REAL world mining scenarios before they send their products out to customers. i sure as hell don't want to get an ASIC miner only to find out that there is a critical bug that cripples my ability to mine...

i don't know how big each company's development team is , but I'd say it'd be reasonable to have maybe 3-5 of their test machines running on main net during development. of course, once they've finalized their development and testing, there is little need to keep those machines attached to main net. then before shipping, customer machines can be "burned-in" on a test net .